Compare commits

..

537 Commits

Author SHA1 Message Date
Marco Costalba a7fcdfd6bf Stockfish 1.7
Signatures are:

./stockfish bench 128 1 12 default depth
8299338

./stockfish bench 128 1 13 default depth
15694903

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-04-07 13:08:50 +02:00
Tord Romstad 41816b7ced Fix PowerPC and ARM compatibility. 2010-04-06 10:19:09 +02:00
Tord Romstad 13224e1d9d Add -mdynamic-no-pic to CFLAGS when compiling with GCC under OS X.
Without this flag, the __cpuid() function doesn't compile correctly
in 32-bit mode.
2010-04-05 21:47:28 +02:00
Marco Costalba 10c1ae8da0 Fix one gcc 4.4 warning
Properly fix previous warning. Patch from Joona.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-04-05 20:11:41 +01:00
Marco Costalba 3a62738174 Fix a warning in HT_enabled()
Under gcc we have:

warning: dereferencing type-punned pointer will break
strict-aliasing rules

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-04-05 15:36:06 +01:00
Marco Costalba 84451191f3 Store score in TT when null search fails high
Use full depth, not reduced one. This allows
to avoid to do a null search when in the same
position and at the same or bigger depth the
null search failed high.

A very small increase, if any.

After 963 games at 1+0
Mod vs Orig: +158 =657 -147  +4 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-04-04 11:23:18 +01:00
Marco Costalba 2ed3358faf Cleanup pawn storm code
In this form it is even more evident we have some
issue there to be fixed sooner then later....

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-04-02 11:51:39 +01:00
Marco Costalba 08634b06a3 Fix a comment in evaluate.cpp
Function name is wrong.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-04-02 11:45:12 +01:00
Marco Costalba 0e33fc6fd4 Change poll() signature
After previous patch we don't need any more the call parameters.

This fixes a couple of warnings under MSVC.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-31 06:43:12 +01:00
Tord Romstad a5a8830e97 Remove several unnecessary UCI options: All king safety options
except "Aggressiveness" and "Cowardice", and "UCI_ShowCurrLine".
No functional change compared to the previous version with the
default settings.
2010-03-30 15:15:01 +02:00
Marco Costalba 2a14123550 Revert LMR reduction based on thinking time
After 500 games at 5+0 on my QUAD (3 days) there
is no difference with old version, so probably it
is a feature that doesn't scale with search depth.

So revert for now, perhaps we should readd under a
different form.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-28 12:04:41 +01:00
Marco Costalba 8fabd69d4a Small comments tweaks in search.cpp
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-28 11:53:30 +01:00
Marco Costalba 1fc88071d1 Sync static null conditions with real one
Almost no functional change, but it seems more
in line with the meaning of static null pruning.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-28 11:53:16 +01:00
Marco Costalba 7dca461927 Silence a couple of warnings
MSVC complains about an implicit conversion from double to int.

Also small comments tweaks.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-23 23:34:31 +01:00
Joona Kiiski 7618ee2df1 Vary reduction aggressiveness as a function of thinking time
In the beginning use milder reduction and at the end be
more aggressive.

After 1500 games on Joona's QUAD
Mod - Orig: 791 - 720 +16 elo

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-23 23:29:23 +01:00
Joona Kiiski 661d48c27b Base work for different reduction schemes
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-23 23:25:58 +01:00
Joona Kiiski 42de93ac15 Do not return unproven mate scores from null move search
Causes very small functional change which is not observable with
our usual set of test positions.

However change is observable fx. with following position:
4k3/3r4/5Q2/6K1/8/8/8/8 w - - 0 1
go depth 24

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-22 07:27:05 +01:00
Marco Costalba 426f55b78d Use fail soft in null search
If null search fails high return null value instead of beta.

With TT hash there may be a small advantage for fail-soft since
storing slightly better bounds may cause slightly more hash hits.

After 990 games at 1+0
Mod vs Orig +171 =665 -154  +6 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-21 15:12:44 +01:00
Marco Costalba b638f6b035 Remove castleRightsMask[] hack
Array castleRightsMask[] is not static because it can
be different for different positions, so let it be
a Position member data. This allows to remove tricky
hacks to take in account that although it was defined
static it could change.

Theoretically now copying a position is a bit slower because
we need to copy also an array of 64 integers, but because in
split() we don't copy the position anymore, but just keep the
pointer, the added burden is not mesurable even in MP case.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-20 11:59:22 +01:00
Marco Costalba 3de0bc43a2 Retire Position::fast_copy()
It is never used and could be tricky, so remove it.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-20 11:45:04 +01:00
Marco Costalba 9fc602bae7 Updated copyright year to 2010
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-20 11:27:07 +01:00
Marco Costalba 49c50399fe Fix POPCNT detection gcc compile error
Also don't use __cpuid() intrinsic for Intel under
Linux because gives wrong results when detecting HT,
use the gcc version instead. Finally clean up the code.

Error was due to changed __cpuid() signature for
gcc compiler.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-17 21:04:56 +01:00
Marco Costalba a4551c59e0 Fix __cpuid() compile error with gcc
Use same __cpuid() signature used under Windows.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-17 13:22:28 +01:00
Marco Costalba c853b87c08 Add hyper-threading detection
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-14 12:52:58 +01:00
Marco Costalba 92bada1a32 Move __cpuid() definition for gcc in types.h
This will allow to use the function also for other
purposes then detecting POPCNT.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-14 12:52:56 +01:00
Marco Costalba eaed535c5f Introduce captured_piece()
It will be used by future patches and also rearranges some
half cooked code that mistakenly ended up in master in the
past.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-13 11:07:10 +01:00
Marco Costalba 49dfc50b12 Reduce increase progression of aspiration window
Currently, in case of fail high/low we research with
a window increased by 2*AspirationDelta at first
attempt, this patch instead makes the research be
done with an increase of just AspirationDelta size,
in case of a consecutive fail we will widen to
2*AspirationDelta and so on.

After Joona's test:
Orig - Mod: 850 - 890

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-13 11:05:27 +01:00
Marco Costalba c835ee8853 Use separated research counters in root_search()
One for failing highs and one for failing lows, this
should reduce average window size in case of positions
that fail first high and then low (or the contrary).

After ~2000 games on Joona's quad we have:

Mod - Orig: 1012- 975 (+6 elo)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-13 11:00:12 +01:00
Marco Costalba 7ff9678651 Group time management globals initialization
Instead to leave uninitialized or scattered in the code
as is the case for ExtraSearchTime.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-13 10:59:03 +01:00
Marco Costalba 4ef068a506 Highlight that alpha and beta could change in root_search()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 18:17:04 +01:00
Marco Costalba f23a9e8f88 Fix a comment and add an assert in root_search()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 17:51:56 +01:00
Marco Costalba cc2a249952 Retire RootMoveNumber and use FirstRootMove instead
It is more clear why we use that global flag.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 17:34:36 +01:00
Joona Kiiski 55f0d6377f Save mateThreat flag in splitPoint and make use of it
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 12:18:03 +01:00
Joona Kiiski aeb664e0ea Document one test result
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 12:17:41 +01:00
Joona Kiiski 8abdb131c8 Synchronize root_search() with other search routines
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 12:14:46 +01:00
Joona Kiiski 43c93cb151 Remove obsolete code snippet from root_search
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 12:07:37 +01:00
Marco Costalba 8d1d9f7181 Sort again root moves after a fail low
Currently we use original sorting after a fail low to
research at wider window. This patch instead sorts the
moves according to the last available move's scores.

Strangely no functional change, but should be.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 11:43:31 +01:00
Marco Costalba a303bde26c Additional search.cpp cleanup
Changed FutilityMarginsMatrix dimensions to be a power of two
so that compiler can produce a faster accessing code.

Introduced print_pv_info() to remove some redundant code in
root_search()

Remaining stuff is triviality and documentation tweaks.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-03-06 11:14:38 +01:00
Marco Costalba 0f50f10327 Destroy all locks before to exit
And use platform-independent functions
where possible.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-28 23:36:02 +01:00
Marco Costalba 8286e6ded2 We don't need lpThreadId parameter in CreateThread()
Under Windows we use CreateThread() to setup threads and
we pass a pointer to a variable that receives the thread
identifier, but this parameter is optional and we don't
use it, so remove it.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-28 23:36:01 +01:00
Marco Costalba 3a558a3d8b Function init_thread() should return an integer under Windows
It happens that NULL is 0, but the conventional meaning is of
a zero pointer, so repleace with an explicit 0 integer value.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-28 23:36:01 +01:00
Marco Costalba 14dbeb22dd Try bad captures before non-captures
Consider sligtly negative captures as good if at low
depth and far from beta.

After 999 games at 1+0
Mod vs Orig +169 =694 -136  +11 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-28 23:35:51 +01:00
Marco Costalba 68eb7e77f1 Revert previous patch
It raises an assert under Windows, it is not clear why but it
happens that idle_loop() is called with incorrect threadID and
the assert triggered is:

assert(threadID >= 0 && threadID < MAX_THREADS);

So revert the patch for now, but we should understand why it
fails.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-27 17:24:58 +01:00
Marco Costalba 57340c109b Do not wait for sleeping in init_threads()
We can't do it with full guarantee anyway because
there is always a possible race between the setting of
state to THREAD_SLEEPING and actual sleeping.

So just remove the not perfect code to avoid misunderstandings.
This reflects what we have done in wake_sleeping_threads() in
the previous patch.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-27 13:26:04 +01:00
Marco Costalba 111aa44662 Remove an incorrect assert in wake_sleeping_threads()
Currently there is no guarantee that threads are sleeping
when calling wake_sleeping_threads() because put_threads_to_sleep()
returns without waiting for threads to actually sleep.

Assert can be easily triggered calling put_threads_to_sleep() and
wake_sleeping_threads() in a tight loop.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-27 12:01:07 +01:00
Joona Kiiski 29fb389760 Add some commentary
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:43:19 +01:00
Joona Kiiski 0d292d1a2d Clean up common adjustments
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:41:40 +01:00
Joona Kiiski 5bb9da9287 Remove "Threat Depth" ucioption
This option likely has very low meaning for playing strength and style,
so I see no need to keep this configurable

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:41:31 +01:00
Joona Kiiski 34c7f1387d Cleanup steps 12, 14
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:39:08 +01:00
Joona Kiiski fc23466236 Clean up step 11
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:38:34 +01:00
Joona Kiiski 01b228b5e1 Clean steps 8 and 9.
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:37:56 +01:00
Joona Kiiski 2142be7d7f Clean razoring code (step 6)
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:37:09 +01:00
Joona Kiiski 3888d14bd4 Synchronize variable listing of 4 different search routines
search() is used as a "leading star" and other routines
are modified according to it.

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:36:29 +01:00
Joona Kiiski 7bcd97933a Remove current line printing in SMP mode
Was broken and fixing would be too messy.
Now this option is only activated in single thread mode

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-26 00:35:35 +01:00
Joona Kiiski 9d4abbc6eb Synchronize sp_search() with search() part I
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-25 07:30:19 +01:00
Joona Kiiski c3b3dcc31a Rename staticValue to refinedValue
Just to avoid misunderstandings.
True staticValue is available through search stack

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-25 07:30:09 +01:00
Joona Kiiski e6f2d43b8a Fix repetition detection bug
Bug spotted by Jouni Uski and fix suggested by Pablo Vazquez

Also add note that we are not always handling fifty move rule correctly

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:56:48 +01:00
Joona Kiiski 1a03f0b0d3 Synchronize sp_search_pv() with search_pv()
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:42:24 +01:00
Joona Kiiski 62f6d39204 Synchronize sp_search() with search() part II
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:41:51 +01:00
Joona Kiiski 936cd5b83d Simplify locking in splitpoint search
One rule: Always lock before picking up a move.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:37:05 +01:00
Joona Kiiski 3c31776a20 Synchronize search_pv() with search take II
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:36:21 +01:00
Joona Kiiski 0980f43ab0 Synchronize search_pv() with search take I
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:34:55 +01:00
Joona Kiiski 9eedc0a463 Search code documentation, take III
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:30:02 +01:00
Joona Kiiski 195b54c312 Search code documentation take II
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:26:05 +01:00
Joona Kiiski 89b4ad6433 Separate razoring from null move
I cannot see connection between the two.

Also add one FIXME for illogical behaviour

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:24:10 +01:00
Joona Kiiski 77bb9a94ae Split search() in independent sections
I don't know if enumerating sections is a good idea,
but for me code is more readable this way

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:20:26 +01:00
Joona Kiiski 8a78ac84f3 Avoid research in case thread has already been asked to stop
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 19:19:52 +01:00
Joona Kiiski 5c944fb3b4 Add one assert
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 07:28:50 +01:00
Joona Kiiski c974d9ef33 Do not wait for threads falling asleep
I cannot see any reason to do this. Even this is not enough to fix
theoretical race case on Windows which doesn't seem to cause any
problems in practice anyhow

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 07:27:45 +01:00
Joona Kiiski 12feb5866f Remove unnecessary conditions from if-clauses and replace them with asserts
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-24 07:26:28 +01:00
Joona Kiiski 80810e4951 Fix crash in debug mode
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-23 07:23:03 +01:00
Joona Kiiski c67b9916f1 Fix some races
Resurrect extra check for sleeping in POSIX code.
This necessary to prevent ugly races between
thread_broadcast and thread_cond_wait.

After thread has woken up, it marks itself as available.
Another thread must not do this, because of possible race.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-23 07:22:56 +01:00
Joona Kiiski 78c6bb1079 Fix one assert
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-23 07:22:49 +01:00
Joona Kiiski 85e60bfc8e Fix compile errors in debug mode
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-23 07:22:42 +01:00
Marco Costalba 79b57dd4ca Document struct SplitPoint fields constness
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-21 16:10:19 +01:00
Marco Costalba b9537edbb0 Beta is never changed after an sp_search()
So we can use a const value instead of a pointer in
split().

Also pass NULL instead of a faked address of alpha in
case split is called from a non-PV node.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-21 15:32:39 +01:00
Marco Costalba d38f4f61e7 Supress make warning on missing .depend file
This is generated by make itself, so the warning
is useless.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-21 15:18:35 +01:00
Marco Costalba 27c74c5245 Fix an icc warning
remark #1599: declaration hides variable "i" (declared at line 2651)

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-21 15:07:53 +01:00
Marco Costalba 2f2e8a68d8 Code style triviality in split()
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-21 14:05:38 +01:00
Marco Costalba 1ea70dd9dd Fix a warning with POPCNT and MSVC
Intrinsic __popcnt64() returns an unsigned __int64, cast
to an integer and silence the warning.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-21 13:44:36 +01:00
Marco Costalba 13c096f839 Revert "Recursive lock"
Joona says that sp_update_pv() does not pass the split point
boundaries, so there is no risk to corrupt data from another
split point. Also the race on thread_should_stop() is harmless
because of this.

So revert the patch and come back to single lock.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-20 23:59:47 +01:00
Marco Costalba 2de2b76896 Remove a couple of useless thread_should_stop() calls
We test for it anyway few lines below and even under lock
protection.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-20 23:48:54 +01:00
Marco Costalba 8b99e94562 Revert small state change optimization in idle_loop()
Joona says:

1. We should not be afraid of "AllThreadsShouldExit" flag.
Because when this is set to true we _must_not_ be searching (= All
splits must have been undone).
And if we are not searching it's impossible that some other thread
could give us work to do. So setting state to THREAD_AVAILABLE
doesn't do any harm. If you want to add check for this, you could do
it like this:

 if (threads[threadID].state == THREAD_WORKISWAITING)
 {
+    assert(!AllThreadsShouldExit)
     threads[threadID].state = THREAD_SEARCHING;

2a. If waitSp->cpus == 0, setting state to THREAD_AVAILABLE makes
no harm either, because helpful master concept dictates that _only_
our own slave can book us. If we don't have any slaves, noone has the
right to book us.

2b. If point (2a) is not correct then your extra check only adds extra race:
In smp code checking for waitSp->cpus > 0 is not enough. It's possible that
our slave immediately exits and another thread
books us as a slave when our state is still
THREAD_AVAILABLE. So instead of adding extra level of security we have
just introduced extra race.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-20 18:36:07 +01:00
Marco Costalba 512a4e4ff0 Recursive lock all split point's chain
When we found a cut-off then lock all the split point chain,
not only current one to avoid races in case two threads running
on different split points where one is ancestor then the other,
find a beta cut-off at the same time, in this case we want only
one to call sp_update_pv().

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-20 18:29:53 +01:00
Marco Costalba 2da290d72b Retire per-thread stopRequest flag
This is a per split-point request, not per-thread. When we find
a beta cut-off in current thread's split point or in or in some
ancestor of the current split point then threads should stop
immediately the search and return to idle_loop().

The check is done by thread_should_stop() that now looks only
at split point's chain.

No functional change and a good semplification.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-20 17:52:09 +01:00
Marco Costalba b39a24ecca Use state instead of flags to track threads
This is easier to follow and also reduces the points
where state changes to mainly idle_loop() and split().

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-20 16:57:24 +01:00
Marco Costalba 189a005a0b Rename THREAD_MAX in MAX_THREADS
Also rename idle_thread_exists() in available_thread_exists()

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-20 13:38:04 +01:00
Joona Kiiski 7c61b8ad2a Search negative SEE moves in qsearch in PV
After 2704 games on slow single core
mod - orig: 1381 - 1323

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-15 19:15:44 +01:00
Joona Kiiski a093f33154 Use zero null move margin when depth < 4 * OnePly
This is because when we are below 4 * OnePly, the null move
will directly jump to qsearch and if we are below beta,
our opponent is above beta and will get immediate
stand pat cut off.

So basically this patch is just optimizing away useless
evaluation calls. dbg_hit_on() runs show that this heuristic
is correct >99% of cases. Transposition table probably causes
some inaccurary?

After 1148 games on QUAD
mod-orig: 583 - 565 +5 elo

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-15 19:14:41 +01:00
Marco Costalba 0895f1ac71 Fix another setting of a flag out of lock protection
In this case is dangerous because in split() we reset the flag to
false, but if it was set due to a cut-off higher in the tree we
completely miss that and go on with the full search.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-14 16:31:30 +01:00
Marco Costalba b29198354c Rename flag 'stop' in 'stopRequest'
Instead of other flags this is not a state flag, i.e. does
not defines a state for the thread, but a request because
after we raise 'stopRequest' flag the corresponding thread is
not stopped, but continues to run for a while until it returns
from sp_search() in idle_loop.

It is important the name reflects this.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-14 16:04:53 +01:00
Marco Costalba 40a7ffd53f Reset thread flags to a known state before to exit think()
Among them 'stop' and 'printCurrentLineRequest' could have
random value, so reset to a known state before to leave the
search.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-14 15:59:12 +01:00
Marco Costalba 8eae6a95fb Fix 'stop' flag changed out of lock protection
This is the first nice effect of previous patch !

Because thread_should_stop() should be declared 'const' we
need to remove the setting of 'stop' flag to true that
turns out to be a bug because thread_should_stop() is called
outside from lock protection while 'stop' flag is a volatile
shared variable so cannot be changed when not in lock.

Note that this bugs fires ONLY when we use more then 2 threads,
so commonly only in a QUAD or OCTAL machine.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-14 13:43:26 +01:00
Marco Costalba 2b740f5495 Introduce ThreadsManager class
Main aim of this patch is to consolidate all the thread related stuff
behind a single class interface so to avoid messing with global flags
and having thread code scattered among non-thread related stuff.

Another advantage is that now access to thread's variables is
more controlled, in particular we can differentiate between
read and write accesses by the mean of different interfaces, it
is so simpler to understand how a function is related to threads.

Lastly this rewrite is the base for future code consolidations and
semplifications that are easier now that we have only one thread's
access point.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-14 12:53:27 +01:00
Marco Costalba fb5ba1d329 Fix compile error under gcc
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-13 13:40:23 +01:00
Marco Costalba 78e494fcbc Ensure function boundaries for threads state changes
Ensure threads are sleeping when leaving init_threads() and
the newly introduced put_threads_to_sleep().

Also ensure threads are not sleeping when leaving
wake_sleeping_threads().

As a side effect we now leave think() with all the threads
(but the main one) guaranteed to sleep. So when we enter
again in think(), after the opponent next move, we know
threads must be sleeping.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-13 12:40:29 +01:00
Marco Costalba a16415f44d Rename stop_threads() to exit_threads()
More stick to what actually happens.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-13 12:07:13 +01:00
Marco Costalba 8a504d36f9 Be sure threads are woken in wake_sleeping_threads()
Wait inside wake_sleeping_threads() for the threads to be
effectively and reliably woken up.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-13 12:02:34 +01:00
Marco Costalba 093dd8fe88 Use Thread c'tor to properly init the struct
This is what c'tors are for.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-13 11:41:34 +01:00
Marco Costalba 6382324afd Add 'sleeping' flag to struct Thread
Will be used by future patches. Also:

- Renamed Idle in AllThreadsShouldSleep

- Explicitly inited AllThreadsShouldExit and AllThreadsShouldSleep
  in init_thread() instead of use an anonymous global initialization.

- Rewritten idle_loop() while condition to avoid a 'break' statement

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-13 11:28:33 +01:00
Marco Costalba cb08413dc4 Allow build on HP-UX 11.X
Patch from Richard Lloyd (slightly edited from me), following the list
of changes as described by the author:

src/Makefile:
- Added PREFIX and BINDIR for the install: rule.
- Added a "make hpux" line to the help: rule.
- Added "make test"/"make check" rule that runs the $(PGOBENCH) command.
- "make clean" now additionally removes core and bench.txt.
- Added an hpux: rule.
- Added an install: rule to mkdir $(BINDIR), copy $(EXE) to $(BINDIR) and
 then strip it.
- "make strip" now ensures that $(EXE) is built first before trying to
 strip it.
- Hide errors and output from the g++ command used by the .depend: rule and
 then touch .depend in case g++ isn't available.
- Hide errors from the "include .depend" in case .depend doesn't exist
 (e.g. directly after a "make clean").

src/book.cpp and src/book.h:
- HP-UX's aCC really didn't like the const keywords used for the
 Book::file_name() definitions, so they were removed. I checked that this
 didn't affect a Linux build and it was still fine.

src/misc.cpp:
- HP-UX uses <sys/pstat.h> and pstat_getdynamic() to determine the number of
 CPU cores, so added conditional code for that (if pstat_getdynamic() fails,
 set the number of cores to 1).

src/tt.cpp:
- <xmmintrin.h> and _mm_prefetch() seem highly specific to the Intel x86(_64)
 and gcc platforms - neither exist in HP-UX, so conditionally avoid that
 code in HP-UX's case. Perhaps some sort of define is needed here
 such as -DHAS_MM_PREFETCH that could be #ifdef'ed for instead?

Even after these changes, it's more convenient for HP-UX users to edit the
default: rule in the Makefile to run "$(MAKE) hpux" before they build
stockfish, but that's not a big deal if they're warned about that first (the
same applies to all other builds other than the standard "$(MAKE) gcc" one).

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-12 06:49:16 +01:00
Marco Costalba 711ef615c7 Fix a couple of new MSVC 2010 warnings
Compiler complains because in Book we have a d'tor but not
copy c'tor and assignement operator (warning C4511 and C4512),
note that after adding them (just declared) you now need also
default c'tor !

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-10 21:27:44 +01:00
Marco Costalba 74203e181d Retire EvalInfo* in SearchStack
It is an hidden bug waiting to fire. The main problem is
that ss[ply] is overwritten by search() and qsearch() called
from IID and razoring, so that we cannot hold a pointer to a
local EvalInfo variable.

For instance if we go razoring then we overwrite the pointer
with the address of a variable local to qsearch(), when we return
from qsearch() variable goes out of scope and now ss[ply].evalInfo
holds a stale pointer !

Because we are not looking for troubles we go through the
safe route and we remove it entirely.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-08 10:50:17 +01:00
Marco Costalba 97fe0ac777 Small code style triviality
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-07 13:23:00 +01:00
Joona Kiiski 7ae16a193b Implement init_search()
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-07 13:09:03 +01:00
Joona Kiiski 8261f61964 Document lookup tables
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-07 13:08:56 +01:00
Joona Kiiski 6e1cb6e45b Implement futility move count array
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-07 13:08:48 +01:00
Joona Kiiski 4bfa0c429e Implement futility margins matrix
No functinal change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-07 13:08:40 +01:00
Marco Costalba 2e70a2873f Use gain table to order non-captures
Gain value is multiplied by 16 to be of comparable magnitudo
of negative history, on average.

This patch shows very good results in tactical tests, but
started very bad in real games, so I have run two test matches.

After 896 games at 1+0
Mod vs Orig +187 =525 -184 +1 ELO

After 999 games at 1+0
Mod vs Orig +223 =590 -186 +13 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-07 08:55:51 +01:00
Joona Kiiski 9429d2d028 Use posKey instead of pos.get_key() after NonPVIID
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-06 17:40:01 +01:00
Joona Kiiski d44fa46082 Use opening book when pondering
Otherwise we will not use move given by opening book
when we receive 'ponderhit'-command.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-06 17:39:53 +01:00
Marco Costalba c7866a4215 Delay sorting of negative scored non-captures
We can do this only when needed, if we get a cut-off
before we skip sorting entirely. This reduces sorting
time of about 20%.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 18:31:09 +01:00
Joona Kiiski 69644d3f73 Copy 4 SearchStack items in split()
In search routines we use information from previous ply
and init killers two plies ahead.

So for me it seems correct to copy 4 searchstack items
in split:

ply - 1, ply, ply + 1, ply + 2

Because
a) we do not split at root (ply == 0)
b) ply < PLY_MAX and SearchStack size is PLY_MAX_PLUS_2
there should be no risk of underflows or overflows

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 18:06:41 +01:00
Marco Costalba b0858877ae Remove sorting optimization for many zeroes
With negative history we don't have anymore a
lot of zeroes to score, so just split moves in
positives and non-positives sets.

Speed up is almost zero, we cannot test speed directly
because node count changed due to reorder, but I have
verified sorting is correct. With a profiler I have
seen we gain a little in sort_moves() and lose a little
in insertion_sort(), so the net effect is almost zero,
but code is simpler.

No real change, just move reordering.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 18:04:08 +01:00
Joona Kiiski 321f6d1d19 Give FailLow flag more descriptive name
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 15:21:21 +01:00
Joona Kiiski e738fa7d10 Remove Problem variable
It was only used to control StopOnPonderHit variable.
Now use FailLow variable instead.

Patch has a minor effect on time management when ponder is on.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 15:21:12 +01:00
Joona Kiiski c5d546e18e Remove unused failHighPly1 flag
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 15:21:05 +01:00
Joona Kiiski 8d65fcc0f3 Remove unused FailHigh flag
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 15:20:57 +01:00
Joona Kiiski d8e5b8c133 Simplify time management
noProblemFound condition is never true.
This was verified by running 800 games 1+0 match in 1 CPU computer.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 15:20:47 +01:00
Marco Costalba 6fe36d13de Be sure negative see evasions are at the bottom
Because H.move_ordering_score() can return negative values
some negative see moves could be searched before non-negative
see moves with negative history.

This patch restores proper ordering.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 08:19:39 +01:00
Marco Costalba 337ec0f3d0 Score non-captures only by history
Now that history can go negative and is almost alwyas
non zero we have no more reasons to use also psqt term.

After 994 games at 1+0
Mod vs Orig +204 =597 -193 +4 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-05 00:34:17 +01:00
Joona Kiiski 5b1043ee11 Reduction lookup table
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-04 19:09:51 +01:00
Marco Costalba 439aea9ab7 Convert gains to use a piece-to mapping
Instead of piece-from-to, in this way it is similar
to what we already do for history.

Almost no change, but seems a bit simpler in this way.

After 995 games at 1+0
Mod vs Orig +207 =596 -192 +5 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-03 19:33:59 +01:00
Marco Costalba 770db27164 Fix a compile error from previous patch
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-02 19:01:24 +01:00
Joona Kiiski 21d32aa7fe Fix indentations
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-02 18:58:00 +01:00
Joona Kiiski 95d33aef9f Retire outdated aspiration search code
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-02 18:57:50 +01:00
Marco Costalba b5a4edd86f Renamed stand pat as 'static null move pruning'
It seems more standard conformant. Also added a bit of
description directly from Tord.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-02 08:51:54 +01:00
Marco Costalba 35ada63174 Save futilityMargin for both colors
It will be needed by future patches.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-01 20:06:56 +01:00
Marco Costalba 06a695d5b8 Fix duplicated scaling function
We erroneusly added two times the same scaling function
to endgame's map.

Fix detected by valgrind becasue resulted in a memleak
of the first added scaling function.

Bug introduced by 30e8f0c9ad6a473 of 13/02/2009

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-01 19:33:02 +01:00
Marco Costalba 4b55d3d883 Increase TT size limit to 8 GB
We had an overflow due to use an integer for hash size,
now we use a size_t as we should, so we can increase to
an higher limit.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-01 16:53:10 +01:00
Marco Costalba 8008f78415 Check bounds in set_option_value()
Normally it's up to the GUI to check for option's limits,
but we could receive the new value directly from the user
by teminal window. So let's check the bounds anyway.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-02-01 14:17:37 +01:00
Marco Costalba 3941e4bdb3 Some code style triviality in root search
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-31 11:14:42 +01:00
Marco Costalba d7d2c1b7e3 Add hardware POPCNT support for gcc
With new target 'make gcc-popcnt' it is now
possible to compile with enabled hardware POPCNT
support also with gcc. Until now was possible only
for Intel and MSVC compilers.

When this instruction is supported by CPU, for instance
on Intel i7 or i5 family, produced binary is a bit faster.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-31 09:39:32 +01:00
Marco Costalba 17502a5659 Revert "Remove pointless gcc flag when generating dependencies"
This reverts commit c43c5fe9e0.

Produces following build error after 'make clean', 'make icc' under Mandriva
with icc version 11.0

Makefile:306: .depend: No such file or directory
In file included from tt.cpp:28:
/usr/lib/gcc/i586-manbo-linux-gnu/4.3.2/include/xmmintrin.h:35:3: error: #error "SSE instruction set not enabled"

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-30 10:37:03 +01:00
Marco Costalba e84488ed6b Retire captures pruning
Futility captures alone does not seem an improvment.

Perhaps is a combination of stand pat + futility that is winning,
so revert for now and continue testing starting from a standard
base until we find the correct receipe.

After 999 games at 1+0
Mod vs Orig +231 0506 -201  +10 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-30 09:59:28 +01:00
Marco Costalba 252844e899 Avoid search tree explosion in qsearch
Under some rare cases we can have a search tree explosion
due to a perpetual check or to a very long non-capture TT
sequence.

This avoids the tree explosion not following TT moves that
are not captures or promotions when we are below the
'generate checks' depth.

Idea suggested by Richard Vida.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-29 16:47:04 +01:00
Joona Kiiski b651e5334b Correct qsearch() TT save
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-29 11:03:06 +01:00
Marco Costalba bd358533a4 Stricter conditions in main search stand pat
Not a biggie but is a reduced pruned patch that doesn't
seems to hurt, so it is welcomed ;-)

After 999 games at 1+0
Mod vs Orig +207 =601 -191  +6 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-29 00:12:03 +01:00
Marco Costalba e4d3a15656 Use float instead of double in reduction parameters
This is faster on 32 bit CPU and precision is enough.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 12:47:48 +01:00
Marco Costalba 1d7a3f26e0 Micro optimize reduction_parameters()
At ply == OnePly (common case) we avoid some useless
floating point computation.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 12:39:15 +01:00
Marco Costalba a0005ba45f Avoid to calculate reduction for each move
This is slow because some floating point operation is
involved.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 12:34:28 +01:00
Joona Kiiski bdd61b1744 Remove useless variable 'PostFutilityValueMargin'
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 11:17:59 +01:00
Joona Kiiski 973e574e1f Precalculate FutilityMargins
This way we don't need to copy+paste formula everywhere

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 11:16:56 +01:00
Joona Kiiski 27393ebae2 Use calculate_reduction() function to simplify code
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 11:15:41 +01:00
Joona Kiiski 3c7eebb48d Bugfix: reduction was not set to zero in full depth search
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 11:15:18 +01:00
Joona Kiiski 548bae80bd Implement calculate_reduction function
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-28 11:14:43 +01:00
Joona Kiiski 2360c8aa2f Standardize set_option function
Previously input like "setoption name Use Search Log value true "
(note space at the end of the line) didn't work.

Now parse value same way as option name. This way we implicitly
left- and right-trim value.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 20:03:44 +01:00
Joona Kiiski 32bd6e44f0 Do not initialize RootPosition at startup
Initializing high-level object at startup is very dangerous,
because low-level snippets are not yet initialized.

For example Position's constructor calls find_checkers() which
calls attackers_to() which depends on various global bitboard arrays
which are not yet initialized. I think we are lucky not to crash.

RootPosition.from_fen(StartPosition); is called immediately after
all initializations are made at uci_main_loop() which is the
correct behaviour

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 20:00:28 +01:00
Marco Costalba c83fd08fd4 Aspiration window rewrite
Joona new aspiration window. Main idea is to always
research aspiration fail highs/low at the same
ply and use much smaller aspiration window than previously.

Testing result is very positive.

1CPU:
953-1149

4CPU:
545 - 656

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 19:59:34 +01:00
Marco Costalba 62b43130e2 Be sure we exit while loop with lock held
This fixes an hang introduced by recent locking
rewrite patch.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 19:23:16 +01:00
Joona Kiiski cd112ee8eb Fix capture pruning
We forgot to update bestValue previously

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 18:58:42 +01:00
Marco Costalba bb968fd42a Simplify locking in sp_search and sp_search_pv
Avoid to take the lock two times in a tight sequence, the first
in get_next_move() and the second to update sp->moves.

Do all with one lock and so retire the now useless locked version
of get_next_move().

Also fix some theorical race due to comparison sp->bestValue < sp->beta
is done out of lock protection. Finally fix another (harmless but time
waster) race that coudl occur because thread_should_stop() is also
called outside of lock protection.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 17:58:04 +01:00
Marco Costalba 307909bed8 Temporary revert "captures pruning" due to an assert
In debug run with 2 threads it happens to be following
assert after some minutes:

assert(value > -VALUE_INFINITE && value < VALUE_INFINITE);

in search(), line 1615.

I am not able to understand why, anyhow reverted for the moment.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 17:57:04 +01:00
Marco Costalba b95ba7b37e Added some FIXME to track needed tests
This avoid us to forget some very needed tests now that
futility has changed in a whole big chunk we need to fine
tuning every splitted change.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 11:49:03 +01:00
Marco Costalba f37741cc83 Integrate gains table in History
This will be useful to use gains table in move
ordering along with history table.

No functional change and big code remove.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 11:22:38 +01:00
Marco Costalba 54b3d44194 Introduce update_gains() and refactor some code
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 11:09:07 +01:00
Joona Kiiski ea53006a9d Fix some silly bugs
SelectiveDepth was ignored

Test results for the whole futility pruning series:

4CPU:
Orig - Mod: 959 - 1027 (+12 elo)

1CPU:
Orig - Mod: 763 - 830 (+15 elo)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 10:14:20 +01:00
Joona Kiiski 6247f27a05 MaxGain based futility pruning for captures
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:57:51 +01:00
Joona Kiiski 5b2fc1e1c0 MaxGain based pruning
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:57:41 +01:00
Joona Kiiski 05a8c318b8 Implement post futility pruning
and prevent futility pruning from pruning
castling moves

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:57:31 +01:00
Joona Kiiski 22b8dc8c98 Collect MaxGain statistics
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:57:24 +01:00
Joona Kiiski cfe59de27d Implement MaxGain table
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:57:14 +01:00
Joona Kiiski c43c5fe9e0 Remove pointless gcc flag when generating dependencies
When generating dependencies there is absolutely no point
to pass -msse flag to gcc, so remove it.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:53:30 +01:00
Joona Kiiski d0daa16769 Remove InfiniteSearch hack
With current search control system, I can see absolutely no
reason to classify fixed time search as infinite search.

So remove old dated hack

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:52:47 +01:00
Joona Kiiski 87303d7ed3 Remove last use of uip.eof()
Value of uip.eof() should not be trusted.
input like "go infinite searchmoves " (note space in the end of line)
causes problems.

Check the return value of (uip >> token) instead

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-27 09:52:15 +01:00
Marco Costalba cf9bf4e58f Reduce lock contention in sp_search_pv()
In less then 1% of cases value > sp->bestValue, so avoid
an useless lock in the common case. This is the same change
already applied to sp_search().

Also SplitPoint futilityValue is not volatile because
never changes after has been assigned in split()

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-26 12:12:11 +01:00
Marco Costalba 5ca4284027 Fix a possible crash in thread_is_available()
When we have more then 2 threads then we do an array
access with index 'Threads[slave].activeSplitPoints - 1'
This should be >= 0 because we tested the variable just
few statements before, but because is a shared variable
it could be that the 'slave' thread set the value to zero
just after we test it, so that when we use the decremented
variable for array access we crash.

Bug spotted by Bruno Causse.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 23:34:21 +01:00
Marco Costalba dac1bcab90 Small split() cleanup
Unify start loop for master and slave threads. Also guarantee
that all the 'stop' flags are set to false before first slave
is started, should be no harm because only master thread can
reset 'stop' flag of slaves to true, so should be no race but
better safe then sorry.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 19:59:33 +01:00
Marco Costalba 81cfd81366 Prune evasions with negative SEE in qsearch
Only pure blocking evasions are candidate
for pruning.

After 998 games at 1+0

Mod vs Orig +215 =596 -187  +10 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 16:31:56 +01:00
Marco Costalba d844a75d2c Avoid copy a Position to get a move's san notation
In move_to_san() we create by copy a new position just
to detect if move gives check. This could be very costly in
line_to_san() that calls move_to_san() for every move, so
create the position only once and pass a reference to move_to_san()

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 16:22:50 +01:00
Marco Costalba c5e71f5150 Fix a race in idle_loop() exiting
When pondering threads are put to sleep, but when thinking
the threads are parked in idle_loop in a tight polling loop
checking for workIsWaiting falg.

So before we set the slave's flag workIsWaiting we have to
guarantee that all the slave data is already setup because
slave can start in any moment from there.

Rearrange the last loop to fix this race.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 15:02:06 +01:00
Marco Costalba c5858ff9ae In split() release the lock before slow search stack copy
Once we have allocated our slave threads and we have removed
master from available threads we can safely remove the lock
so that the lenghty search stack copy operation will not
impact lock contention.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 14:12:30 +01:00
Marco Costalba 0ff91e16da Do not copy master position in split()
A pointer is enough because after a split point has been
setup master and slaves thread end up calling sp_search() or
sp_search_pv() and here a full copy of split point position is
done again, note that even master does another copy (of itself)
and this is done before any do_move() call so that master Position
is never updated between split() and sp_search().

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 12:46:08 +01:00
Marco Costalba c2df60048e Use fast_copy() instead of full copy in sp_search
And detach splitPoint Position from the master one.

So we duplicate StateInfo only once in split() instead
of one for each thread in sp_search

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 12:13:16 +01:00
Marco Costalba 84ec1f7331 Better document how Position c'tor works
Renamed a bit the functions to be more clear what
we actually are doing when we craete a Position object
and explained how StateInfo works.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 12:04:00 +01:00
Marco Costalba b1ac6c69a0 Fix a couple of MSVC casting warnings
Also removed some trailing whitespaces and aligned
indentation to current standard.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 11:27:32 +01:00
Marco Costalba 56e09b4cc8 Copy only the search stack tail in split()
Only the previous, the current and the next ply SearchStack
are copied.

This reduces split overhead especially at low depth (high ply)
and with many threads.

Possibly no functional change (it is not easy to prove in SMP)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-25 11:11:22 +01:00
Tord Romstad 5ed0a60203 Merge branch 'master' of ssh://free2.projectlocker.com/sf 2010-01-24 16:09:54 +01:00
Tord Romstad 1588a4e846 Fixes a Chess960 bug when playing with more than one search thread.
The init_eval() function corrupted the static array castleRightsMask[]
in the Position class, resulting in instant crashes in most Chess960
games. Fixed by repairing the damage directly after the function is
called. Also modified the Position::to_fen() function to display
castle rights correctly for Chess960 positions, and added sanity checks
for uncastled rook files in Position::is_ok().
2010-01-24 16:09:32 +01:00
Marco Costalba 5894c759cd Check for thread creation successful completion
It is a good programming practice to verify a system
call has indeed succeed.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-24 14:33:51 +01:00
Marco Costalba 3975a2b94f Fix some races in SMP code
When a search fails high then sp->alpha is increased and
slave threads are requested to stop.

So we have to check for a stop request before to start a search
otherwise we could end up with sp->alpha >= sp->beta
leading to an assert in debug run in search_pv().

This patch fixes the assert and get rid of some of possible races.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-24 14:30:09 +01:00
Tord Romstad eb6ddd54f1 Make sure we make a move at the end of the search when reaching
maximum depth during a "go movetime ..." search. This prevents
Stockfish from hanging forever after finding a mate in two or
three while running a test suite at a level of a few seconds
per move.

No functional change when playing games at normal time controls.
2010-01-22 13:42:33 +01:00
Marco Costalba 01ebb3d996 If near beta generate checks at -OnePly
In qsearch() try to get a cutoff with the help of an
extra check if we are already very near.

Small increase in actual games but a good result in tactical
test sets where this patch makes SF more tactical.

Mod vs Orig +197 =620 -181 +6 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-21 00:31:01 +01:00
Marco Costalba 285df57a9a Retire LMRPVMoves and LMRNonPVMoves
Are no used anymore.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-20 18:43:29 +01:00
Marco Costalba 806c1d8723 Fix enum Value issue with gcc 4.4
Louis Zulli reports a miscompile with g++-4.4 from MacPorts.

Namely enum Value is compiled as unsigned instead of signed integer
and this yields an issue in score_string() where float(v) is incorrectly
casted when Value v is negative.

This patch ensure that compiler choses a signed variable to store a Value.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-19 23:30:23 +01:00
Marco Costalba a1b8c8109b Small lnArray[] cleanup
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-19 11:52:13 +01:00
Marco Costalba c01af56769 Silence some silly MSVC warnings
Value is never used un-initialized, but MSVC is not
smart enough to detect itself :-(

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-19 11:23:00 +01:00
Marco Costalba 66d5c13a88 Order check moves used in qsearch
Use the same scoring system used for evasions. Small if any
increase, but should be in at least for completeness.

After 999 games at 1+0
Mod vs Orig +208 =601 -190 +6 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-19 11:22:49 +01:00
Marco Costalba 0edad63b44 Avoid an useless evaluate() call
Now that we have position static score we don't
need to call evaluate() a second time.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-18 15:47:43 +01:00
Marco Costalba b833c8247a Allow SearchStack to link an EvalInfo object
This will allow to have wider access to attack
information, for instance from MovePicker.

Note that 'eval' field become obsolete, it is kept
just becasue when we get a position score from TT
we update 'eval' even without an EvalInfo object.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-18 10:03:09 +01:00
Joona Kiiski 419c5b69ca Make reduction search code SMP-friendly
In sp_search_pv() we do a LMR search using sp->alpha, at the end
we detect a fail high with condition (value > sp->alpha), but if
another thread has increased sp->alpha during our LMR search we
could miss to detect a fail high event becasue value will be equal
to old alpha and so smaller then new one.

This patch fixes this SMP-bug and changes also the non SMP versions
of the search to keep code style in sync.

Bug spotted by Bruno Causse.

No functional change (for single CPU case)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-18 09:11:58 +01:00
Joona Kiiski 72e1e9b986 Small cleanup of unused code in sp_search
futilityValue is now calculated immediately after
staticValue, so remove small bunch of unused code

No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-17 14:21:14 +01:00
Marco Costalba 1062459029 Fix silly MSVC warning
MSVC raises an "use of partially uninitialized variable" for futilityValue
and staticValue but this is not rue becasue when !isCheck variables
are never used, anyhow silence the warning.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-17 13:54:52 +01:00
Marco Costalba b5d38ad1e5 Initialize futilityMargin in EvalInfo c'tor
This is less prone to bugs because now it's up to the
compiler don't forget this important initialization.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-17 13:24:40 +01:00
Joona Kiiski 000a975eaf Retire quick_evaluate()
No change in functionality signature

The only functional change is that when we reach PLY_MAX,
we now return VALUE_DRAW instead of evaluating position.

But we reach PLY_MAX only when position is dead drawn and
transposition table is filled with draw scores, so this
shouldn't matter at all.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-17 13:15:44 +01:00
Joona Kiiski d457594197 Razor at depth one, but do razoring only when not in check
This way razoring is always based on exact evaluation and
follows simple formula.

Joona's test results are positive:

32-bit 1CPU:

Mod - Orig: 1073 - 993

64-bit 4CPU:

Mod - Orig: 759 - 721

Functionality Signature: 11448962

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-17 11:46:40 +01:00
Marco Costalba 942c18ef66 Allow negative history values
Don't clamp to zero if a move continues to fail.

After 946 games at 1+0
Mod vs Orig +208 =562 -176 +12 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-13 22:10:43 +01:00
Marco Costalba 007285be2d Store node evaluation in SearchStack
This info will be used by future patches.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-11 21:18:30 +01:00
Marco Costalba e817a55bc6 Decrease NullMoveMargin and adjust razoring
Also retire razoring margins vector and use
a simpler formula instead.

Now that we use a more accurate static evaluation
try to avoid useless null searches when we are well
below beta. And for teh same reason increase a bit
the razoring.

After 972 games at 1+0
Mod vs Orig +224 =558 -190 +12 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-11 21:17:02 +01:00
Marco Costalba 638f3d31cc Do not wait when AbortSearch is set
It means we have already received "stop" or "quit" commands.

This fixes an hang in tactical test in Fritz GUI. Bug
introduced by previous bug fix :-(

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-11 11:19:12 +01:00
Marco Costalba 55745f4105 Fix sending of best move during an infinite search
According to UCI standard once engine receives 'go infinite'
command it should search until the "stop" command and do not exit
the search without being told so, even if PLY_MAX has been reached.

Patch is quite invasive because it cleanups some hacks used
by fixed depth and fixed nodes modes, mainly during benchmarks.

Bug found by Pascal Georges.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-10 13:18:43 +01:00
Marco Costalba 968c3de8e0 Fix threads count setting
Was broken after "Optimal tune for 8 cores" patch.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-09 16:49:54 +01:00
Marco Costalba 6a6facc058 Optimal tune for 8 cores
After deep tests Louis Zulli found on his OCTAL machine that
best setup for an 8 core CPU is as following

"Threads" = 8
"Minimum Split Depth" = 6 or 7 (mSD)
"Maximum Number of Threads per Split Point" = not important (MNTpSP)

Here are testing results:

mSD7 (8 threads) vs mSD4 (8 threads): 291 - 120 - 589
mSD6 vs mSD7: 168 - 188 - 644
mSD6-MNTpSP5 vs mSD6-MNTpSP6: 172 - 172 - 656
SF-7threads vs SF-8threads: 179 - 204 - 617

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-08 12:58:51 +01:00
Marco Costalba 5b21c10afb Sync qsearch with search
So to have the same layout and be as much similar as
possible. The only functional change is that now we
try ttMove as first also in PV nodes and at the end
we save the ttMove, as it happens in search. This
should have almost zero impact on ELO but it seems
the correct thing to do.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-08 12:03:47 +01:00
Marco Costalba dba072c449 Use full evaluation in null search
This is an important design change because we know
compute evaluation in each node.

This is a 2.0 type change!

After 977 games at 1+0

Mod vs Orig +236 =538 -202 51.74%  +12 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-08 10:20:35 +01:00
Joona Kiiski e2e360fcbc Slow down reductions
After testing on Joona QUAD the whole LMR series:

Orig - Mod: 335 - 405 (+33 elo)

Functionality Signature: 12581900

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-07 15:49:48 +01:00
Joona Kiiski 49b1c5dccd Use logarithmic LMR also at root
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-07 15:40:04 +01:00
Joona Kiiski c99d963fa5 Logarithmic LMR
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-07 15:35:57 +01:00
Marco Costalba 7c0679ad61 Fix 'position ..... moves ' parsing bug
If after 'moves' there is a space then we crash.

The problem is that operator>>() trims whitespaces so that
after 'moves' has been extract we are still not at eof()
but remaining string contains only spaces. So that the next
extarction operation uip >> token ends up with unchanged token
value that remains 'moves', this garbage value is then feeded
to RootPosition.do_move() through move_from_string() that does
not detect the invalid move value leading to a crash.

This bug is triggered by Shredder 12 interface under Mac that
puts a space after 'moves' without any actual move list.

Bug fixed by Justin Blanchard

After reviewing UCI parsing code I spotted other possible weak
points due to the fact that we don't test if the last extract
operation has been succesful. So I have extended Justing patch
to fix the remaining possible holes in uci.cpp

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-07 11:59:32 +01:00
Marco Costalba cc974fa7a4 Fix en-passant parsing from fen string
According to standard en-passant is recorded in fen string regardless
of whether there is a pawn in position to make an en passant capture.

Instead internally we set ep square only if the pawn can be captured.
So teach from_fen() to correctly handle this difference.

Bug reported and fixed by Justin Blanchard.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-06 10:07:07 +01:00
Marco Costalba 807844eab1 Introduce refine_eval()
Try to get a position evaluation better then
the quick one with the help of the TT table.

This allows the null search conditions and
chosen reductions to be more accurate.

After 908 games at 1+0
Mod vs Orig +209 =526 -173 +14 ELO

Functionality Signature: 16627355

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-05 14:57:59 +01:00
Marco Costalba 860260c3b1 Increase null reduction at high depths
Linear rule, less aggressive then Dann's one.

It seems it scales well with depth. We will need to
verify against weaker engine if it keeps the score.

After 999 games at 1+0 on my Dual Core
Mod vs Orig +232 =534 -207  +9 ELO

After 1000 games by Martin Thoresen on his QUAD at 1+0
Mod vs Orig 521/479 52.10%

Functionality Signature: 17655312

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-05 08:09:09 +01:00
Marco Costalba d11af1de11 Fix a compile error under gcc
And some warnings on the picky icc.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-04 17:19:42 +01:00
Marco Costalba 721d557681 Last round of search.cpp cleanup
The most interesting thing is a bit of rewrite
and semplification in connected_moves()

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-04 12:39:13 +01:00
Marco Costalba 0e15b0f1d3 Space inflate bottom part of search.cpp
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-03 21:30:46 +01:00
Marco Costalba 9e6d38d224 Rename MaxActiveSplitPoints
And move in thread.h togheter with THREAD_MAX

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-03 15:37:23 +01:00
Marco Costalba c8af7a867e Retire 'finished' from MovePicker
It is not useful becasue it is safe to call
get_next_move() multiple times when phase == PH_STOP

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-03 15:37:17 +01:00
Marco Costalba 5c8f571459 Rename SingleReplyExtension in SingleEvasionExtension
Because that's the correct meaning. Note that also the
corresponding UCI option has been renamed.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-03 15:37:03 +01:00
Marco Costalba 0256db2a11 Small cleanup in search.cpp
Also clarify some comments.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-03 15:36:40 +01:00
Marco Costalba 12d8f74242 Retire approximateEval field from SplitPoint
It is not used anymore after the futility pruning
rewrite in a66f31f12.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2010-01-02 10:30:05 +01:00
Marco Costalba 7824603549 Double HistoryMax and reduce aging
After history accounting rewrite in 1.6, a small
tweak of history parameters seems positive.

Note that these are not to be considered the optimal
values, just a wild guess that proved good.

Finding the optimal values would require a much longer
testing time.

After 967 games at 1+0

Mod vs Orig 240 529 198 +15 ELO

Functionality Signature: 21222553

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-31 13:45:57 +01:00
Marco Costalba 0f39e5c4ff Fix a little warning under gcc compiler
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-30 13:25:22 +01:00
Marco Costalba 990d83a72d Optimized bitScanReverse32()
Should be a bit faster then previous one.
Hacked by Pascal Georges.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-30 13:25:21 +01:00
Marco Costalba cf486cf229 Restore development version
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-30 13:25:21 +01:00
Marco Costalba e0a8b36436 Stockfish 1.6.2
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-30 13:25:20 +01:00
Marco Costalba 8d724220a7 Better fix for gcc optimization issue
According to the standard, compiler is free to choose
the enum type as long as can keep its data.
Also cast to short and right shift are implementation
defined in case of a signed integer.

Normally all the compilers implement this stuff in
the "usual" way, but gcc with -O3 and -O2 pushes
aggressively the language to its limits to squeeze
even the last bit of speed. And this broke our
not 100% standard conforming code.

The fix is to rewrite the Score enum and the 16 bits
word extracting functions in a way that is 100% standard
compliant and with no speed regression on gcc and also on
the other compilers.

Verified it works on all compilers and with equivalent
functionality.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-30 13:25:02 +01:00
Marco Costalba 0973cc2ef6 Score enum should be at least 32 bits
The compiler is allowed to chose the size of an enum variable
based on the values it is expected to store. So force the compiler
to use at least a 32 bit integer type for the Score.

MSVC and Intel do not change, while gcc under -O3 is affected
by this change.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-27 19:45:19 +01:00
Marco Costalba 3f14f9a478 Revert small pop_1st_bit() optimization
We cannot cast a pointer type to an unrelated pointer type.
This is a violation of the strict aliasing rules.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-27 14:07:08 +01:00
Marco Costalba aa86d81f79 Remove a bogus assert
It is not clear why is not true, even in single thread
case, but as a matter of fact it is not!

So remove it.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-27 13:54:46 +01:00
Marco Costalba b884351cc7 Use THREAD_MAX instead of hardcoded 8
This will allow to change THREAD_MAX value in the future.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-27 13:52:29 +01:00
Marco Costalba 4d9e9ac3d4 Restore development version
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-27 08:35:44 +01:00
Marco Costalba 3dc9f95225 Set maximum hash table size to 2GB
We cannot allocate more then 2 GB, so let the limit
reflect this.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-27 00:44:08 +01:00
Marco Costalba bc0871acbc Stockfish 1.6.1
Workaround a gcc optimization bug.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-26 19:39:22 +01:00
Marco Costalba 2643f1552f Workaround optimization bug in gcc
Unfortunatly we need to slow down to -O1 to be sure
it works always.

Note that sometime it works also with -O2 or even -O3,
but user has to try himself.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-26 19:39:13 +01:00
Marco Costalba ba07b95ee0 Fix description of Score enum
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-26 19:39:04 +01:00
Marco Costalba ef58551a2d Fix a typo in ReducedStateInfo
It happened to work by accident because Score and
Value are both integer.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-26 19:38:53 +01:00
Marco Costalba 7d34e7bf84 Stockfish 1.6
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-22 22:11:10 +01:00
Marco Costalba 12aeac5e14 Score definition gives a compile error under gcc
For enum definitions a parenthesis is required.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-21 11:44:11 +01:00
Marco Costalba 734fb9a13b Setup Release Candidate 1
To be used by Jim for testing different compiles settings.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-19 17:00:33 +01:00
Marco Costalba c12364bb67 Fix a comment in HistoryMax description
Was obsoleted out some time ago.

Spotted by Justin Blanchard

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-19 10:43:12 +01:00
Marco Costalba ad5b5cef4a Fix book name is hard coded as book.bin
Instead should be read by the corresponding UCI
option "Book File".

Bug reported and fixed by Justin Blanchard (Arch Linux)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-18 10:11:50 +01:00
Marco Costalba 0d88b832e3 In non-PV IID don't call evaluate when in check
Was a long standing hidden bug from Glaurung times,
triggered only now that we enable IID at non PV nodes.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-15 12:07:23 +01:00
Marco Costalba 14c3da5cad Fix a compile error in debug mode
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-15 11:52:57 +01:00
Marco Costalba 0b9b34655f Enable IID at non-PV nodes
We want to increrase the opportunities
of doing an exclusion search.

After 999 games at 1+0
Mod vs Orig +216 =574 -209 50.35%  503.0/999  +2 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-14 11:44:37 +01:00
Joona Kiiski a66f31f129 Synchronize pruning rules in search and sp_search
Regression test passed:

Mod - Orig: 365 - 351

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-13 10:03:46 +01:00
Marco Costalba 2161d8b0b3 Remove history counters
Instead decrement history value on failure.

After 999 games at 1+0

Mod vs Orig  +236 =558 -204 51.60% +11 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-12 19:23:10 +01:00
Joona Kiiski 7bc72d092f Fix overflow risk in split point
Sizeof of search stack should be PLY_MAX+2 instead of PLY_MAX.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-10 19:00:19 +01:00
Joona Kiiski b056e5d40a Re-enable TT.insert_pv()
This time make sure that valuable TTentries are not overwritten.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-09 19:43:12 +01:00
Joona Kiiski d0b8bc5fdf Disable insert_pv
This way we avoid overwriting valuable TT entries which
are needed to calculate exclusion search extension for pv.

Mod - Orig: 483 - 410 (+28 elo!)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-08 18:13:38 +01:00
Marco Costalba 9a46ac6b2c Set IncrementalFutilityMargin to 8
After 999 games we are almost equal (+2 ELO),
but we have a good result against Rybka

Rybka 2.3.2a mp 32-bit vs Mod  254.5 - 242.5 +152/-140/=205 51.21%
Rybka 2.3.2a mp 32-bit vs Orig 259.5 - 236.5 +151/-128/=217 52.32%

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-08 11:16:14 +01:00
Joona Kiiski 0fc9d9ef61 Replace 100 with PLY_MAX in ok_to_use_TT
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-03 15:12:50 +01:00
Joona Kiiski bd618941ce Adjust SingleReplyMargin 0x64 -> 0x20
Mod - Orig: 920 - 890 (+6 elo)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-02 19:49:50 +01:00
Marco Costalba 403db5a6e9 Don't clear hash at the beginning of a new game
After 900 games at 1+0
Mod vs Orig +217 =480 -196 +8 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-02 15:13:13 +01:00
Marco Costalba 3f3365221b Try to prune also when approximateEval < beta
Now we always try to filter out moves, we will have
more wasted evaluation calls, but also more pruned
nodes.

After 786 games

Mod vs Orig +196 =413 -177 +8 ELO

Verified also against Rybka it increases score to 50-51%

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-12-01 11:49:33 +01:00
Joona Kiiski ae0b965711 Do not crash if we are asked to search mate or stalemate position.
We might be asked to ponder mate or stalemate position.
This being the case, simply wait for stop or ponderhit.
Currently we crash.

UCI specs aren't clear on the issue, but it cost nothing to
add little check.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-28 21:41:57 +01:00
Marco Costalba 455993b289 Fix get_option_value() for strings with spaces
Problem is that

istream& operator>> (istream& is, char* str );

according to C++ documentation "Ends extraction when the
next character is either a valid whitespace or a null character,
or if the End-Of-File is reached."

So if the parameter value is a string with spaces the currently
used instruction 'ss >> ret;' copies the chars only up to the first
white space and not the whole string.

Use a specialization of get_option_value() to fix this corner case.

Bug reported by xiaozhi

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-28 17:31:56 +01:00
Marco Costalba c7a77dd3c0 Retire FutilityMargins[] array
Now we use a formula to calculate margins on the fly.

Node count has changed because we fixed a leftover when
we still where using FutilityMargins to calculate futilityValue
in the case that we had the evaluation score in TT.

Also small indentation fix.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-28 11:56:08 +01:00
Marco Costalba af3dd21e90 IncrementalFutilityMargin to 4 and increased pruning
Increase pruning at low depths while tone downa bit at
higher depths (linearize a bit the logaritmic behaviour)

This goes togheter with IncrementalFutilityMargin decreased
to 4 compensate the bigger pruning effect.

Total pruned nodes are more or less the same. We go from 36%
of nodes after prune to 37% with this patch.

After 999 games at 1+0
Mod vs Orig +250 =526 -223 +9 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-27 21:17:34 +01:00
Joona Kiiski 1a7047f544 Drop OnlyMoveExt PV-condition from 8 plies to 6 plies
Orig - Mod: 731 - 750

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-27 15:56:26 +01:00
Marco Costalba 25c22ffe7a Use move not ttMove in exclude search
If we arrive until the exclusion search call then
we know that move == ttMove == tte->move()

But using ttMove in search call while, during excluded search
conditions we have used tte->Move()could be a little bit suboptimal.
On the other side using tte->move() also in search call is a bit ugly
so opt for the third choice that is the most clean becasue from the
conditions the reader easily understands that we are talking of ttMove
and that we ant to exclude the move we are evaluating in that moment.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-27 15:38:51 +01:00
Marco Costalba ae6157fcf3 Better document previous patch
If tte->move() != MOVE_NONE then tte->move() == ttMove

What could happen is that we have a ttMove without a tte, or,
we have a tte but tte->move() == MOVE_NONE

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-27 11:54:07 +01:00
Marco Costalba 5bec768d42 Fix a possible crash in excluded search condition
Due to IID we could have a ttMove and not a tte, or,
even if we have a tte they could belong to different
searches so that the depth and type of tte don't
have the same origin of the ttMove.

To fix this we always use tte entry in excluded search
condition and, after an IID, we reprobe the TT table.

No functional change. Apart from possible crash fix.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-26 13:58:55 +01:00
Marco Costalba 6ae30e7cb1 Document why we don't use TT to prune in search_pv()
From a Joona' s post on talkchess.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-25 17:42:52 +01:00
Joona Kiiski 5ea8167921 Revert last Only move extensions tweaks
They gave bad results:

Mod - Orig: 361 - 404

Master is now verified to be functional equivalent with F_63

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-25 17:42:51 +01:00
Marco Costalba 55b5b03273 Speed up sorting of non-captures
Becasue we have a lot of zero scores (around 30% of moves)
it is a good idea to do a couple a presorting loops across
the move list and shuffle the moves a bit so that with a
small effort we end up with 3 groups of moves: positives
scores, zero scores and negative scores.

We have two advantages

1) We don't need to sort zero scores

2) Sort two small groups is faster then sort a single big one

Speed up is of about 2%

Because equal scored moves could be reordered in a different way
this is not a "no functional change" although I have verified
the output list is always correctly sorted.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-25 17:42:41 +01:00
Marco Costalba 850c021f86 Rewrite messy LSN-code take 2
We already reset loseOnTime flag at the beginning of
a new game, so we can simplify a bit the ligic there.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-25 14:35:32 +01:00
Joona Kiiski cd0c7373cd Rewrite messy LSN-code
* New version is documented and logic should be easier to follow
* Add extra check to not use LSN with x moves / y seconds time control
* New code fixes some rear cases where old code (still) causes program to lose on time at move 1.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-25 12:10:47 +01:00
Joona Kiiski 9a59535962 Remove RootMoveList::scan_for_easy_move()
* The function is called only in one place
* It must not be called elsewhere
* The function call easily replaced with simple one line condition

No functional change (tested with usual set + 2000 random positions)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-25 11:57:20 +01:00
Joona Kiiski 48246468f2 Remove 2 FIXMEs from search.cpp
* First one is without any documentation, code is working just fine,
  so there seems to be nothing that really should be fixed.

* Second one requesting emergency measures on aspiration fail low
  when we are running out of time and we are without good move.
  After very long time, I've come to conclusion that this is
  impossible to fix, so remove request.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-25 11:52:09 +01:00
Marco Costalba 2b6bc70f7b Document and cleanup new effective-single-reply code
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 21:12:49 +01:00
Joona Kiiski 8b3fdec7ec Always extend full ply in PV
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 21:02:34 +01:00
Joona Kiiski 58452de86d Add mild extension in low depths
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 21:02:27 +01:00
Joona Kiiski 93c9f342ca Fix currentMove bug
Orig vs Master: +15 elo 887.5 - 812.5 (1700 games, finished) [4CPU]

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 21:01:05 +01:00
Joona Kiiski 16acf57773 Only move extension based on exclusion search
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 21:00:13 +01:00
Joona Kiiski 77eec9f9cb Base work for exclusion search
No functional change

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 21:00:05 +01:00
Marco Costalba bc35f4c42d Tone down a bit futility parameters
After 999 games at 1+0

Mod vs Orig +239 =542 -218  +7 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 20:59:27 +01:00
Marco Costalba 889c8538a8 Remove 4*IncrementalFutilityMargin from futilityValue
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 20:59:26 +01:00
Marco Costalba c52da3b806 Logaritmic futility margins
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 20:59:24 +01:00
Marco Costalba b599da01fa Exponential futility margins
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 20:59:23 +01:00
Marco Costalba 52bca81dcb History pruning exponential limit
Use an exponenital law instead of a linear one for
history pruning.

This should prune more at low depths and a bit less
at high depths.

After 965 games

Mod vs Orig +233 =504 -228 +2 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 20:59:21 +01:00
Marco Costalba 0eedf47661 Incremental Futility Margin
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-23 20:59:20 +01:00
Marco Costalba 989833205f In razor qsearch use corrected beta
Correct beta by razor margin when callin qsearch

After 1019 games on Joona's QUAD

Mod - Orig: 524 - 495 (+10 elo)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-22 21:15:13 +01:00
Marco Costalba 87507121d5 Code style triviality
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-22 21:12:33 +01:00
Marco Costalba 89fe8bc0a6 Micro-optimize get_material_info()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-15 09:35:22 +01:00
Marco Costalba 4c58db0dab Convert pawns evaluation to Score
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-14 17:57:50 +01:00
Marco Costalba 71e852ea81 Move game phase computation to MaterialInfo
Game phase is a strictly function of the material
combination so its natural place is MaterialInfo,
not position.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-14 17:57:49 +01:00
Marco Costalba 314faa905a Null move dynamic reduction based on value
After 994 games at 1+0

Mod vs Orig +244 =521 -229 50.75%  504.5/994 +5 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-14 17:57:34 +01:00
Marco Costalba a530fc2b60 Use a more standard perft UCI interface
Call directly 'perft 6' to search up to depth 6*OnePly
instead of the old 'perft depth 6'.

It is more in line to what other engines do. Also a bit
of cleanup while there.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-13 10:35:56 +01:00
Marco Costalba 7d0e0ff95e Better document king safety evaluation
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-13 10:10:22 +01:00
Marco Costalba 764229a2e2 Rearrange table layout in evaluate.cpp
A bit more cache friendly.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-13 10:10:21 +01:00
Marco Costalba 5e340346db Remove dcCandidates data member from SplitPoint
It is no more used now that we have CheckInfo.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-13 10:10:21 +01:00
Marco Costalba bf395c6be1 Remove update_checkers()
Now that we have CheckInfo we don't need it anymore.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-11 22:26:29 +01:00
Marco Costalba ad44ff2bca Retire evaluate_mobility()
Move the code to the caller and also move mob_area
computation out of evaluate_pieces(). It is more clear
the code flow and it is also faster.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-11 22:25:08 +01:00
Marco Costalba 8e96149c8c Small sort_moves() deobfuscation
Write the for loop in a more idiomatic way, no assembly
change and of course no functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-11 22:25:07 +01:00
Marco Costalba 3c085775d7 Don't futility-prune ttMove
After 933 games
Mod vs Orig +219 =505 -208 +4 ELO

A small increase as expected.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-11 22:24:30 +01:00
Marco Costalba dd5a3ae4a6 Propagate "move is check" info to do_move()
When false (common case) we avoid to update checkers
bitboard that although not so costly slows down a bit
this very hot and critical path.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-10 18:12:19 +01:00
Marco Costalba f7f09b91ea Small update_checkers() cleanup
And is a bit faster too.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-10 17:18:00 +01:00
Marco Costalba 8a116ce691 Small update to pop_1st_bit()
Avoid a 64 bit load using a pointer. It saves a couple of push/pop
instructions so advantage is only theorical, but anyway we use
pop_1st_bit() as a reference implementation for 32 bit systems so
we keep it more for documentation purposes then for other reasons.

Idea of pointer is of Eric Mullins.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-10 17:18:00 +01:00
Marco Costalba 16626dd655 Small CheckInfo fallout
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 21:48:02 +01:00
Marco Costalba 2f01d67a92 Fully convert move_is_check() internally
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 21:38:39 +01:00
Marco Costalba 975d5e9c64 Convert move_is_check() to take a CheckInfo reference
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 21:02:07 +01:00
Marco Costalba 30075e4abc Use CheckInfo to compute dcCandidates
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 20:54:45 +01:00
Marco Costalba 37398d9456 Introduce CheckInfo struct
Keeps info used to speed-up move_is_check()

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 20:50:02 +01:00
Marco Costalba e05039156c Fix operator/(Score s, int i)
And remove some useless declarations

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 09:00:24 +01:00
Marco Costalba f35ddb04af Don't copy the key in do_move
It will be overwritten anyway.

Also other little small touches that seem to increase
speed more then the whole enum Score patch series :-(

Optimization is really a black art.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 08:45:30 +01:00
Marco Costalba ef6fca98a0 Define Score as an enum
Increases performance because now we use one integer
for both midgame and endgame scores.

Unfortunatly the latest patches seem to have reduced a bit
the speed so at the end we are more or less at the same
performance level of the beginning. But this patch series
introduced also some code cleanup so it is the main reason
we commit anyway.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-09 08:43:34 +01:00
Marco Costalba fea46a8212 Change Score definition to avoid the union
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 21:10:49 +01:00
Marco Costalba 776c7df30c Revert "Do not extend at low depths if not in PV"
On Joona's QUAD:
Orig - Mod: 414 - 373

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 19:47:48 +01:00
Marco Costalba 5e112f16da Revert "IID in pv also when TT move depth is too small"
After almost 900 games we are at -2 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 19:46:36 +01:00
Marco Costalba 15ec3e911e Last conversions to Score in evaluate.cpp
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 17:08:28 +01:00
Marco Costalba 1ecd8e13ee Convert ThreatBonus to Score
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 15:45:49 +01:00
Marco Costalba 444c7c5183 Convert RookOn7thBonus and QueenOn7thBonus to be Score
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 15:35:11 +01:00
Marco Costalba e9757f7610 Convert mobility bonus tables to Score
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 15:28:02 +01:00
Marco Costalba 1ab01f1c14 Convert apply_weight() to handle Score
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 14:33:40 +01:00
Marco Costalba 4626ec2890 Convert MaterialInfo and PawnInfo to use Score
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 14:17:10 +01:00
Marco Costalba dda7e4639a Introduce PieceSquareTable[16][64]
Instead of MgPieceSquareTable[16][64] and EgPieceSquareTable[16][64]

This allows to fetch mg and eg values from adjacent words in memory.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 13:14:24 +01:00
Marco Costalba 1ae8c59c0b Convert Position to use Score struct
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 12:40:48 +01:00
Marco Costalba 06f06a9be8 Introduce Score struct
Save mid and end game scores in an union so to
operate on both values in one instruction.

This patch just introduces the infrastructure and changes
EvalInfo to use a single Score value instead of mgValue
and egValue.

Speed is more or less the same because we still don't use
unified midgame-endgame tables where the single assignment
optimization can prove effective.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 12:35:04 +01:00
Marco Costalba 2f5ee9e4e8 Fix correct name of int64_t type
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-07 10:08:28 +01:00
Marco Costalba dd884b65b7 Do not extend at low depths if not in PV
Only check extensions are allowed.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 17:52:12 +01:00
Marco Costalba 0fdc75c0bd IID in pv also when TT move depth is too small
Try an internal iterative deepening not only when we don't
have a TT move but also if search depth is more then 4*OnePly
higher then TT move depth.

On some tests it seems that in around 20% of cases ttMove changes !

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 17:52:04 +01:00
Marco Costalba dae7cacd3b Better big-endian support wording in Makefile
Suggested by Joona.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 17:50:38 +01:00
Marco Costalba 7c0cb8e73d Enable POPCNT only through Makefile
Also remove some fallback templates that prevent a
compile error in case the user runs 'make icc-profile-popcnt'
from a non supported machine.

We want to loudly fail in that case instead of silently
fallback in a non-popcount compilation.

Updated documentation too.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 17:50:24 +01:00
Joona Kiiski 53ce6ce49c Add popcnt-support in Makefile
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 17:50:15 +01:00
Marco Costalba 7a68916ff9 Small code-style touches in movegen.cpp
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 14:42:48 +01:00
Marco Costalba 82a1e2d5fc Fix a small warning under icc
Variable 'f' in 'for' loop scope hides same named
one in outer scope.

Of curse no functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 10:39:33 +01:00
Marco Costalba dc286d2673 Big-endian compatible pop_1st_bit()
Thanks to Eric Mullins we have now endian friendly
pop_1st_bit() and also is removed the need to use
-fno-strict-aliasing compiler option with GCC.

Speed is almost as fast, very small difference if any in
perft test, so I assume almost no difference in real games.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 10:21:15 +01:00
Marco Costalba a9e536a7eb Fix a compile error in debug mode
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-06 09:23:47 +01:00
Marco Costalba e59d053984 Enable PH_TT_MOVES during evasion generation
This allow us to avoid the generation of the
evasion moves if we already have a TT move, and
in case we have a cut-off we skip evasion generation
altoghter.

Node count is changed because now we try TT move _before_
to generate evasions. The search on the TT move alters the
piece lists so that when we come back to generate evasions
we build the move list with a diferent order and this alters
the node count.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-05 19:14:17 +01:00
Marco Costalba 1c73c1c150 Extend move_is_legal() to work also when in check
This patch is a prerequisite to use TT phase
during evasions.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-05 14:58:57 +01:00
Marco Costalba 423b8b9ded Move locals definitions at the function start
It seems to me function are easier to read now.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-05 14:58:57 +01:00
Marco Costalba 94dcac1fee Retire MovePicker::discovered_check_candidates()
It is now no more needed to know dc candidates
inside MovePicker, so avoid calculating there.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-05 07:03:48 +01:00
Marco Costalba 0855d93de8 Rewrite generate_pawn_moves() and simplify evasions
Big cleanup and semplification of pawns evasions that
now are pseudo-legal as the remaining moves. This
allow us to remove a lot of tricky code.

Verified against perft: no functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-05 07:03:47 +01:00
Marco Costalba deecb3757c Generate pseudo-legal moves in generate_evasions()
This allow a big semplification in move generation
that will be committed with the next patch. And makes
handling of evasions similar to the other type of moves.

This patch plus the next seem to improve also on
the performance side because after 640 games to
verify there are no hidden regressions we are at +9 ELO

Verified with perft no functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-05 07:03:36 +01:00
Marco Costalba 53c2bf0697 Optimize generate_evasions()
Generate captures of checking piece and blocking
evasions in one go.

Also reduce of one indentation level early returning
when we have a double check.

Verified with perft no functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-04 11:18:43 +01:00
Marco Costalba 483a257618 Speed up perft
There is no need to do / undo the move at the last ply

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-04 11:18:05 +01:00
Marco Costalba 12461996a5 Remove SEE optimizations
Don't seem to help, perhaps because we
return an approximate SEE score instead of the
real negative score so that we have some bad capture
or evasion sub-optimal ordering that compensates
the speed up.

Anyhow after 999 games at 1+0
Mod vs Orig +240 =514 -245 -2 ELO

So almost no harm to remove and make the code simpler.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-04 11:17:42 +01:00
Marco Costalba 70b7404a63 Reorder evasions
Always try ttMove as first. Then try good captures ordered
by MVV/LVA, then non-captures if destination square is not
under attack, ordered by history value, and at the end
bad-captures and non-captures with a negative SEE. This
last group is ordered by the SEE score.

After 999 games at 1+0
Mod vs Orig +254 =546 -199 +19 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-01 21:08:30 +01:00
Marco Costalba dddaeff7d8 Another see() shortcut
Because we only generate legal moves we can assume
a king cannot be recaptured, so we can safely return
immediately with the captured piece score. If the move
turns out to be illegal it will be pruned anyhow,
independently from SEE value. This gives a good speed up
especially now that we SEE-test all the evasions that
are always legal and very often are king moves.

Another optimization catches almost 15% of cases, unfortunatly
we have already calculated the very expensive attacks, so
benefits are not so big anyway.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-01 21:06:26 +01:00
Marco Costalba 941d923bf8 Shortcut see_sign() when SEE is known negative
This patch cuts 30% of SEE calculations, as a drawback
a returned negative value is no more always correct if
a shortcut is found.

This could impact move order when based on negative see
score as example bad captures and evasions.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-01 21:06:14 +01:00
Marco Costalba 23de3e16f1 Remove castling moves in check generation
Check generation is used only in qsearch and
only at Depth(0), castling moves that give check
are very rare overall and even almost not exsistent
at Depth(0).

So retire this almost never used code that adds
a small but consistent slow down in the normal path.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-01 17:19:04 +01:00
Marco Costalba 8ebe5075eb Optimize check generation
Because discovery checks are very rare it is better to handle
them all in one go and strip from usual check generation
function.

Also rewrite direct checks generation to use piece lists instead
of pop_1st_bit()

On perft test we have a +6% of speed up and is verified we
generate the same moves, although in a different order.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-01 17:13:01 +01:00
Joona Kiiski fa49311b36 Implemented perft
Patch from Joona with extension to benchmark and inclusion
of Depth(0) moves generation by me.

Note that to test also qsearch and in particulary checks
generations a change in the end condition is needed.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-11-01 17:05:00 +01:00
Marco Costalba d9b920acfb Evaluation threat values after 39089 games
Verified against tuning branch.

After 100 games at 1+0 on Joona QUAD

Mod - Orig: 527.5 - 471.5 (+20 elo)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-31 09:42:59 +01:00
Marco Costalba 12b0517d1b Fix build under gcc
Also some warnings squashed.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-25 10:22:03 +01:00
Marco Costalba b38685625a Add threat evaluation
Give a bonus for each kind of attacked piece. Bonus
value is based on the type of attacked piece and the
type of attacking one.

Penalize pieces attacked by enemy pawns, also in
this case penality value depends on the type of
attacked piece.

This patch oboletes as redundant the increased mobility
count of the attcked squares that is then removed.

After 956 games at 1+0
Mod vs Orig  +262 =462 -232 51.57%  493.0/956 +11 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-25 08:21:35 +01:00
Marco Costalba 73da3a431c Micro optimize mobility calculation
Take out of mobility loop a constant expression.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-23 11:24:53 +01:00
Marco Costalba b50921fd5c Unify capture and promotion tests
Small code cleanup and a bit faster too.

The only functional change is that in extension
in pv node we extend promotions and not only captures
when condition met.

This is practically an undetectable change and has
no impact on strenght.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-22 07:22:54 +01:00
Marco Costalba a72c55283d Don't prune TT move in qsearch even if SEE < 0
Even if SEE is negative there is always a good possibility
that TT move is a cut move anyway. For instance a lot of
BXN exchanges that have negative SEE can very easily be
good exchanges.

A nice side effect is a bit reduced frequency of
see_sign() calls.

After 643 games at 1+0
Mod vs Orig +174 =327 -142 52.49%  337.5/643 +17 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-22 07:22:44 +01:00
Marco Costalba cf4df0327a Pick best moves one per cycle instead of sorting
When the move list is very small, like captures normally
are, it is faster to pick the best move with a linear
scan, one per cycle.

This has the added advantage that the picked capture move is
very possibly a cut-off move, so that other searches are
avoided. For non-captures it is still faster to sort in
advance.

Because scan-and-pick alghortim is not stable, node count
has changed.

After 885 games at 1+0
Mod vs Orig +196 =510 -179 50.96%  451.0/885

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-22 07:18:41 +01:00
Marco Costalba 51c3af9dd0 Avoid a needless locking in sp_search()
Only in less then 2% of cases we have a new sp->bestValue,
so check before to lock and save a costly locking
most of the times.

Patch suggested by Joona.

No functional search.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-19 07:47:18 +01:00
Joona Kiiski f0b0a3b135 Similarize pruning code in search() and sp_search()
Use futility pruning also in split points.
Do not use history pruning in split points when
getting mated.

After 1000 games on Joona QUAD
Orig - Mod: 496 - 504

Added an optimization to avoid a costly lock in the
very common case that sp->futilityValue <= sp->bestValue.
A test on a dual CPU shows only 114 hits on 23196 events,
so avoid a lock in all the other cases.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-18 21:31:19 +01:00
Marco Costalba 4dd7fccfd1 Use an homegrown insertion sort instead of std::sort()
It is stable and it is also a bit faster then std::sort()
on the tipical small move lists that we need to handle.

Verified to have same functionality of std::stable_sort()

After 999 games at 1+0
Mod vs Orig +240 =534 -225 50.75%  507.0/999  +5 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-17 09:24:58 +01:00
Marco Costalba c1ea5ed6f7 Do not prune the move if we are still under mate
If after the first tried 2 + int(depth) moves we still
have no any move that takes us out of a mate then do
not prune the following move, it is more important to
escape mate then speed up search.

This fixes an odd behaviour regarding mates, as example
the following diagram is a mate in 4, not in 3 as bogusly
reported before this patch.

1B2n3/8/2R5/5p2/3kp1n1/4p3/B3K3/8 w - - bm #4;

The performance impact should be minimal, the increment
in searched nodes is less then 0.1 %%

Idea and patch by Joona

After 999 games at 1+0
Mod vs Orig +193 =604 -202  -3 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-15 18:52:21 +01:00
Tord Romstad 53f882ff1a Minor improvement in eval of unstoppable pawns promoting one ply apart.
Marco's new code for evaluating two unstoppable passed pawns where
one pawn promotes a single ply before the other tried to detect
cases where the pawn that promotes first could immediately capture
the pawn that promotes a ply later, but didn't work in cases where
the two pawns are on the same file. An example of this is the
following position:

8/8/3K4/2P5/2p5/3k4/8/8 w - -

With the new code, such positions are handled correctly.
2009-10-15 12:39:55 +02:00
Marco Costalba d8e7ce1863 Fix a crash when reaching PLY_MAX in a check position
In this case we call evaluate() being in check and this
is not allowed.

Bug found testing with reduced PLY_MAX value as suggested
by Miguel A. Ballicora on talkchess.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-12 15:32:50 +01:00
Marco Costalba 181d34e5a0 Add a new rule on promoting pawns in evaluate_passed_pawns()
Add a rule about the situation when one side queens exactly
one ply before the other. To avoid difficult (but luckly rare)
cases we only handle the case of free paths to queen for
both sides.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-12 10:09:06 +02:00
Marco Costalba 2655f93c32 Fix x-ray attack from behind in evaluate_passed_pawns()
Fix a condition for x-ray attack of a queen or
a rook behind a pawn of us. Previous condition does
not check if the enemy slider behind our pawn is
really attacking the pawn.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-12 09:36:35 +02:00
Marco Costalba ab4d26f9bd Small cleanup and in evaluate_passed_pawns()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-12 09:36:30 +02:00
Marco Costalba d6b04c2798 Revert "Use std::stable_sort() instead of std::sort()"
Unfortunatly std::stable_sort() implementation in gcc is
horrendously slow. We have a big performance regression on
Linux systems (-20% !)

So revert the commit and wait to fix the issue in a different
way, perhaps with an our home grown sorting, that should be
comparable in speed with std::sort()

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-12 09:36:23 +02:00
Marco Costalba e2e249eabd Use std::stable_sort() instead of std::sort()
Standard does not mandate std::sort() to be stable, so we
can have, and actually do have different node count on
different platforms.

So use the platform independent std::stable_sort() and gain
same functionality on any platform (Windows, Unix, Mac OS)
and with any compiler (MSVC, gcc or Intel C++).

This sort is teoretically slower, but profiling shows only a very
minimal drop in performance, probably due to the fact that
the set to sort is very small, mainly only captures and with
less frequency non-captures, anyhow we are talking of 30-40 moves
in the worst average case. Sorting alghortims are build to work on
thousands or even milions of elements. With such small sets
performance difference seems not noticable.

After 999 games at 1+0

Mod vs Orig +234 =523 -242 -3 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-10 15:45:46 +01:00
Marco Costalba ed19a9f909 Unroll color loops in evaluate_passed_pawns()
Speed increase is on 1.5% on Intel pgo build.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-10 14:56:56 +01:00
Marco Costalba eddfd46a10 Use piece_list to scan the pawns in evaluate_pawns()
No functional change and small speed increase.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-10 14:56:10 +01:00
Marco Costalba a806d7c3d6 Fix pieceList initialization in Position::clear()
We want piece list to be terminated with SQ_NONE.

This happens with all the pieces but the pawns that
being 8 make the inner loop exit just before writing
the SQ_NONE value at the tail of the list.

This bug was hidden because currently we don't use
piece list to scan pawns, but this will change in the
future and in any case an initialization should be done
correctly for the whole array to avoid subtle bugs in
the future.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-10 09:49:41 +01:00
Marco Costalba ccdb634b77 Unroll color loops in get_pawn_info
This allow to resolve a lot of addresses at compile time
instead of an indirect access at runtime.

Speed up on pgo compile is of 1.3%

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-09 16:48:45 +01:00
Marco Costalba 564ed5b38c Small micro-optimization in get_pawn_info()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-09 11:29:11 +01:00
Marco Costalba f05d059b17 Restore pliesFromNull counter
It is not equivalent because the check for the
50 moves rule get badly affected.

// Draw by the 50 moves rule?
if (st->rule50 > 100 || (st->rule50 == 100 && !is_check()))
    return true;

So we _really_ need two counters.

Thanks to Joona and Tord to be patience with a silly guy ;-)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-09 10:11:43 +01:00
Marco Costalba 06a5b602dc Fix an off-by-one bug in extract_pv()
In case we reach ply == PLY_MAX we exit the function
writing

pv[PLY_MAX] = MOVE_NONE;

And because SearchStack is defined as:

struct SearchStack {
  Move pv[PLY_MAX];
  Move currentMove;
  .....

We end up with the unwanted assignment

SearchStack.currentMove = MOVE_NONE;

Fortunatly this is harmless because currentMove is not used where
extarct_pv() is called. But neverthless this is a bug that
needs to be fixed.

Thanks to Uri Blass for spotting out this.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-09 10:04:55 +01:00
Marco Costalba d892063cd3 Rewrite previous patch using only one counter
Use only rule50 and retire pliesFromNull.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-09 04:35:11 +01:00
Marco Costalba 64d6ba2e98 Do not claim repetition after null move
Null moves can artificially create a repetition
draw where instead there is no one.

So use a second counter to reset history after
a null move.

Idea from Joona.

After 999 games at 1+0

Mod vs Orig +238 =553 -208 51.50%  514.5/999  +10 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-09 03:55:10 +01:00
Marco Costalba 5a5dc6fa10 Restore development version
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-08 15:16:47 +01:00
Marco Costalba 8ee0842c81 Stockfish 1.5.1
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-08 15:16:34 +01:00
Marco Costalba e59ff49a55 Fix the polling frequency when pondering
When pondering InfiniteSearch == false but myTime == 0 so that
NodesBetweenPolls = 1000 instead of the standard.

The patch fixes the bug and is more robust because checks
directly myTime for a non-zero value, without relying on
an indirect test (InfiniteSearch in this case).

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-08 09:09:19 +01:00
Tord Romstad 225dcfeeb7 Use slightly lower polling frequency in the last few seconds.
Instead of checking the time every 100 nodes in the last second,
and every 1000 nodes in the last five seconds, Stockfish now checks
every 1000 nodes in the last second and every 5000 nodes in the last
five seconds.  This was tested in 1036 games at a time control of
40 moves/10 seconds, and no losses on time occured.

Also fixed a bug pointed out by Marco:  In infinite mode, myTime
is actually 0, but of course we still don't want to check the time
more frequently than the standard once per 30000 nodes in this
case.
2009-10-08 08:55:25 +02:00
Tord Romstad 8dd01fda12 Minor change to time management code, to make sure we don't lose on
time at the last move before the time control when there is very
little time left.
2009-10-07 18:27:00 +02:00
Tord Romstad 18cd83a380 Display fail high/fail low in search log file. 2009-10-06 12:51:15 +02:00
Marco Costalba fd2b3df770 Fix bogus comment in extract_pv()
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-06 11:15:05 +01:00
Marco Costalba da948cc94e Fix use of an initialized SearchStack
In RootMoveList c'tor we allocate a search stack and then
call directly qsearch.

There is called init_node() that clears all the fields of the search
stack array that refers to current ply but not the the killer moves.

The killer moves cleared correspond to ply+2.

In id_loop() this is not a problem because killer moves of
corresponding ply are cleared anyway few instructions later,
but in RootMoveList c'tor we leave them uninitialized.

This patch fixes this very old bug. It comes direclty
from Glaurung age.

Bug spotted by Valgrind.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-06 11:12:41 +01:00
Marco Costalba e49b21eacb Remove a redundant assignment in PawnInfo c'tor
We don't need to set key to 0 because clear() already
takes care of that.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-06 09:18:15 +01:00
Marco Costalba 60bc30275d Small code reformat in TranspositionTable::extract_pv()
In particular don't use an array of StateInfo, this
avoids a possible overflow and is in any case redundant.

Also pass as argument the pv[] array size to avoid a second
possible overflow on this one.

Fix suggested by Joona.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-06 07:14:12 +01:00
Tord Romstad 32dfaa56b0 Fixed an embarassing Chess960 bug found by Alexander Schmidt.
It turned out that we used do_move_bb to update the king and rook
bitboards when making and unmaking castling moves, which obviously
doesn't work in Chess960, where the source and destination squares
for the king or rook could be identical.

No functional change in normal chess.
2009-10-05 16:46:18 +02:00
Marco Costalba 3701a8e57d Restore development version
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-05 07:15:13 +01:00
Marco Costalba aaa07fb161 Stockfish 1.5
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-04 07:09:12 +01:00
Marco Costalba 1361ba75cb Small touches to increased mobility patch
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-04 06:59:06 +01:00
Marco Costalba da9c423989 Move a comment where it belongs in SEE
No functional change of course.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-03 10:48:20 +01:00
Marco Costalba 3713bb26ef Don't increase mobility if attacked piece is defended by a pawn
If an enemy piece is defended by a pawn don't give the
extra mobility bonus in case we attack it.

Joona says that "Paralyzing pawn" is usually worth of nothing.

On Joona QUAD after 964 games:
Orig - Patch_2: 191 - 218 - 555 (+ 10 elo)

On my PC after 999 games at 1+0:
Mod vs Orig +227 =550 -222 50.25%  502.0/999  +2 ELO

In both cases we tested against the original version (without
increased mobility), not against the previous patch that instead
seems to fail on Joona QUAD:
Orig vs. Prev.Patch: 237 - 217 - 627 (-6 elo)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-10-03 10:48:07 +01:00
Marco Costalba cff9ff2198 Count two times number of attacked pieces in mobility
Now in mobility we count enemy attacked pieces as
empty squares.

With this patch we try to give an higher score to positions
where the number of attacked pieces is higher.

After 999 games at 1+0

Mod vs Orig +262 =517 -219 52.15% 520.5/998 +15 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-30 16:11:45 +01:00
Marco Costalba a6c6037813 Optimize futilityValue calculation
Avoid calling evaluate() if we already have the score in TT

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-30 16:11:41 +01:00
Marco Costalba e677185567 Store pawn attacks in PawnInfo
They are pawn structure invariant so has a sense to
store togheter with pawn info instead of recalculating
them each time evaluate() is called.

Speed up is around 1%

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-30 16:11:37 +01:00
Tord Romstad 237dd331d5 Fixed a couple of typos in a comment.
No functional change, of course.
2009-09-30 09:53:29 +02:00
Marco Costalba 98c8a83bb8 Fix a MSVC warning in search.cpp
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-29 16:48:50 +01:00
Tord Romstad 73be819426 Temporarily removed the unfinished UCI_Elo code in preparation for
the release of the public Stockfish 1.5.
2009-09-29 13:40:00 +02:00
Marco Costalba 2940abdac8 Print RootMoveList startup scoring
This satisfies a specific user request of 28/8/2009

"The only issue I have is that during multiPV analysis, the depth 1
best move score is not reported by the engine (reporting for the best
move begins at depth 2).  I need it at depth 1 also. Would it be
possible to make this modification in future versions? This would be
of great help as otherwise I will have to use a lesser engine.

The goal of my project is to calculate the ELO performance in a game
and also the ELO rating of individual moves. For this I need depth 1
scores for lower rated performances. I intend to distribute the program
for free upon completion.

Thanks, Jack Welbourne"

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-29 10:14:43 +01:00
Marco Costalba fa0bffeafa Retire compute_weight() in evaluation.cpp
Is used only in weight_option() so inline there.
Unroll color loop also for evaluate_space() and
finally also some assorted code style fixes.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-28 17:56:04 +01:00
Marco Costalba d56345c9ae Unroll color loops in evaluate
Use templates to manually unroll the loops so that
many values could be calculated at compile time or at
runtime but with a fast direct memory access instead of
an indirect one.

This change gives a speed up of 3.5 % on pgo build !!!  :-)

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-28 11:44:12 +01:00
Marco Costalba 60e23693f0 Change back file mode of misc.cpp
It was erroneusly changed by 6bf22f35 from
mode 100644 to 100755.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-27 07:58:28 +01:00
Marco Costalba 91f0c08789 Update piece list iteration also in evaluate_pieces()
Move to what we already do in generate_piece_moves()

This simple patch gives a spped up of 1.4% !!

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-26 15:49:04 +02:00
Marco Costalba 6bf22f354f Retire faked Windows version of gettimeofday()
Use equivalent Windows function _ftime() instead.

This patch also removes two long standing warnings
under MSVC.

No functional change and no change for non-Windows systems.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-26 15:48:58 +02:00
Marco Costalba 48b74142ef Micro optimization of generate_piece_moves()
This patch make the piece list always terminated by SQ_NONE,
so that we can use a simpler and faster loop in move
generation.

Speedup is about 0.6%.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-24 07:11:39 +01:00
Marco Costalba dcb323bf0d Retire kingSquare[] array
It is redundant. Use pieceList[c][KING][0] instead.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 17:47:03 +01:00
Marco Costalba 44cb792c76 Reorder data layout and optimize access patern
With this very simple patch we get a speed boost
of 0.8% on my PC !

Sometime we find the most complex tricks to increase speed
when instead the best results come from the simplest solutions.

No functional change of course ;-)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 17:33:24 +01:00
Marco Costalba e68e135771 Fix a couple of Intel compiler warnings
And avoid calculating emptySquares for pawns captures
case.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 17:01:59 +01:00
Marco Costalba 46141b078c Fix a piece_of_color_and_type() / pieceS_of_color_and_type() typo
Bug introduced in 17c51192

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 17:01:30 +01:00
Marco Costalba 02fd34a5e8 Rename generate_piece_moves() in generate_piece_evasions()
A better and more specific name. Also a bit of code reshuffle.

Verified No functional change and No performance change
for the whole series.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 14:23:07 +01:00
Marco Costalba 20cac227bb Retire generate_pawn_captures()
And unify in generate_pawn_noncaptures() renamed
generate_pawn_moves()

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 11:18:55 +01:00
Marco Costalba 4346445be3 Retire generate_pawn_blocking_evasions()
And unify in generate_pawn_noncaptures()

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 10:47:11 +01:00
Marco Costalba 21850536d5 Standardize generate_pawn_blocking_evasions()
Rewrite in the form normally used in other similar
functions like generate_pawn_noncaptures()

This allow an easier reading of the pawn moves generators
and simplify a bit the code.

No functional change (tested on more then 100M nodes).

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-23 10:06:32 +01:00
Marco Costalba 0179a32cf5 Code style and subtle fix in move_is_legal()
A bunch of trivial code style and comment fixes.

Among them there is a real fix for a subtle case
involving promotion moves.

We currently check that a pawn push to 8/1th rank
must be a promotion, but we don't check the contary,
i.e. that a pawn push on a different rank must NOT be
a promotion. Note that, funny enough, we perform this
control for all the other pieces, but not for the pawns!

This patch fixes this really corner case.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-22 07:07:19 +01:00
Marco Costalba 8487069058 Simplify move legality check for uncommon cases
Remove a bunch of difficult and tricky code to test
legality of castle and ep moves and instead use a slower
but simpler check against the list of generated legal moves.

Because these moves are very rare the performance impact
is small but code semplification is ver big: almost 100 lines
of difficult code removed !

No functionality change. No performance change (strangely enough
there is no even minimal performance regression in pgo builds but
instead a slightly and unexpected increase).

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-22 07:07:18 +01:00
Marco Costalba 43ca5c926d Enable functionality of previous patch
Now under-promotion checks are generated.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-22 07:07:18 +01:00
Marco Costalba aed542d74c When generating checks add possibly under-promotions
In qsearch at depth 0 we generate only captures and checks.
Queen promotion moves are generated among the captures, but
under-promotion moves (both captures and non-captures) are
never generated even if they could give a discovery check.

This patch fixes this limitation extending generate_pawn_noncaptures()
to generate also check moves when required.

Apart for adding the (rare) case of an under-promotion that gives
discovery check, the patch is also a good cleanup because removes
generate_pawn_checks() altoghter.

This patch does the code clean-up but not enables the functional
change so to allow an easier debug.

No functional change and no performance change (actually a very
very small speed increase).

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-22 07:07:17 +01:00
Marco Costalba aaffcf973e Fix a bug in generate_piece_checks()
We are generating also king moves that give check !

Of course these moves are illegal so are in any case
filtered out in MovePicker. Neverthless we should avoid
to generate them.

Also simplify a bit the code.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-22 07:07:16 +01:00
Marco Costalba 746bcb348f Small micro optimization in generate_evasions()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-22 07:07:16 +01:00
Marco Costalba a7cb05b1eb Change evaluation GrainSize from 4 to 8
Idea from Joona.

After 999 games at 1+0 on my Intel Core 2 Duo
Orig - Mod: +215 =538 -226 (+11 ELO)

On Joona QUAD after 845 games at 1+0
Orig - Mod: 151 - 181 - 513 (+13 elo)

So it seems a good change !

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-22 07:06:52 +01:00
Marco Costalba 9741694fca Save static evaluation also for failed low nodes
When a node fails low and bestValue is still equal to
the original static node evaluation, then save this
in TT along with usual info.

This will allow us to avoid a future costly evaluation() call.

This patch extends to failed low nodes what we already do
for failed high ones.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 20:05:40 +01:00
Marco Costalba e145c0d3e2 Revert evaluation drift
Still not clear if it helps and, especially, how it
helps. So revert for now to avoid any influence on
future feature now under test.

With this patch we come back to be functional
equivalent to patch e33c94883 F_53.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 19:39:54 +01:00
Marco Costalba 24cc3a97a4 Evaluation drift: add always 7 instead of ply
After 828 games at 1+0

Mod vs Orig +191 =447 -190 50.06%  414.5/828

So almost no difference. Patch is committed more for
documentation purposes then for other reasons.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 19:32:53 +01:00
Marco Costalba e4277c06bf Rename piece_attacks_from() in attacks_from()
It is in line with attackers_to() and is shorter and
piece is already redundant because is passed as template
parameter anyway.

Integrate also pawn_attacks_from() in the attacks_from()
family so to have an uniform attack info API.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 14:55:28 +01:00
Marco Costalba dd80b9abaf Remove undefined pinned_pieces(Color c, Bitboard& p)
It was added in revision 5f142ec2 but never used.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 11:01:56 +01:00
Marco Costalba 84d6fe0f31 Retire attackers_to(Square s, Color c)
Use the definition in the few places where is needed.

As a nice side effect there is also an optimization in
generate_evasions() where the bitboard of enemy pieces
is computed only once and out of a tight loop.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 10:47:59 +01:00
Marco Costalba 6845397c5c Rename piece_attacks() in piece_attacks_from()
It is a bit longer but much easier to understand especially
for people new to the sources. I remember it was not trivial
for me to understand the returned attack bitboard refers to
attacks launched from the given square and not attacking the
given square.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 10:26:54 +01:00
Marco Costalba f74f42b298 Cleanup piece_attacks_square() functions
Most of them are not required to be public and are
used in one place only so remove them and use its
definitions.

Also rename piece_attacks_square() in piece_attacks()
to be aligned to the current naming policy.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 10:12:56 +01:00
Marco Costalba 0e0adfe2e1 Rename attacks_to() in attackers_to()
These functions return bitboard of attacking pieces,
not the attacks themselfs so reflect this in the name.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 09:31:48 +01:00
Marco Costalba 049139d025 Change pawn_attacks() API
Instead of pawn_attacks(Color c, Square s) define as
pawn_attacks(Square s, Color c) to be more aligned to
the others attack info functions.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 09:09:27 +01:00
Marco Costalba 62a8f393f1 Clean up API for attack information
Remove undefined functions sliding_attacks() and ray_attacks()
and retire square_is_attacked(), use the corresponding definition
instead. It is more clear that we are computing full attack
info for the given square.

Alos fix some obsolete comments in move generation functions.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 08:48:10 +01:00
Marco Costalba c5f44ef45b Move kingSquare[] array to StateInfo
This avoids to reverting back when undoing the move.

No functional change. No performance change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-20 07:32:00 +01:00
Marco Costalba 7c55b0e880 Don't compensate TT for evaluation drift
It seems that it works better without compensation
of drifted value when saving static evaluation in TT.

After 818 games at 1+0

Mod vs Orig +217 =429 -172 52.75%  431.5/818  +19 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-19 12:52:57 +01:00
Marco Costalba 77ac1e7953 Use WIN32_LEAN_AND_MEAN in lock.h
This avoids inclusion of a bunch of not very commonly
used headers from windows.h

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-17 14:18:44 +01:00
Joona Kiiski cddda7cd19 Make static value saved in TT independent from ply
After 963 games at 1+0

Mod vs Orig +246 =511 -206 52.08%  501.0/962  +14 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-17 14:16:16 +01:00
Marco Costalba c81010a878 Evaluation drift
Increase evaluation score with ply.

After 940 games at 1+0

Mod vs Orig +247 =487 -206  +15 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-15 09:04:16 +01:00
Marco Costalba 6709b01903 Fix semantic of piece_attacks<PAWN>
Return the bitboard with the pawn attacks for both colors
so to be aligned to the meaning of the others piece_attacks<Piece>
templates.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-13 16:13:49 +01:00
Marco Costalba 3863cd191c Indirectly prefetch board[from]
One of the most time critical functions is move_is_check()
and in particular the call to type_of_piece_on(from) in the
switch statement.

This call lookups in board[] array and can be slow if board[from]
is not already cached. Few instructions before in the execution stream,
we check the move for legality with pl_move_is_legal().

This patch changes pl_move_is_legal() to use type_of_piece_on(from)
for checking for a king move so that board[from] is automatically
cached in L1 and ready to be used by the near follower move_is_check()

Another advantage is that the call to king_square(us) in pl_move_is_legal()
is avoided most of the times.

Speed up of this nice and tricky patch is 0.7% !

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-13 11:35:48 +01:00
Marco Costalba f205fe1fe5 Retire piece_is_slider(PieceType pt)
Is not used in any part of the sources.

No functional change, of course ;-)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-04 08:21:07 +01:00
Marco Costalba 9f28d8a854 Second take at unifying bitboard representation access
This patch is built on Tord idea to use functions instead of
templates to access position's bitboards. This has the added advantage
that we don't need fallback functions for cases where the piece
type or the color is a variable and not a constant.

Also added Joona suggestion to workaround request for two types
of pieces like bishop_and_queens() and rook_and_queens().

No functionality or performance change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-04 08:21:06 +01:00
Marco Costalba 76bed11f7b Templetize functions to get pieces by type
Use a single template to get bitboard representation of
the position given the type of piece as a constant.

This removes almost 80 lines of code and introduces an
uniform notation to be used for querying for piece type.

No functional change and no performance change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-04 08:21:05 +01:00
Marco Costalba e33c94883f Set LMRPVMoves to 10 instead of 14
After 934 games at 1+0

Mod vs Orig +228 =493 -213 50.80%  474.5/934   +6 ELO

So it seems not negative and there is also the added
benefit to unify LMRPVMoves use in search_pv() and in
root list.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-04 08:21:04 +01:00
Marco Costalba 46ffea46ea Fix poly values mismerge
I managed to completely mismerge correct values
for QuadraticCoefficientsOppositeColor table :-(

Now it correspond to tuning branch for real.

After 999 games at 1+0

Mod vs Orig +247 =512 -240 50.35%  503.0/999  +2 ELO

So almost no change, but the new values comes from the
same tuning session of the others, so has more sense to
use these ones.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-09-04 08:21:02 +01:00
Tord Romstad 03d6a86900 Bug fix for discovered checks in connected_moves().
Because of a hard-to-spot single-character bug in connected_moves(),
the discovered check code had no effect whatsoever. The condition
in the if (...) statement at the beginning of the code would always
return false.

Thanks to Edsel Apostol for pointing out this bug!
2009-09-02 09:58:15 +02:00
Marco Costalba 17c5119222 Retire pieces_of_color_and_type()
It is used mainly in a bunch of inline oneliners
just below its definition. So substitute it with
the explicit definition and avoid information hiding.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-31 16:23:04 +02:00
Marco Costalba cf71efc34b MovePicker: rename number_of_moves() in number_of_evasions()
It is more clear that only in that case the move number is
correct, otherwise is only a partial quantity: the number of
moves of that phase.

In case of PH_EVASIONS instead we have only one phase.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-31 15:32:31 +02:00
Marco Costalba c9d364f9ca Use pointers instead of array indices also for badCaptures
To have uniformity with moves array handling.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-31 12:33:44 +02:00
Marco Costalba 97dd7568ed Document index[] and pieceList[] are not invariants
Array index[] and pieceList[] are not guaranteed to be
invariant to a do_move() + undo_move() sequence when a
capture move is involved.

The reason is that the captured piece is removed form
the list and substituted with the last one in do_move()
while in undo_move() is added again but at the end of
the list.

Because index[] and pieceList[] are used in move generation
to scan the pieces it means that moves will be generated
in a different order before and after a do_move() + undo_move()
sequence as, for instance, the one in Position::has_mate_threat()

After latest patches, move generation could now be invoked
also by MovePicker c'tor and this explains why order of
picked moves is different if MovePicker object is istantiated
before or after a Position::has_mate_threat() call.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-31 11:02:28 +02:00
Marco Costalba af220cfd52 Workaround a bug in Position::has_mate_threat()
It seems that pos.has_mate_threat() changes the position !

So that calling MovePicker c'tor before or after the
has_mate_threat() call changes the things !

Bug was unhidden by previous patch that makes MovePicker c'tor
to generate, score and sort good captures under some circumstances.

Because scoring the captures is position dependent it seems that
the moves returned by MovePicker are different when c'tor is
called before has_mate_threat()

Of course this is only a workaround because the real bug is still
hidden :-(

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-30 20:10:09 +01:00
Marco Costalba 1130c8d815 Skip TT_MOVES phase when possible
If we don't have tt moves to search skip the
useless loop associated with TT_MOVES phase.

Another 1% speed boost that brings this series
to a +6.2% against original revision 595a90df

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-30 20:10:09 +01:00
Marco Costalba 607ac0687a Movepicker: take move's loop out of switch statement
This not only cleans up the code but gives another
speed boost of 1.8%

From revision 595a90dfd0 we have increased pgo compiled binary
speed of a whopping +5.2% without any functional change !!

This is really awsome considering that we have also
cut line count by 25 lines.

Sometime we spend days for getting an extra 1% from move
generation while instead the biggest optimizations come
from anonymous and apparently dull parts of the code.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-30 20:10:08 +01:00
Marco Costalba e9de96f0e4 Revert "null move reorder" series
Does not seem to improve on the standard, latest results
from Joona after 2040 games are negative:

Orig - Mod: 454 - 424 - 1162

And is more or less the same I got few days ago.

So revert for now.

Verified same functionality of 595a90dfd

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-30 20:09:58 +01:00
Marco Costalba 9ab84a8165 Convert handling of tt moves and killers to standard form
Use the same way of loop along the move list used for
the others move kinds so to be consistent in get_next_move()

And a bit of the usual clean up too, but just a bit.

It is even a bit (+0.3%) faster now. ;-)

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-29 19:51:00 +01:00
Marco Costalba ac65b14d30 Try null move before captures
Always after TT move but before captures.

This seems a better setup against version before this
patch.

After 999 games at 1+0

Mod - Orig +252 =527 -220 +11 ELO

Unfortunatly it does not seems to improve on the standard
version, with null move outside of movepicker (595a90df) with
the latest speed-up patches added in.

After 999 games at 1+0

Mod - Standard +244 =506 -249 -2 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-29 07:17:09 +01:00
Marco Costalba 9e4befe3f1 Use pointers instead of array indices in MovePicker
This avoids calculating the array entry position
at each access and gives another boost of almost 1%.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-29 06:48:31 +01:00
Marco Costalba 6cf28d4aa7 Change the flow in wich moves are generated and picked
In MovePicker we get the next move with pick_move_from_list(),
then check if the return value is equal to MOVE_NONE and
in this case we update the state to the new phase.

This patch reorders the flow so that now from pick_move_from_list()
renamed get_next_move() we directly call go_next_phase() to
generate and sort the next bunch of moves when there are no more
move to try. This avoids to always check for pick_move_from_list()
returned value and the flow is more linear and natural.

Also use a local variable instead of a pointer dereferencing in a
time critical switch statement in get_next_move()

With this patch alone we have an incredible speed up of 3.2% !!!

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-27 19:56:26 +01:00
Marco Costalba 129cde008c Disable again null move at depth == OnePly
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-26 16:59:58 +01:00
Joona Kiiski b088f0aefd Use special null move technique in low depth.
Try good captures before null move when depth < 3 * OnePly.
Use this kind of null move also in Depth == OnePly.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-26 16:30:39 +01:00
Joona Kiiski a5d699d62f Use nullMove only through MovePicker.
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-26 16:30:35 +01:00
Joona Kiiski f6d2452916 Add Null move support to MovePicker.
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-26 16:29:18 +01:00
Joona Kiiski 268c53ac51 Create useNullMove local variable
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-26 15:42:58 +01:00
Marco Costalba 595a90dfd0 Clean killers handling in movepicker
Original patch from Joona with added optimizations
by me.

Great cleanup of MovePicker with speed improvment of 1%

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-26 15:38:47 +01:00
Marco Costalba e217407450 Micro-optimze extension()
Explicitly write the conditions for pawn to 7th
and passed pawn instead of wrapping in redundant
helpers.

Also retire the now unused move_is_pawn_push_to_7th()
and the never used move_was_passed_pawn_push() and
move_is_deep_pawn_push()

Function extension() is so time critical that this
simple patch speeds up the pgo compile of 0.5% and
it is also more clear what actually happens there.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-25 15:11:05 +01:00
Marco Costalba 2078878376 Merge branch 'master' of git-Stockfish@free2.projectlocker.com:sf 2009-08-23 18:57:11 +01:00
Marco Costalba d1d4437699 Remove a local variable from pop_1st_bit()
Remove the 'b' uint32_t local variable.
Optimized assembly is more or less the same
(one 'mov' instruction less), but now it is
written in a way more similar to the final assembly
flow so it should be easier for compiler to optimize.

Also guarantee that BitTable[] is always aligned.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-23 18:55:07 +01:00
Marco Costalba ba04eb0446 Poly ampli+bias values after 73831 games
Verified correct against tuning branch.

After 999 games at 1+0

Mod vs Orig +257 =510 -232 51.20%  +9 ELO

Very small increase but an increase anyway !

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-23 18:51:01 +01:00
Tord Romstad ed347e7cbd Added a few new targets to the Makefile for OS X with icpc.
The following new targets were added:
   * osx-icc32: 32-bit x86 compiled with icpc.
   * osx-icc64: 64-bit x86 compiled with icpc.
   * osx-icc32-profile: 32-bit x86 compiled with icpc and pgo.
   * osx-icc64-profile: 64-bit x86 compiled with icpc and pgo.
2009-08-21 10:50:34 +02:00
Marco Costalba 95af1e28be Fix some asserts raised by is_ok()
There were two asserts.

The first was raised because is_ok() was called at the
beginning of do_castle_move() and this is wrong after
the last code reformatting because at that point the state
is already modified by the caller do_move().

The second, raised by debugIncrementalEval, was due to a
rounding error in compute_value() that occurs because
TempoValueEndgame was updated in an odd number by patch

"Merge Joona Kiiski evaluation tweaks" (3ed603cd) of 13/3/2009

This line in compute_value() is the guilty one:

result += (side_to_move() == WHITE)? TempoValue / 2 : -TempoValue / 2;

The fix is to increment TempoValueEndgame so to be even.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-20 17:48:52 +01:00
Tord Romstad e9aa20ad13 Fixed incorrect material key update when making promotion moves. 2009-08-20 16:54:20 +02:00
Marco Costalba e01fefbbaf More use of memset() in Position::clear()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-18 21:21:28 +01:00
Marco Costalba e4fc957898 Little do_move() micro optimizations
Also a few remaining style touches.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-18 08:58:19 +01:00
Marco Costalba 693b38a5e7 Better clarify how pieceList[] and index[] work
Rearrange the code a bit to be more self-documenting.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-17 23:15:35 +01:00
Marco Costalba fbec55e52e Unify patch series summary
This patch seems bigger then what actually is.

It just moves some code around and adds a bit of coding style fixes
to do_move() and undo_move() so to have uniformity of naming in both
functions.

The diffstat for the whole patch series is

239 insertions(+), 426 deletions(-)

And final MSVC pgo build is even a bit faster:

Before 448.051 nodes/sec

After 453.810  nodes/sec (+1.3%)

No functional change (tested on more then 100M of nodes)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-17 15:09:20 +01:00
Marco Costalba 05e70d6740 Unify undo_ep_move(m)
Integrate undo_ep_move in undo_move() this reduces line count
and code readibility.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-17 14:48:45 +01:00
Marco Costalba b4cb1a3a9e Unify undo_promotion_move()
Integrate do_ep_move in undo_move() this reduces line count
and code readibility.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-17 14:48:33 +01:00
Marco Costalba ec14fb1b33 Unify do_promotion_move()
Integrate do_promotion_move() in do_move() this reduces line count
and code readibility.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-17 14:48:20 +01:00
Marco Costalba cb506d3b16 Unify do_ep_move()
Integrate do_ep_move in do_move() this reduces line count
and code readibility.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-17 14:47:12 +01:00
Marco Costalba e0c47a6ceb L1/L2 friendly PhaseTable[]
In Movepicker c'tor we access during initialization one of
MainSearchPhaseIndex..QsearchWithoutChecksPhaseIndex globals.

Postpone definition of PhaseTable[] just after them so that
when PhaseTable[] will be accessed later in get_next_move()
it will be already present in L1/L2.

It works like an implicit prefetching of PhaseTable[].

Also shrink PhaseTable[] to fit an L1 cache line of 16 bytes
using uint8_t instead of int.

This apparentely innocuous patch gives an astonish speed
up of 1.6% under MSVC 2010 beta, pgo optimized !

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-15 16:09:10 +01:00
Marco Costalba f3d0b76feb Use optimized pop_1st_bit() under Windows 64 with icc
Intel compiler can handle this code even under Windows.

So lift the costrain.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-14 12:47:49 +01:00
Marco Costalba bfd4421f49 Better naming and document some endgame functions
In particular the generic scaling functions.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-14 08:19:55 +01:00
Marco Costalba fd12e8cb23 Finally fix prefetch on Linux
It was due to a missing -msse compiler option !

Without this option the CPU silently discards
prefetcht2 instructions during execution.

Also added a (gcc documented) hack to prevent Intel
compiler to optimize away the prefetches.

Special thanks to Heinz for testing and suggesting
improvments. And for Jim for testing icc on Windows.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-14 08:13:42 +01:00
Marco Costalba 166c09a7a0 Reuse 5 slots instead of 4
But this time with the guarantee of an always aligned
access so that prefetching is not adversely impacted.

On Joona PC
1+0, 64Mb hash:

Orig - Mod: 174 - 237 - 359

Instead after 1000 games at 1+0 with 128MB hash size
we are at + 1 ELO (just 4 games of difference).

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-14 08:13:13 +01:00
Marco Costalba 8d369600ec Double prefetch on Windows
After fixing the cpu frequency with RightMark tool I was
able to test speed all the different prefetch combinations.

Here the results:

OS Windows Vista 32bit, MSVC compile
CPU Intecl Core 2 Duo T5220 1.55 GHz
bench on depth 12, 1 thread, 26552844 nodes searched
results in nodes/sec

no-prefetch
402486, 402005, 402767, 401439, 403060

single prefetch (aligned 64)
410145, 409159, 408078, 410443, 409652

double prefetch (aligned 64) 0+32
414739, 411238, 413937, 414641, 413834

double prefetch (aligned 64) 0+64
413537, 414337, 413537, 414842, 414240

And now also some crazy stuff:

single prefetch (aligned 128)
410145, 407395, 406230, 410050, 409949

double prefetch (aligned 64) 0+0
409753, 410044, 409456

single prefetch (aligned 64) +32
408379, 408272, 406809

single prefetch (aligned 64) +64
408279, 409059, 407395

So it seems the best is a double prefetch at the addres + 32 or +64,
I will choose the second one because it seems more natural to me.

It is still a mystery why it doesn't work under Linux :-(

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-10 22:35:08 +01:00
Marco Costalba f4140ecc0c Avoid Intel compiler optimizes away prefetching
Without this hack Intel compiler happily optimizes
away the gcc builtin call.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-10 13:49:12 +01:00
Marco Costalba 60b5da4cc8 Use aligned prefetch address
Prefetch always form a chache line boundary. It seems
that if prefetch address is not cache line aligned then
performance is adversely impacted.

Hopefully we will resuse that 32 bits of padding for something
useful in the future.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-10 13:49:00 +01:00
Marco Costalba 55c46b2399 Remove old BishopPairBonus constants
Now that we have poly imbalance these ones
are no more used.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-10 13:47:39 +01:00
Marco Costalba 76ae0e36be Enable prefetch also for gcc
This fix a compile error under Linux with gcc when
there aren't the intel dev libraries.

Also simplify the previous patch moving TT definition
from search.cpp to tt.cpp so to avoid using passing a
pointer to TT to the current position.

Finally simplify do_move(), now we miss a prefetch in the
rare case of setting an en-passant square but code is
much cleaner and performance penalty is almost zero.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-10 01:42:35 +01:00
Marco Costalba 4251eac860 Try to prefetch as soon as position key is ready
Move prefetching code inside do_move() so to allow a
very early prefetching and to put as many instructions
as possible between prefetching and following retrieve().

With this patch retrieve() times are cutted of another 25%

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-09 16:45:37 +01:00
Marco Costalba cd4604b05c Add TT prefetching support
TT.retrieve() is the most time consuming function
because almost always involves a very slow RAM access.

TT table is so big that is never cached. This patch
prefetches TT data just after a move is done, so that
subsequent TT.retrieve will be very  fast.

Profiling with VTune shows that TT:retrieve() times are
almost cutted in half !

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-09 14:18:15 +01:00
Marco Costalba e6863f46de Use 5 TTEntry slots instead of 4
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-09 04:42:26 +01:00
Marco Costalba 6f1475b6fc Use 32 bit key in TT
Shrink key to 32 bits instead of 64. To still avoid
collisions use the high 32 bits of position key as the
TT key and the low 32 bits to retrieve the correct
cluster index in the table.

With this patch size og TTentry shrinks to 96 bits instead
of 128 and the cluster of 4 TTEntry sums to 48 bytes instead
of 64.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-09 04:42:07 +01:00
Marco Costalba 4a777954e1 Makefile: added 'make strip' target
Binaries are always built with symbol table in to easy
debugging and profiling.

It is now possible to run:

make strip

To remove symbol table from the compiled binary. This
could be useful to prepare the release version.

Patch by Heinz van Saanen.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 17:37:13 +01:00
Marco Costalba 54382f8b07 Let LMR at root be independent of MultiPV value
Current formula enable LMR when

i + MultiPV >= LMRPVMoves

It means that, for instance, if MultiPV == 1 then LMR
will be started to be considered at move i = LMRPVMoves - 1,
while if MultiPV == 3 then it will start before,
at move i = LMRPVMoves - 3.

With this patch the formula becomes

i >= MultiPV + LMRPVMoves - 2

So that LMR will always start after LMRPVMoves - 1 moves
from the last PV move.

No functional change when MultiPV == 1

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 17:30:46 +01:00
Marco Costalba 339bb8a524 Speed up polynomial material imbalance loop
Access pos.piece_count() only once and avoid some
branches in the inner loop.

Profiling with VTune shows a 20% speed improvement in
get_material_info(), and it is also a bit more cleaned
up this way ;-)

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 14:12:04 +01:00
Marco Costalba aa925a0e29 There is no need to special case KNNK ending
It is always draw, so use the corresponding proper
evaluation function.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 13:10:10 +01:00
Marco Costalba 23ceb66950 Move halfOpenFiles[] calculation out of a loop
And put it in an already existing one so to
optimze a bit.

Also additional cleanups and code shuffles
all around the place.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 09:21:42 +01:00
Marco Costalba 565d12bf42 Compile without DEBUG flag by default
And build also symbol table. It can easily stripped
after .exe is done and it is necessary for profiling.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 09:21:29 +01:00
Marco Costalba 00eab73399 Revert material balance values after 100000 games
After Joona's direct testing with ~2000 games it seems
values after 100.000 games does not give any advantage,
so revert for now.

Score of Stockfish_0 vs Stockfish_15: 491 - 392 - 1102
Score of Stockfish_0 vs Stockfish_40: 461 - 439 - 1076
Score of Stockfish_0 vs Stockfish_65: 442 - 518 - 1018 (13 elo)
Score of Stockfish_0 vs Stockfish_100: 504 - 502 - 984

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 03:49:49 +01:00
Joona Kiiski 5be3d98d17 Do not adjust Minimum Split Depth automatically
Currently minimum split depth is set automatically to 6
when number of CPUs is more than 4. I believe this is a bad
idea since for example my quad (4CPU with hyperthreading) is
detected as 8CPU computer. I've manually lowered down the number
of Threads, but so far I have played all games with Minimum
Split Depth set to 6!

Since 4CPU computers with hyperthreading are quite common and
8 CPU computers extremely rear (I expect we can get a direct
jump to 16 or 32 cores), this automatic adjusting is likely
to do more harm than good. Add a note in Readme.txt, so that
those rear 8CPU owners can manually tweak the "Minimum Split
Depth" parameter

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 03:36:20 +01:00
Marco Costalba 5b3fcab1ad Polished Makefile for *nix
Greately improved Makefile from Heinz van Saanen

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-08-08 03:30:27 +01:00
Tord Romstad 977ca40d6d Supply the "upperbound" and "lowerbound" parameters in UCI search
output when the score is outside the root window.
2009-08-07 16:26:24 +02:00
Tord Romstad ae49677446 Fixed a bug in PV extraction from the transposition table: The
previous used move_is_legal to verify that the move from the TT
was legal, and the old version of move_is_legal only works when
the side to move is not in check. Fixed this by adding a separate,
slower version of move_is_legal which works even when the side to
move is in check.
2009-08-06 18:07:32 +02:00
Tord Romstad 2fff532f4e Moved the code for extracting the PV from the TT to tt.cpp, where
it belongs.
2009-08-06 14:02:53 +02:00
Tord Romstad da854fe83a Added a new function build_pv(), which extends a PV by walking
down the transposition table.

When the search was stopped before a fail high at the root was
resolved, Stockfish would often print a very short PV, sometimes
consisting of just a single move. This was not only a little
user-unfriendly, but also harmed the strength a little in
ponder-on games: Single-move PVs mean that there is no ponder
move to search.

It is perhaps worth considering to remove the pv[][] array
entirely, and always build the entire PV from the transposition
table. This would simplify the source code somewhat and probably
make the program infinitesimally faster, at the expense of
sometimes getting shorter PVs or PVs with rubbish moves near
the end.
2009-08-06 13:27:49 +02:00
Tord Romstad a1096e55cf Initial work towards adjustable playing strength.
Added the UCI_LimitStrength and the UCI_Elo options, with an Elo
range of 2100-2900. When UCI_LimitStrength is enabled, the number
of threads is set to 1, and the search speed is slowed down according
to the chosen Elo level.

Todo:

1. Implement Elo levels below 2100 by blundering on purpose and/or
   crippling the evaluation.
2. Automatically calibrate the maximum Elo by measuring the CPU speed
   during program initialization, perhaps by doing some bitboard
   computations and measuring the time taken.

No functional change when UCI_LimitStrength is false (the default).
2009-08-04 11:31:25 +02:00
Tord Romstad dad632ce5b Added LMR at the root.
After 2000 games at 1+0

Mod vs Orig +534 =1033 -433 52.525%  1050.5/2000  +18 ELO
2009-08-03 09:08:59 +02:00
Joona Kiiski 2f7723fd44 Remove useless mate value special handling in null search
After 1200 games (1CPU), time control 1+0:

Mod vs Orig: +331 =564 -277  +16 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-26 18:55:17 +01:00
Marco Costalba 152f3b13b7 Yet another small touch to endgame functions handling
It is like a never finished painting. Everyday a little touch
more.

But this time it is very little ;-)

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-26 17:42:48 +01:00
Marco Costalba bb1b049b83 Remove unused members in Application class
Also rearrange a bit the remining methods.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-26 16:11:20 +01:00
Marco Costalba 50f92bed06 Fix a spurious extra space
This morning it seems there is nothing better to do...

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-26 09:07:42 +01:00
Marco Costalba bdb586ac2b Micro optimize extension() in search.cpp
Small micro-optimization in this very
time critical function.

Use bitwise 'or' instead of logic 'or' to avoid branches
in the assembly and use the result to skip an handful of checks.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-25 16:48:28 +01:00
Marco Costalba 1b0303b6e9 Polynomial material balance after 100.000 games
Verified it is equivalent to the tuning branch results
with parameter values sampled after 100.000 games.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-24 14:26:49 +01:00
Marco Costalba 5f232e0667 Revert Makefile changes
Some unwanted changes to Makefile slept in in patch
"Introduced the UCI_AnalyseMode option".

Revert them. No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-24 14:18:03 +01:00
Marco Costalba 080a4995a3 Simplify king shelter cache handling
This is more similar to how get_material_info() and
get_pawn_info() work and also removes some clutter from
evaluate_king().

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-24 14:13:13 +01:00
Marco Costalba 20224a5bbf Delay costly SEE call during captures ordering in MovePicker
When ordering moves we push all captures with negative SEE values
to badCaptures[] array during the scoring phase.

This patch delays the costly SEE call up to when the move has been
picked up in pick_move_from_list(), this way we save some SEE calls
in case we get a cutoff.

It seems we have a speed gain of about 1-1.5 % in terms of nodes/sec
and profiling seems to confirm the small but real speed increase.

Idea from Pablo Vazquez on talkchess.com
http://www.talkchess.com/forum/viewtopic.php?t=29018&start=20

It would be a no functional change but actually it is not because
now sorting set is different and so std::sort(), that is not a
stable sort, does not guarantees the order of same scored moves to
remain the same as before.

After 952 games at 1+0 we are below error bar, almost equal just
6 games of difference (+2 ELO)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-24 14:12:33 +01:00
Marco Costalba 8654fee18c Microptimization in do_evaluate()
Do not call count_1s_max_15() if not necessary, as is
not in the common case (>95%).

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-23 22:01:42 +01:00
Marco Costalba 8b45b60327 Use do_move_bb() helpers when doing a castle
Small cleanup.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-23 10:43:58 +01:00
Marco Costalba 044ad593b3 Add Tord's polynomial material balance
Use a polynomial weighted evaluation to calculate
material value.

This is far more flexible and elegant then applying
a series of single euristic rules as before.

Also correct a design issue in which we returned two
values, one for middle game and one for endgame, while
instead, because game phase is a function of board
material itself, only one value should be calculated and
used both for mid and end game.

Verified it is equivalent to the tuning branch results with
parameter values sampled after 40.000 games.

After 999 games at 1+0

Mod vs Orig +277 =482 -240 51.85%  518.0/999  +13 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-23 00:03:30 +01:00
Marco Costalba 5600d91cff Rename int32 in int32_t
To use the same naming rule of the other types and
to be compatible with inttypes.h, used under Linux.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-20 10:53:41 +01:00
Marco Costalba 1cc44bcaae Correctly set mateThreat in search()
We do not accept null search returned mate values,
but we always do a full search in those cases.

So the variable mateThreat that is set only if null move
search returns a mate value is always false.

Restore the functionality of mateThreat moving the
assignement where it can be triggered.

After 999 games at 1+0

Mod vs Orig +253 =517 -229 51.20%  +8 ELO

Bug reported by xiaozhi

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-20 08:05:48 +01:00
Marco Costalba 15eb59683e Use increased LMR horizont also in PV search
Tord says that using a lower horizon at PV nodes
looks strange and inconsistent with the general
philosophy of our search (i.e. always being more
conservative at PV nodes). So set LMR at 3 also
on search_pv().

Test result after 601 games seems to confirm this.

Mod vs Orig +156 =318 -127 52.41%  315.0/601  +17 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-18 12:47:37 +02:00
Marco Costalba 620cfbb676 Reintroduce null move dynamic reduction
Test extension of LMR horizon to 3 plies alone, without
touching null move search. To keep the patch minimal we still
don't change LMR horizon in PV search. This will be the object
of the next patch.

Result seems good after 998 games:

Mod vs Orig  +252/=518/-228 51.20%  511.0/998 +8 ELO

So dynamic null move reduction seems a bit stronger then
fixed reduction even with LMR horizon set to 3.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-18 06:08:06 +01:00
Marco Costalba fe523b2d18 Use increased LMR horizont only after a null move
Revert to LMR horizont of 2 plies. Only if parent move
is a null move increase to 3 so to avoid the bad combination
of null move reduction + LMR reduction. This is a more
aggressive patch then previous one, but it seems we are
going in the wromg direction.

After 531 games result is not good:

Mod vs Orig  +123/=265/-143 48.12%  255.5/531  -13 ELO

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-18 06:08:02 +01:00
Marco Costalba 2a203d8d6f Combine increased LMR horizont and fixed null move reduction
Set null move reduction to R=4, but increase the LMR horizon
to 3 plies. The two tweaks are related and should compensate
the combined effect of null move + LMR reduction at shallow
depths.

Idea from Tord.

After 999 games at 1+0

Mod vs Orig  +251 =522 -225 51.30% + 9 ELO

On Tord iMac Core 2 Duo 2.8 GHz, one thread,
Mac OS X 10.6, at 1+0 time control we have:

Mod vs Orig 994-1006  -1.4 ELO

But Orig version is pgo compiled and Mod is not.
The PGO compiled version is about 8% faster, which
corresponds to about 7 Elo points. This means that
results are reasonably consistent.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-18 06:07:58 +01:00
Tord Romstad b8326edea3 Introduced the UCI_AnalyseMode option, and made the evaluation function
symmetrical in analyse mode.

No functional change when playing games.
2009-07-17 22:26:01 +02:00
Marco Costalba 20e8738901 Fix two compile errors in new endgame code
Code that compiles cleanly under MSVC triggers one
compile error (correct) under Intel C++ and two(!)
under gcc.

The first is the same complained by Intel, but the second
is an interesting corner case of C++ standard (there are many)
that is correctly spotted only by gcc.

Both MSVC and Intel pass this silently, probably to avoid
breaking people code.

Now we are fully C++ compliant ;-)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-17 19:29:25 +01:00
Marco Costalba b3b1d3aaa7 Move constant bitboard arrays from header to cpp file
This avoid to duplicate storage allocation for every file
where they are used.

Note that simple numeric constant can remain in header because
are automatically folded by the compiler.

Patch suggested by Tord.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-17 16:25:53 +01:00
Marco Costalba 0d69ac33ff Remove even more redundancy in endgame functions handling
Push on the templatization even more to chip out some code
and take the opportunity to show some neat template trick ;-)

Ok. I would say we can stop here now....it is quickly becoming
a style exercise but we are not boost developers so give it a stop.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-17 16:05:19 +01:00
Tord Romstad 342c8c883c Removed an incorrect assert() statement in search.cpp, which asserted that
a static eval cached in the transposition table would always equal the static
eval of the current position. This is in general not true, because the cached
value could be from a previous search with different evaluation parameter
settings, or from a search from the opposite side (Stockfish's evaluation
function is assymmetric by default).
2009-07-17 09:12:59 +02:00
Marco Costalba 297c12e595 Simplify endgame functions handling
We really don't need to have global endgame functions. We can
allocate them on the heap at initialization time and store the
corresponding pointer directly in the functions maps. To avoid
leaks we just need to remember to deallocate them in map d'tor.

These functions are always created in couple, one for each color,
so remove a lot of redundant hard coded info and just use the minimum
required: the type and the corresponding named string.

This greatly simplifies the code and also it is less error prone,
now is much simpler to add a new endgame specialized function: just
add the corresponding enum in endgame.h and the obvious add_xx()
call in EndgameFunctions c'tor, and of course, the most important part,
the EvaluationFunction<xxx>::apply() specialization in endgame.cpp

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-17 07:55:51 +01:00
Tord Romstad 5c20f59788 Renamed the variable 'looseOnTime' to 'loseOnTime', because I'm a pedant.
No functional change.
2009-07-15 11:01:49 +02:00
Marco Costalba ea06200423 Remove "Last seconds noise" filtering UCI option
This feature makes sense during development, but
It doesn't seem to make sense for normal users.

Also fix a possible race where the GUI adjudicates
the game a fraction of second before the engine sets
looseOnTime flag so that it will bogusly waits until
it ran out of time at the beginning of the next new game.

The fix is to always reset looseOnTime at the beginning
of a new game.

Race condition spotted by Tord.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-15 08:35:00 +01:00
Marco Costalba 3849beb979 Introduce SERIALIZE_MOVES_D() macro and use it for pawn moves
This is another moves serialization macro but this time
focused on pawn moves where the 'from' square is given as
a delta from the 'to' square.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-14 10:28:41 +01:00
Marco Costalba 20ed03fc0b Micro optimize pawn moves generation
It is very rare we have pawns on 7(2) rank, so we
can skip the promotion handling stuff in most cases.

With this patch pawn moves generation is almost 20% faster.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-14 10:28:29 +01:00
Marco Costalba 2a461b4b74 Introduce see_sign() and use it to shortcut full see()
Mostly of times we are interested only in the sign of SEE,
namely if a capture is negative or not.

If the capturing piece is smaller then the captured one we
already know SEE cannot be negative and this information
is enough most of the times. And of course it is much
faster to detect then a full SEE.

Note that in case see_sign() is negative then the returned
value is exactly the see() value, this is very important,
especially for ordering capturing moves.

With this patch the calls to the costly see() are reduced
of almost 30%.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-12 08:37:43 +01:00
Marco Costalba 6f39a3fc80 Move some global variables to local scope in search.cpp
Some variables were global due to some old and now removed code,
but now can be moved in local scope.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-12 08:37:43 +01:00
Marco Costalba 7eefc1f6cc Joona tweaks of Weights and limits
Verification test give unusless result

After 999 games at 1+0
Mod vs Orig +250 =503 -246 50.20% +1 ELO

So we are well below our radar level. Neverthless
there are 100.000 games on Joona QUAD that we could
take in account and that shows that this tweak perhaps
has something good in it, altough very little.

Verification tests shows should not be a regression, at
least not a big one even in the worst case, so apply the
change anyway and keep the finger crossed ;-)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-12 08:37:28 +01:00
Marco Costalba 7622793080 Small tidy up of previous patch
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-10 18:50:43 +01:00
Tord Romstad 174b40c28d Strip whitespace from beginning of string sent to set_option_value().
It turned out that the input sent to set_option_value() when it is called by
set_option() in uci.cpp always started with at least one whitespace. In most
cases, this is not a problem, because the majority of UCI options have numeric
values. It did, however, cause a problem for UCI options with non-numerical
values, like options of type CHECK and COMBO. In particular, changing the
value of an option of type CHECK didn't work, because the comparisons with
"true" and "false" would always return false. This means that the "Ponder"
and "UCI_Chess960" options haven't been working for a while.
2009-07-10 18:34:56 +02:00
Marco Costalba 03f524c591 Revert last tweaks
Tests show no improvment, so revert for now.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-09 16:45:39 +01:00
Marco Costalba 3444b94735 Joona tweaks of tempos and misc parameters
Unfortunatly this tweak does not give good results.

After 894 games at 1+0 we have:

Mod vs Orig  +205/-236/=453 48.27%  -12 ELO !!

Perhaps we should test again, but in the mean time
we are going to revert this.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-09 16:45:17 +01:00
Marco Costalba 2693db616d Restore development versioning and LSN filtering
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-06 11:20:05 +01:00
Marco Costalba a5fce1958b Fix generation of check blocking promotion
A promotion move is not considered a possible evasion as it could be.

Bug introduced by patch

Convert also generate_pawn_blocking_evasions() to new API (7/5/2009)

Bug spotted by Kenny Dail.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2009-07-06 09:41:22 +01:00
59 changed files with 5350 additions and 5277 deletions
+90 -89
View File
@@ -1,89 +1,90 @@
1. Introduction
---------------
Stockfish is a free UCI chess engine derived from Glaurung 2.1. It is
not a complete chess program, but requires some UCI compatible GUI
(like XBoard with PolyGlot, eboard, Jos, Arena, Sigma Chess, Shredder,
Chess Partner, or Fritz) in order to be used comfortably. Read the
documentation for your GUI of choice for information about how to use
Stockfish with your GUI.
This version of Stockfish supports up to 8 CPUs, but has not been
tested thoroughly with more than 2. The program tries to detect the
number of CPUs on your computer and set the number of search threads
accordingly, but please be aware that the detection is not always
correct. It is therefore recommended to inspect the value of the
"Threads" UCI parameter, and to make sure it equals the number of CPU
cores on your computer.
2. Files
--------
This distribution of Stockfish consists of the following files:
* Readme.txt, the file you are currently reading.
* Copying.txt, a text file containing the GNU General Public
License.
* src/, a subdirectory containing the full source code, including a
Makefile that can be used to compile Stockfish on Unix-like
systems. For further information about how to compile Stockfish
yourself, read section 4 below.
* polyglot.ini, for using Stockfish with Fabien Letouzey's PolyGlot
adapter.
3. Opening books
----------------
This version of Stockfish has experimental support for PolyGlot opening
books. For information about how to create such books, consult the
PolyGlot documentation. The book file can be selected by setting the
UCI parameter "Book File".
4. Compiling it yourself
------------------------
On Unix-like systems, it should usually be possible to compile
Stockfish directly from the source code with the included Makefile.
The exception is computer with big-endian CPUs, like PowerPC
Macintoshes. Some of the bitboard routines in the current version of
Stockfish are endianness-sensitive, and won't work on a big-endian CPU.
Stockfish has POPCNT instruction runtime detection and support. This can
give an extra speed on Core i7 or similar systems. To enable this feature
(disabled by default) simply uncomment #define USE_POPCNT in bitcount.h
before to compile.
On 64 bit Unix-like systems the 'bsfq' assembly instruction will be used
for bit counting. Detection is automatic at compile time, but in case you
experience compile problems you can comment out #define USE_BSFQ line in types.h
5. Terms of use
---------------
Stockfish is free, and distributed under the GNU General Public License
(GPL). Essentially, this means that you are free to do almost exactly
what you want with the program, including distributing it among your
friends, making it available for download from your web site, selling
it (either by itself or as part of some bigger software package), or
using it as the starting point for a software project of your own.
The only real limitation is that whenever you distribute Stockfish in
some way, you must always include the full source code, or a pointer
to where the source code can be found. If you make any changes to the
source code, these changes must also be made available under the GPL.
For full details, read the copy of the GPL found in the file named
Copying.txt.
6. Feedback
-----------
The author's e-mail address is mcostalba@gmail.com
1. Introduction
---------------
Stockfish is a free UCI chess engine derived from Glaurung 2.1. It is
not a complete chess program, but requires some UCI compatible GUI
(like XBoard with PolyGlot, eboard, Jos, Arena, Sigma Chess, Shredder,
Chess Partner, or Fritz) in order to be used comfortably. Read the
documentation for your GUI of choice for information about how to use
Stockfish with your GUI.
This version of Stockfish supports up to 8 CPUs, but has not been
tested thoroughly with more than 2. The program tries to detect the
number of CPUs on your computer and set the number of search threads
accordingly, but please be aware that the detection is not always
correct. It is therefore recommended to inspect the value of the
"Threads" UCI parameter, and to make sure it equals the number of CPU
cores on your computer. If you are using more than four threads, it
is recommended to raise the value of "Minimum Split Depth" UCI parameter
to 6.
2. Files
--------
This distribution of Stockfish consists of the following files:
* Readme.txt, the file you are currently reading.
* Copying.txt, a text file containing the GNU General Public
License.
* src/, a subdirectory containing the full source code, including a
Makefile that can be used to compile Stockfish on Unix-like
systems. For further information about how to compile Stockfish
yourself, read section 4 below.
* polyglot.ini, for using Stockfish with Fabien Letouzey's PolyGlot
adapter.
3. Opening books
----------------
This version of Stockfish has experimental support for PolyGlot opening
books. For information about how to create such books, consult the
PolyGlot documentation. The book file can be selected by setting the
UCI parameter "Book File".
4. Compiling it yourself
------------------------
On Unix-like systems, it should usually be possible to compile
Stockfish directly from the source code with the included Makefile.
For big-endian machines like Power PC you need to enable the proper
flag changing from -DNBIGENDIAN to -DBIGENDIAN in the Makefile.
Stockfish has POPCNT instruction runtime detection and support. This can
give an extra speed on Core i7 or similar systems. To enable this feature
compile with 'make icc-profile-popcnt'
On 64 bit Unix-like systems the 'bsfq' assembly instruction will be used
for bit counting. Detection is automatic at compile time, but in case you
experience compile problems you can comment out #define USE_BSFQ line in types.h
5. Terms of use
---------------
Stockfish is free, and distributed under the GNU General Public License
(GPL). Essentially, this means that you are free to do almost exactly
what you want with the program, including distributing it among your
friends, making it available for download from your web site, selling
it (either by itself or as part of some bigger software package), or
using it as the starting point for a software project of your own.
The only real limitation is that whenever you distribute Stockfish in
some way, you must always include the full source code, or a pointer
to where the source code can be found. If you make any changes to the
source code, these changes must also be made available under the GPL.
For full details, read the copy of the GPL found in the file named
Copying.txt.
6. Feedback
-----------
The author's e-mail address is mcostalba@gmail.com
+292 -107
View File
@@ -18,133 +18,318 @@
# along with this program. If not, see <http://www.gnu.org/licenses/>.
###
### Files
###
### Executable name. Do not change
EXE = stockfish
### Installation dir definitions
PREFIX = /usr/local
BINDIR = $(PREFIX)/bin
### ==========================================================================
### Compiler speed switches for both GCC and ICC. These settings are generally
### fast on a broad range of systems, but may be changed experimentally
### ==========================================================================
GCCFLAGS = -O3 -msse
ICCFLAGS = -fast -msse
ICCFLAGS-OSX = -fast -mdynamic-no-pic
### ==========================================================================
### Enable/disable debugging, disabled by default
### ==========================================================================
GCCFLAGS += -DNDEBUG
ICCFLAGS += -DNDEBUG
ICCFLAGS-OSX += -DNDEBUG
### ==========================================================================
### Remove below comments to compile for a big-endian machine
### ==========================================================================
#GCCFLAGS += -DBIGENDIAN
#ICCFLAGS += -DBIGENDIAN
#ICCFLAGS-OSX += -DBIGENDIAN
### ==========================================================================
### Run built-in benchmark for pgo-builds with: 32MB hash 1 thread 10 depth
### These settings are generally fast, but may be changed experimentally
### ==========================================================================
PGOBENCH = ./$(EXE) bench 32 1 10 default depth
### General compiler settings. Do not change
GCCFLAGS += -g -Wall -fno-exceptions -fno-rtti
ICCFLAGS += -g -Wall -fno-exceptions -fno-rtti -wd383,869,981,10187,10188,11505,11503
ICCFLAGS-OSX += -g -Wall -fno-exceptions -fno-rtti -wd383,869,981,10187,10188,11505,11503
### General linker settings. Do not change
LDFLAGS = -lpthread
### Object files. Do not change
OBJS = application.o bitboard.o pawns.o material.o endgame.o evaluate.o main.o \
misc.o move.o movegen.o history.o movepick.o search.o piece.o \
position.o direction.o tt.o value.o uci.o ucioption.o \
mersenne.o book.o bitbase.o san.o benchmark.o
###
### Rules
###
### General rules. Do not change
default:
$(MAKE) gcc
help:
@echo ""
@echo "Makefile options:"
@echo ""
@echo "make > Default: Compiler = g++"
@echo "make gcc-popcnt > Compiler = g++ + popcnt-support"
@echo "make icc > Compiler = icpc"
@echo "make icc-profile > Compiler = icpc + automatic pgo-build"
@echo "make icc-profile-popcnt > Compiler = icpc + automatic pgo-build + popcnt-support"
@echo "make osx-ppc32 > PPC-Mac OS X 32 bit. Compiler = g++"
@echo "make osx-ppc64 > PPC-Mac OS X 64 bit. Compiler = g++"
@echo "make osx-x86 > x86-Mac OS X 32 bit. Compiler = g++"
@echo "make osx-x86_64 > x86-Mac OS X 64 bit. Compiler = g++"
@echo "make osx-icc32 > x86-Mac OS X 32 bit. Compiler = icpc"
@echo "make osx-icc64 > x86-Mac OS X 64 bit. Compiler = icpc"
@echo "make osx-icc32-profile > OSX 32 bit. Compiler = icpc + automatic pgo-build"
@echo "make osx-icc64-profile > OSX 64 bit. Compiler = icpc + automatic pgo-build"
@echo "make hpux > HP-UX. Compiler = aCC"
@echo "make strip > Strip executable"
@echo "make clean > Clean up"
@echo ""
all: $(EXE) .depend
test check: default
@$(PGOBENCH)
clean:
$(RM) *.o .depend stockfish
$(RM) *.o .depend *~ $(EXE) core bench.txt
###
### Compiler:
###
### Possible targets. You may add your own ones here
gcc:
$(MAKE) \
CXX='g++' \
CXXFLAGS="$(GCCFLAGS)" \
all
CXX = g++
# CXX = icpc
gcc-popcnt:
$(MAKE) \
CXX='g++' \
CXXFLAGS="$(GCCFLAGS) -DUSE_POPCNT" \
all
###
### Dependencies
###
icc:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS)" \
all
icc-profile-make:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS)" \
CXXFLAGS+='-prof-gen=srcpos -prof_dir ./profdir' \
all
icc-profile-use:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS)" \
CXXFLAGS+='-prof_use -prof_dir ./profdir' \
all
icc-profile:
@rm -rf profdir
@mkdir profdir
@touch *.cpp *.h
$(MAKE) icc-profile-make
@echo ""
@echo "Running benchmark for pgo-build ..."
@$(PGOBENCH) > /dev/null
@echo "Benchmark finished. Build final executable now ..."
@echo ""
@touch *.cpp *.h
$(MAKE) icc-profile-use
@rm -rf profdir bench.txt
icc-profile-make-with-popcnt:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS) -DUSE_POPCNT" \
CXXFLAGS+='-prof-gen=srcpos -prof_dir ./profdir' \
all
icc-profile-use-with-popcnt:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS) -DUSE_POPCNT" \
CXXFLAGS+='-prof_use -prof_dir ./profdir' \
all
icc-profile-popcnt:
@rm -rf profdir
@mkdir profdir
@touch *.cpp *.h
$(MAKE) icc-profile-make
@echo ""
@echo "Running benchmark for pgo-build (popcnt disabled)..."
@$(PGOBENCH) > /dev/null
@touch *.cpp *.h
$(MAKE) icc-profile-make-with-popcnt
@echo ""
@echo "Running benchmark for pgo-build (popcnt enabled)..."
@$(PGOBENCH) > /dev/null
@echo "Benchmarks finished. Build final executable now ..."
@echo ""
@touch *.cpp *.h
$(MAKE) icc-profile-use-with-popcnt
@rm -rf profdir bench.txt
osx-ppc32:
$(MAKE) \
CXX='g++' \
CXXFLAGS="$(GCCFLAGS)" \
CXXFLAGS+='-arch ppc' \
CXXFLAGS+='-DBIGENDIAN' \
LDFLAGS+='-arch ppc' \
all
osx-ppc64:
$(MAKE) \
CXX='g++' \
CXXFLAGS="$(GCCFLAGS)" \
CXXFLAGS+='-arch ppc64' \
CXXFLAGS+='-DBIGENDIAN' \
LDFLAGS+='-arch ppc64' \
all
osx-x86:
$(MAKE) \
CXX='g++' \
CXXFLAGS="$(GCCFLAGS)" \
CXXFLAGS+='-arch i386 -mdynamic-no-pic' \
LDFLAGS+='-arch i386 -mdynamic-no-pic' \
all
osx-x86_64:
$(MAKE) \
CXX='g++' \
CXXFLAGS="$(GCCFLAGS)" \
CXXFLAGS+='-arch x86_64 -mdynamic-no-pic' \
LDFLAGS+='-arch x86_64 -mdynamic-no-pic' \
all
osx-icc32:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS-OSX)" \
CXXFLAGS+='-arch i386' \
LDFLAGS+='-arch i386' \
all
osx-icc64:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS-OSX)" \
CXXFLAGS+='-arch x86_64' \
LDFLAGS+='-arch x86_64' \
all
osx-icc32-profile-make:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS-OSX)" \
CXXFLAGS+='-arch i386' \
CXXFLAGS+='-prof_gen -prof_dir ./profdir' \
LDFLAGS+='-arch i386' \
all
osx-icc32-profile-use:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS-OSX)" \
CXXFLAGS+='-arch i386' \
CXXFLAGS+='-prof_use -prof_dir ./profdir' \
LDFLAGS+='-arch i386' \
all
osx-icc32-profile:
@rm -rf profdir
@mkdir profdir
@touch *.cpp *.h
$(MAKE) osx-icc32-profile-make
@echo ""
@echo "Running benchmark for pgo-build ..."
@$(PGOBENCH) > /dev/null
@echo "Benchmark finished. Build final executable now ..."
@echo ""
@touch *.cpp *.h
$(MAKE) osx-icc32-profile-use
@rm -rf profdir bench.txt
osx-icc64-profile-make:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS-OSX)" \
CXXFLAGS+='-arch x86_64' \
CXXFLAGS+='-prof_gen -prof_dir ./profdir' \
LDFLAGS+='-arch x86_64' \
all
osx-icc64-profile-use:
$(MAKE) \
CXX='icpc' \
CXXFLAGS="$(ICCFLAGS-OSX)" \
CXXFLAGS+='-arch x86_64' \
CXXFLAGS+='-prof_use -prof_dir ./profdir' \
LDFLAGS+='-arch x86_64' \
all
osx-icc64-profile:
@rm -rf profdir
@mkdir profdir
@touch *.cpp *.h
$(MAKE) osx-icc64-profile-make
@echo ""
@echo "Running benchmark for pgo-build ..."
@$(PGOBENCH) > /dev/null
@echo "Benchmark finished. Build final executable now ..."
@echo ""
@touch *.cpp *.h
$(MAKE) osx-icc64-profile-use
@rm -rf profdir bench.txt
hpux:
$(MAKE) \
CXX='/opt/aCC/bin/aCC -AA +hpxstd98 -DBIGENDIAN -mt +O3 -DNDEBUG' \
CXXFLAGS="" \
LDFLAGS="" \
all
strip:
strip $(EXE)
### Compilation. Do not change
$(EXE): $(OBJS)
$(CXX) $(LDFLAGS) -o $@ $(OBJS)
### Installation
install: default
-mkdir -p -m 755 $(BINDIR)
-cp $(EXE) $(BINDIR)
-strip $(BINDIR)/$(EXE)
### Dependencies. Do not change
.depend:
$(CXX) -MM $(OBJS:.o=.cpp) > $@
-@$(CXX) -msse -MM $(OBJS:.o=.cpp) > $@ 2> /dev/null
include .depend
###
### Compiler and linker switches
###
# Enable/disable debugging, disabled by default
CXXFLAGS += -DNDEBUG
# Compile with full warnings, and symbol names stripped, you can use
# -g instead of -s to compile symbol's table in, useful for debugging.
CXXFLAGS += -Wall -s
# General optimization flags. Note that -O2 might be faster than -O3 on some
# systems; this requires testing.
CXXFLAGS += -O3 -fno-exceptions -fno-rtti -fno-strict-aliasing
# Disable most annoying warnings for the Intel C++ compiler
# CXXFLAGS += -wd383,869,981
# Compiler optimization flags for the Intel C++ compiler in Mac OS X:
# CXXFLAGS += -mdynamic-no-pic -no-prec-div -ipo -static -xP
# Profiler guided optimization with the Intel C++ compiler v11. To use it, first
# create the directory ./profdata if it does not already exist, and delete its
# contents if it does exist. Then compile with -prof_gen, and run the
# resulting binary for a while (for instance, do ./stockfish bench 128 1, and
# wait 15 minutes for the benchmark to complete). Then do a 'make clean', and
# recompile with -prof_use.
# CXXFLAGS += -prof-gen -prof-dir./profdata
# CXXFLAGS += -prof-use -ipo -prof_dir./profdata
# Profiler guided optimization with GCC. I've never been able to make this
# work.
# CXXFLAGS += -fprofile-generate
# LDFLAGS += -fprofile-generate
# CXXFLAGS += -fprofile-use
# CXXFLAGS += -fprofile-use
# General linker flags
LDFLAGS += -lm -lpthread
# Compiler switches for generating binaries for various CPUs in Mac OS X.
# Note that 'arch ppc' and 'arch ppc64' only works with g++, and not with
# the intel compiler.
# CXXFLAGS += -arch ppc
# CXXFLAGS += -arch ppc64
# CXXFLAGS += -arch i386
# CXXFLAGS += -arch x86_64
# LDFLAGS += -arch ppc
# LDFLAGS += -arch ppc64
# LDFLAGS += -arch i386
# LDFLAGS += -arch x86_64
# Backwards compatibility with Mac OS X 10.4 when compiling under 10.5 with
# GCC 4.0. I haven't found a way to make it work with GCC 4.2.
# CXXFLAGS += -isysroot /Developer/SDKs/MacOSX10.4u.sdk
# CXXFLAGS += -mmacosx-version-min=10.4
# LDFLAGS += -isysroot /Developer/SDKs/MacOSX10.4u.sdk
# LDFLAGS += -Wl,-syslibroot /Developer/SDKs/MacOSX10.4u.sdk
# LDFLAGS += -mmacosx-version-min=10.4
# Backwards compatibility with Mac OS X 10.4 when compiling with ICC. Doesn't
# work yet. :-(
# CXXFLAGS += -DMAC_OS_X_VERSION_MIN_REQUIRED=1040
# CXXFLAGS += -DMAC_OS_X_VERSION_MAX_ALLOWED=1040
# CXXFLAGS += -D__ENVIRONMENT_MAC_OS_X_VERSION_MIN_REQUIRED__=1040
# CXXFLAGS += -F/Developer/SDKs/MacOSX10.4u.sdk/
# LDFLAGS += -Wl,-syslibroot -Wl,/Developer/SDKs/MacOSX10.4u.sdk
-include .depend
+5 -9
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -47,9 +47,9 @@ Application::Application() {
init_uci_options();
Position::init_zobrist();
Position::init_piece_square_tables();
MovePicker::init_phase_table();
init_eval(1);
init_bitbases();
init_search();
init_threads();
// Make random number generation less deterministic, for book moves
@@ -59,19 +59,15 @@ Application::Application() {
Application::~Application() {
stop_threads();
exit_threads();
quit_eval();
}
void Application::initialize() {
instance();
}
Application& Application::instance() {
// A static Application object is allocated
// once only when this function is called.
static Application singleton;
return singleton;
}
void Application::exit_with_failure() {
+2 -9
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -29,18 +29,11 @@ class Application {
Application();
Application(const Application&);
~Application();
public:
static void initialize();
static void exit_with_failure();
~Application();
private:
static Application& instance();
void init();
void deallocateAll();
};
#endif // !defined(APPLICATION_H_INCLUDED)
+7 -5
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -84,9 +84,9 @@ void benchmark(const string& commandLine) {
}
csStr >> threads;
csVal >> val;
if (val < 1 || val > THREAD_MAX)
if (val < 1 || val > MAX_THREADS)
{
cerr << "The number of threads must be between 1 and " << THREAD_MAX << endl;
cerr << "The number of threads must be between 1 and " << MAX_THREADS << endl;
Application::exit_with_failure();
}
set_option_value("Hash", ttSize);
@@ -104,7 +104,7 @@ void benchmark(const string& commandLine) {
if (limitType == "time")
secsPerPos = val * 1000;
else if (limitType == "depth")
else if (limitType == "depth" || limitType == "perft")
maxDepth = val;
else
maxNodes = val;
@@ -153,7 +153,9 @@ void benchmark(const string& commandLine) {
int dummy[2] = {0, 0};
Position pos(*it);
cerr << "\nBench position: " << cnt << '/' << positions.size() << endl << endl;
if (!think(pos, true, false, 0, dummy, dummy, 0, maxDepth, maxNodes, secsPerPos, moves))
if (limitType == "perft")
totalNodes += perft(pos, maxDepth * OnePly);
else if (!think(pos, false, false, 0, dummy, dummy, 0, maxDepth, maxNodes, secsPerPos, moves))
break;
totalNodes += nodes_searched();
}
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+110 -20
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -163,6 +163,53 @@ const int RShift[64] = {
#endif // defined(IS_64BIT)
const Bitboard SquaresByColorBB[2] = { BlackSquaresBB, WhiteSquaresBB };
const Bitboard FileBB[8] = {
FileABB, FileBBB, FileCBB, FileDBB, FileEBB, FileFBB, FileGBB, FileHBB
};
const Bitboard NeighboringFilesBB[8] = {
FileBBB, FileABB|FileCBB, FileBBB|FileDBB, FileCBB|FileEBB,
FileDBB|FileFBB, FileEBB|FileGBB, FileFBB|FileHBB, FileGBB
};
const Bitboard ThisAndNeighboringFilesBB[8] = {
FileABB|FileBBB, FileABB|FileBBB|FileCBB,
FileBBB|FileCBB|FileDBB, FileCBB|FileDBB|FileEBB,
FileDBB|FileEBB|FileFBB, FileEBB|FileFBB|FileGBB,
FileFBB|FileGBB|FileHBB, FileGBB|FileHBB
};
const Bitboard RankBB[8] = {
Rank1BB, Rank2BB, Rank3BB, Rank4BB, Rank5BB, Rank6BB, Rank7BB, Rank8BB
};
const Bitboard RelativeRankBB[2][8] = {
{ Rank1BB, Rank2BB, Rank3BB, Rank4BB, Rank5BB, Rank6BB, Rank7BB, Rank8BB },
{ Rank8BB, Rank7BB, Rank6BB, Rank5BB, Rank4BB, Rank3BB, Rank2BB, Rank1BB }
};
const Bitboard InFrontBB[2][8] = {
{ Rank2BB | Rank3BB | Rank4BB | Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank3BB | Rank4BB | Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank4BB | Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank6BB | Rank7BB | Rank8BB,
Rank7BB | Rank8BB,
Rank8BB,
EmptyBoardBB
},
{ EmptyBoardBB,
Rank1BB,
Rank2BB | Rank1BB,
Rank3BB | Rank2BB | Rank1BB,
Rank4BB | Rank3BB | Rank2BB | Rank1BB,
Rank5BB | Rank4BB | Rank3BB | Rank2BB | Rank1BB,
Rank6BB | Rank5BB | Rank4BB | Rank3BB | Rank2BB | Rank1BB,
Rank7BB | Rank6BB | Rank5BB | Rank4BB | Rank3BB | Rank2BB | Rank1BB
}
};
Bitboard RMask[64];
int RAttackIndex[64];
@@ -186,6 +233,8 @@ Bitboard BishopPseudoAttacks[64];
Bitboard RookPseudoAttacks[64];
Bitboard QueenPseudoAttacks[64];
uint8_t BitCount8Bit[256];
////
//// Local definitions
@@ -248,7 +297,8 @@ void init_bitboards() {
#if defined(IS_64BIT) && !defined(USE_BSFQ)
static const int BitTable[64] = {
static CACHE_LINE_ALIGNMENT
const int BitTable[64] = {
0, 1, 2, 7, 3, 13, 8, 19, 4, 25, 14, 28, 9, 34, 20, 40, 5, 17, 26, 38, 15,
46, 29, 48, 10, 31, 35, 54, 21, 50, 41, 57, 63, 6, 12, 18, 24, 27, 33, 39,
16, 37, 45, 47, 30, 53, 49, 56, 62, 11, 23, 32, 36, 44, 52, 55, 61, 22, 43,
@@ -267,7 +317,8 @@ Square pop_1st_bit(Bitboard* b) {
#elif !defined(USE_BSFQ)
static const int BitTable[64] = {
static CACHE_LINE_ALIGNMENT
const int BitTable[64] = {
63, 30, 3, 32, 25, 41, 22, 33, 15, 50, 42, 13, 11, 53, 19, 34, 61, 29, 2,
51, 21, 43, 45, 10, 18, 47, 1, 54, 9, 57, 0, 35, 62, 31, 40, 4, 49, 5, 52,
26, 60, 6, 23, 44, 46, 27, 56, 16, 7, 39, 48, 24, 59, 14, 12, 55, 38, 28,
@@ -285,36 +336,72 @@ union b_union {
Bitboard b;
struct {
#if defined (BIGENDIAN)
uint32_t h;
uint32_t l;
#else
uint32_t l;
uint32_t h;
#endif
} dw;
};
// WARNING: Needs -fno-strict-aliasing compiler option
Square pop_1st_bit(Bitboard* bb) {
b_union u;
uint32_t b;
b_union u;
Square ret;
u.b = *bb;
u.b = *bb;
if (u.dw.l)
{
b = u.dw.l;
*((uint32_t*)bb) = b & (b - 1);
b ^= (b - 1);
}
else
{
b = u.dw.h;
*((uint32_t*)bb+1) = b & (b - 1); // Little endian only?
b = ~(b ^ (b - 1));
}
return Square(BitTable[(b * 0x783a9b23) >> 26]);
if (u.dw.l)
{
ret = Square(BitTable[((u.dw.l ^ (u.dw.l - 1)) * 0x783a9b23) >> 26]);
u.dw.l &= (u.dw.l - 1);
*bb = u.b;
return ret;
}
ret = Square(BitTable[((~(u.dw.h ^ (u.dw.h - 1))) * 0x783a9b23) >> 26]);
u.dw.h &= (u.dw.h - 1);
*bb = u.b;
return ret;
}
#endif
// Optimized bitScanReverse32() implementation by Pascal Georges. Note
// that first bit is 1, this allow to differentiate between 0 and 1.
static CACHE_LINE_ALIGNMENT
const char MsbTable[256] = {
0, 1, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 5, 5,
5, 5, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 7, 7,
7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
7, 7, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8
};
int bitScanReverse32(uint32_t b)
{
int result = 0;
if (b > 0xFFFF)
{
b >>= 16;
result += 16;
}
if (b > 0xFF)
{
b >>= 8;
result += 8;
}
return result + MsbTable[b];
}
namespace {
// All functions below are used to precompute various bitboards during
@@ -335,6 +422,9 @@ namespace {
in_front_bb(c, s) & this_and_neighboring_files_bb(s);
OutpostMask[c][s] = in_front_bb(c, s) & neighboring_files_bb(s);
}
for (Bitboard b = 0ULL; b < 256ULL; b++)
BitCount8Bit[b] = (uint8_t)count_1s(b);
}
+13 -49
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -40,7 +40,6 @@ const Bitboard EmptyBoardBB = 0ULL;
const Bitboard WhiteSquaresBB = 0x55AA55AA55AA55AAULL;
const Bitboard BlackSquaresBB = 0xAA55AA55AA55AA55ULL;
const Bitboard SquaresByColorBB[2] = { BlackSquaresBB, WhiteSquaresBB };
const Bitboard FileABB = 0x0101010101010101ULL;
const Bitboard FileBBB = 0x0202020202020202ULL;
@@ -51,22 +50,6 @@ const Bitboard FileFBB = 0x2020202020202020ULL;
const Bitboard FileGBB = 0x4040404040404040ULL;
const Bitboard FileHBB = 0x8080808080808080ULL;
const Bitboard FileBB[8] = {
FileABB, FileBBB, FileCBB, FileDBB, FileEBB, FileFBB, FileGBB, FileHBB
};
const Bitboard NeighboringFilesBB[8] = {
FileBBB, FileABB|FileCBB, FileBBB|FileDBB, FileCBB|FileEBB,
FileDBB|FileFBB, FileEBB|FileGBB, FileFBB|FileHBB, FileGBB
};
const Bitboard ThisAndNeighboringFilesBB[8] = {
FileABB|FileBBB, FileABB|FileBBB|FileCBB,
FileBBB|FileCBB|FileDBB, FileCBB|FileDBB|FileEBB,
FileDBB|FileEBB|FileFBB, FileEBB|FileFBB|FileGBB,
FileFBB|FileGBB|FileHBB, FileGBB|FileHBB
};
const Bitboard Rank1BB = 0xFFULL;
const Bitboard Rank2BB = 0xFF00ULL;
const Bitboard Rank3BB = 0xFF0000ULL;
@@ -76,35 +59,13 @@ const Bitboard Rank6BB = 0xFF0000000000ULL;
const Bitboard Rank7BB = 0xFF000000000000ULL;
const Bitboard Rank8BB = 0xFF00000000000000ULL;
const Bitboard RankBB[8] = {
Rank1BB, Rank2BB, Rank3BB, Rank4BB, Rank5BB, Rank6BB, Rank7BB, Rank8BB
};
const Bitboard RelativeRankBB[2][8] = {
{ Rank1BB, Rank2BB, Rank3BB, Rank4BB, Rank5BB, Rank6BB, Rank7BB, Rank8BB },
{ Rank8BB, Rank7BB, Rank6BB, Rank5BB, Rank4BB, Rank3BB, Rank2BB, Rank1BB }
};
const Bitboard InFrontBB[2][8] = {
{ Rank2BB | Rank3BB | Rank4BB | Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank3BB | Rank4BB | Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank4BB | Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank5BB | Rank6BB | Rank7BB | Rank8BB,
Rank6BB | Rank7BB | Rank8BB,
Rank7BB | Rank8BB,
Rank8BB,
EmptyBoardBB
},
{ EmptyBoardBB,
Rank1BB,
Rank2BB | Rank1BB,
Rank3BB | Rank2BB | Rank1BB,
Rank4BB | Rank3BB | Rank2BB | Rank1BB,
Rank5BB | Rank4BB | Rank3BB | Rank2BB | Rank1BB,
Rank6BB | Rank5BB | Rank4BB | Rank3BB | Rank2BB | Rank1BB,
Rank7BB | Rank6BB | Rank5BB | Rank4BB | Rank3BB | Rank2BB | Rank1BB
}
};
extern const Bitboard SquaresByColorBB[2];
extern const Bitboard FileBB[8];
extern const Bitboard NeighboringFilesBB[8];
extern const Bitboard ThisAndNeighboringFilesBB[8];
extern const Bitboard RankBB[8];
extern const Bitboard RelativeRankBB[2][8];
extern const Bitboard InFrontBB[2][8];
extern Bitboard SetMaskBB[65];
extern Bitboard ClearMaskBB[65];
@@ -132,6 +93,8 @@ extern Bitboard BishopPseudoAttacks[64];
extern Bitboard RookPseudoAttacks[64];
extern Bitboard QueenPseudoAttacks[64];
extern uint8_t BitCount8Bit[256];
////
//// Inline functions
@@ -360,13 +323,13 @@ inline Bitboard isolated_pawn_mask(Square s) {
#if defined(USE_BSFQ) // Assembly code by Heinz van Saanen
inline Square __attribute__((always_inline)) first_1(Bitboard b) {
inline Square first_1(Bitboard b) {
Bitboard dummy;
__asm__("bsfq %1, %0": "=r"(dummy): "rm"(b) );
return (Square)(dummy);
}
inline Square __attribute__((always_inline)) pop_1st_bit(Bitboard* b) {
inline Square pop_1st_bit(Bitboard* b) {
const Square s = first_1(*b);
*b &= ~(1ULL<<s);
return s;
@@ -386,6 +349,7 @@ extern Square pop_1st_bit(Bitboard* b);
extern void print_bitboard(Bitboard b);
extern void init_bitboards();
extern int bitScanReverse32(uint32_t b);
#endif // !defined(BITBOARD_H_INCLUDED)
+26 -53
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -22,62 +22,22 @@
#if !defined(BITCOUNT_H_INCLUDED)
#define BITCOUNT_H_INCLUDED
// To enable POPCNT support uncomment USE_POPCNT define. For PGO compile on a Core i7
// you may want to collect profile data first with USE_POPCNT disabled and then, in a
// second profiling session, with USE_POPCNT enabled so to exercise both paths. Don't
// forget to leave USE_POPCNT enabled for the final optimized compile though ;-)
//#define USE_POPCNT
#include "types.h"
// Select type of intrinsic bit count instruction to use
#if defined(__INTEL_COMPILER) && defined(IS_64BIT) && defined(USE_POPCNT) // Intel compiler
#include <nmmintrin.h>
inline bool cpu_has_popcnt() {
int CPUInfo[4] = {-1};
__cpuid(CPUInfo, 0x00000001);
return (CPUInfo[2] >> 23) & 1;
}
// Define a dummy template to workaround a compile error if _mm_popcnt_u64() is not defined.
//
// If _mm_popcnt_u64() is defined in <nmmintrin.h> it will be choosen first due to
// C++ overload rules that always prefer a function to a template with the same name.
// If not, we avoid a compile error and because cpu_has_popcnt() should return false,
// our templetized _mm_popcnt_u64() is never called anyway.
template<typename T> inline unsigned _mm_popcnt_u64(T) { return 0; } // Is never called
#define POPCNT_INTRINSIC(x) _mm_popcnt_u64(x)
#elif defined(_MSC_VER) && defined(IS_64BIT) && defined(USE_POPCNT) // Microsoft compiler
#include <intrin.h>
inline bool cpu_has_popcnt() {
int CPUInfo[4] = {-1};
__cpuid(CPUInfo, 0x00000001);
return (CPUInfo[2] >> 23) & 1;
}
// See comment of _mm_popcnt_u64<>() few lines above for an explanation.
template<typename T> inline unsigned __popcnt64(T) { return 0; } // Is never called
#define POPCNT_INTRINSIC(x) __popcnt64(x)
#else // Safe fallback for unsupported compilers or when USE_POPCNT is disabled
inline bool cpu_has_popcnt() { return false; }
// Select type of intrinsic bit count instruction to use, see
// README.txt on how to pgo compile with POPCNT support.
#if !defined(USE_POPCNT)
#define POPCNT_INTRINSIC(x) 0
#elif defined(_MSC_VER)
#define POPCNT_INTRINSIC(x) (int)__popcnt64(x)
#elif defined(__GNUC__)
#endif // cpu_has_popcnt() and POPCNT_INTRINSIC() definitions
#define POPCNT_INTRINSIC(x) ({ \
unsigned long __ret; \
__asm__("popcnt %1, %0" : "=r" (__ret) : "r" (x)); \
__ret; })
#endif
/// Software implementation of bit count functions
@@ -144,10 +104,23 @@ inline int count_1s_max_15(Bitboard b) {
}
// Detect hardware POPCNT support
inline bool cpu_has_popcnt() {
int CPUInfo[4] = {-1};
__cpuid(CPUInfo, 0x00000001);
return (CPUInfo[2] >> 23) & 1;
}
// Global constant initialized at startup that is set to true if
// CPU on which application runs supports POPCNT intrinsic. Unless
// USE_POPCNT is not defined.
#if defined(USE_POPCNT)
const bool CpuHasPOPCNT = cpu_has_popcnt();
#else
const bool CpuHasPOPCNT = false;
#endif
// Global constant used to print info about the use of 64 optimized
+7 -7
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -390,7 +390,7 @@ void Book::close() {
/// Book::file_name() returns the file name of the currently active book,
/// or the empty string if no book is open.
const string Book::file_name() const {
const string Book::file_name() { // Not const to compile on HP-UX 11.X
return is_open() ? fileName : "";
}
@@ -429,11 +429,11 @@ Move Book::get_move(const Position& pos) {
if (!bookMove)
return MOVE_NONE;
MoveStack moves[256];
int n = generate_legal_moves(pos, moves);
for (int j = 0; j < n; j++)
if ((int(moves[j].move) & 07777) == bookMove)
return moves[j].move;
MoveStack mlist[256];
MoveStack* last = generate_moves(pos, mlist);
for (MoveStack* cur = mlist; cur != last; cur++)
if ((int(cur->move) & 07777) == bookMove)
return cur->move;
return MOVE_NONE;
}
+5 -2
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -53,11 +53,14 @@ struct BookEntry {
};
class Book : private std::ifstream {
Book(const Book&); // just decleared..
Book& operator=(const Book&); // ..to avoid a warning
public:
Book() {}
~Book();
void open(const std::string& fName);
void close();
const std::string file_name() const;
const std::string file_name();
Move get_move(const Position& pos);
private:
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+3 -2
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -48,7 +48,8 @@ inline void operator+= (Depth &d, int i) { d = Depth(int(d) + i); }
inline void operator+= (Depth &d1, Depth d2) { d1 += int(d2); }
inline Depth operator- (Depth d, int i) { return Depth(int(d) - i); }
inline Depth operator- (Depth d1, Depth d2) { return Depth(int(d1) - int(d2)); }
inline void operator-= (Depth & d, int i) { d = Depth(int(d) - i); }
inline void operator-= (Depth &d, int i) { d = Depth(int(d) - i); }
inline void operator-= (Depth &d1, Depth d2) { d1 -= int(d2); }
inline Depth operator* (Depth d, int i) { return Depth(int(d) * i); }
inline Depth operator* (int i, Depth d) { return Depth(int(d) * i); }
inline void operator*= (Depth &d, int i) { d = Depth(int(d) * i); }
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+22 -51
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -29,41 +29,6 @@
#include "endgame.h"
////
//// Constants and variables
////
/// Evaluation functions
// Generic "mate lone king" eval
EvaluationFunction<KXK> EvaluateKXK(WHITE), EvaluateKKX(BLACK);
// K and two minors vs K and one or two minors
EvaluationFunction<KmmKm> EvaluateKmmKm(WHITE);
EvaluationFunction<KBNK> EvaluateKBNK(WHITE), EvaluateKKBN(BLACK); // KBN vs K
EvaluationFunction<KPK> EvaluateKPK(WHITE), EvaluateKKP(BLACK); // KP vs K
EvaluationFunction<KRKP> EvaluateKRKP(WHITE), EvaluateKPKR(BLACK); // KR vs KP
EvaluationFunction<KRKB> EvaluateKRKB(WHITE), EvaluateKBKR(BLACK); // KR vs KB
EvaluationFunction<KRKN> EvaluateKRKN(WHITE), EvaluateKNKR(BLACK); // KR vs KN
EvaluationFunction<KQKR> EvaluateKQKR(WHITE), EvaluateKRKQ(BLACK); // KQ vs KR
EvaluationFunction<KBBKN> EvaluateKBBKN(WHITE), EvaluateKNKBB(BLACK); // KBB vs KN
/// Scaling functions
ScalingFunction<KBPK> ScaleKBPK(WHITE), ScaleKKBP(BLACK); // KBP vs K
ScalingFunction<KQKRP> ScaleKQKRP(WHITE), ScaleKRPKQ(BLACK); // KQ vs KRP
ScalingFunction<KRPKR> ScaleKRPKR(WHITE), ScaleKRKRP(BLACK); // KRP vs KR
ScalingFunction<KRPPKRP> ScaleKRPPKRP(WHITE), ScaleKRPKRPP(BLACK); // KRPP vs KRP
ScalingFunction<KPsK> ScaleKPsK(WHITE), ScaleKKPs(BLACK); // King and pawns vs king
ScalingFunction<KBPKB> ScaleKBPKB(WHITE), ScaleKBKBP(BLACK); // KBP vs KB
ScalingFunction<KBPPKB> ScaleKBPPKB(WHITE), ScaleKBKBPP(BLACK); // KBPP vs KB
ScalingFunction<KBPKN> ScaleKBPKN(WHITE), ScaleKNKBP(BLACK); // KBP vs KN
ScalingFunction<KNPK> ScaleKNPK(WHITE), ScaleKKNP(BLACK); // KNP vs K
ScalingFunction<KPKP> ScaleKPKPw(WHITE), ScaleKPKPb(BLACK); // KPKP
////
//// Local definitions
////
@@ -364,7 +329,7 @@ Value EvaluationFunction<KBBKN>::apply(const Position& pos) {
assert(pos.non_pawn_material(strongerSide) == 2*BishopValueMidgame);
assert(pos.piece_count(weakerSide, KNIGHT) == 1);
assert(pos.non_pawn_material(weakerSide) == KnightValueMidgame);
assert(pos.pawns() == EmptyBoardBB);
assert(pos.pieces(PAWN) == EmptyBoardBB);
Value result = BishopValueEndgame;
Square wksq = pos.king_square(strongerSide);
@@ -378,16 +343,23 @@ Value EvaluationFunction<KBBKN>::apply(const Position& pos) {
result += Value(square_distance(bksq, nsq) * 32);
// Bonus for restricting the knight's mobility
result += Value((8 - count_1s_max_15(pos.piece_attacks<KNIGHT>(nsq))) * 8);
result += Value((8 - count_1s_max_15(pos.attacks_from<KNIGHT>(nsq))) * 8);
return (strongerSide == pos.side_to_move() ? result : -result);
}
/// K and two minors vs K and one or two minors or K and two knights against
/// king alone are always draw.
template<>
Value EvaluationFunction<KmmKm>::apply(const Position&) {
return Value(0);
}
template<>
Value EvaluationFunction<KNNK>::apply(const Position&) {
return Value(0);
}
/// KBPKScalingFunction scales endgames where the stronger side has king,
/// bishop and one or more pawns. It checks for draws with rook pawns and a
@@ -395,7 +367,7 @@ Value EvaluationFunction<KmmKm>::apply(const Position&) {
/// returned. If not, the return value is SCALE_FACTOR_NONE, i.e. no scaling
/// will be used.
template<>
ScaleFactor ScalingFunction<KBPK>::apply(const Position& pos) {
ScaleFactor ScalingFunction<KBPsK>::apply(const Position& pos) {
assert(pos.non_pawn_material(strongerSide) == BishopValueMidgame);
assert(pos.piece_count(strongerSide, BISHOP) == 1);
@@ -404,7 +376,7 @@ ScaleFactor ScalingFunction<KBPK>::apply(const Position& pos) {
// No assertions about the material of weakerSide, because we want draws to
// be detected even when the weaker side has some pawns.
Bitboard pawns = pos.pawns(strongerSide);
Bitboard pawns = pos.pieces(PAWN, strongerSide);
File pawnFile = square_file(pos.piece_list(strongerSide, PAWN, 0));
// All pawns are on a single rook file ?
@@ -421,7 +393,6 @@ ScaleFactor ScalingFunction<KBPK>::apply(const Position& pos) {
// The bishop has the wrong color, and the defending king is on the
// file of the pawn(s) or the neighboring file. Find the rank of the
// frontmost pawn.
Rank rank;
if (strongerSide == WHITE)
{
@@ -450,7 +421,7 @@ ScaleFactor ScalingFunction<KBPK>::apply(const Position& pos) {
/// It tests for fortress draws with a rook on the third rank defended by
/// a pawn.
template<>
ScaleFactor ScalingFunction<KQKRP>::apply(const Position& pos) {
ScaleFactor ScalingFunction<KQKRPs>::apply(const Position& pos) {
assert(pos.non_pawn_material(strongerSide) == QueenValueMidgame);
assert(pos.piece_count(strongerSide, QUEEN) == 1);
@@ -461,12 +432,12 @@ ScaleFactor ScalingFunction<KQKRP>::apply(const Position& pos) {
Square kingSq = pos.king_square(weakerSide);
if ( relative_rank(weakerSide, kingSq) <= RANK_2
&& relative_rank(weakerSide, pos.king_square(strongerSide)) >= RANK_4
&& (pos.rooks(weakerSide) & relative_rank_bb(weakerSide, RANK_3))
&& (pos.pawns(weakerSide) & relative_rank_bb(weakerSide, RANK_2))
&& (pos.piece_attacks<KING>(kingSq) & pos.pawns(weakerSide)))
&& (pos.pieces(ROOK, weakerSide) & relative_rank_bb(weakerSide, RANK_3))
&& (pos.pieces(PAWN, weakerSide) & relative_rank_bb(weakerSide, RANK_2))
&& (pos.attacks_from<KING>(kingSq) & pos.pieces(PAWN, weakerSide)))
{
Square rsq = pos.piece_list(weakerSide, ROOK, 0);
if (pos.pawn_attacks(strongerSide, rsq) & pos.pawns(weakerSide))
if (pos.attacks_from<PAWN>(rsq, strongerSide) & pos.pieces(PAWN, weakerSide))
return ScaleFactor(0);
}
return SCALE_FACTOR_NONE;
@@ -645,7 +616,7 @@ ScaleFactor ScalingFunction<KPsK>::apply(const Position &pos) {
assert(pos.non_pawn_material(weakerSide) == Value(0));
assert(pos.piece_count(weakerSide, PAWN) == 0);
Bitboard pawns = pos.pawns(strongerSide);
Bitboard pawns = pos.pieces(PAWN, strongerSide);
// Are all pawns on the 'a' file?
if ((pawns & ~FileABB) == EmptyBoardBB)
@@ -723,9 +694,9 @@ ScaleFactor ScalingFunction<KBPKB>::apply(const Position &pos) {
else
{
Bitboard ray = ray_bb(pawnSq, (strongerSide == WHITE)? SIGNED_DIR_N : SIGNED_DIR_S);
if (ray & pos.kings(weakerSide))
if (ray & pos.pieces(KING, weakerSide))
return ScaleFactor(0);
if( (pos.piece_attacks<BISHOP>(weakerBishopSq) & ray)
if( (pos.attacks_from<BISHOP>(weakerBishopSq) & ray)
&& square_distance(weakerBishopSq, pawnSq) >= 3)
return ScaleFactor(0);
}
@@ -790,13 +761,13 @@ ScaleFactor ScalingFunction<KBPPKB>::apply(const Position& pos) {
if ( ksq == blockSq1
&& square_color(ksq) != square_color(wbsq)
&& ( bbsq == blockSq2
|| (pos.piece_attacks<BISHOP>(blockSq2) & pos.bishops(weakerSide))
|| (pos.attacks_from<BISHOP>(blockSq2) & pos.pieces(BISHOP, weakerSide))
|| rank_distance(r1, r2) >= 2))
return ScaleFactor(0);
else if ( ksq == blockSq2
&& square_color(ksq) != square_color(wbsq)
&& ( bbsq == blockSq1
|| (pos.piece_attacks<BISHOP>(blockSq1) & pos.bishops(weakerSide))))
|| (pos.attacks_from<BISHOP>(blockSq1) & pos.pieces(BISHOP, weakerSide))))
return ScaleFactor(0);
else
return SCALE_FACTOR_NONE;
+7 -28
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -45,11 +45,12 @@ enum EndgameType {
KRKN, // KR vs KN
KQKR, // KQ vs KR
KBBKN, // KBB vs KN
KNNK, // KNN vs K
KmmKm, // K and two minors vs K and one or two minors
// Scaling functions
KBPK, // KBP vs K
KQKRP, // KQ vs KRP
KBPsK, // KB+pawns vs K
KQKRPs, // KQ vs KR+pawns
KRPKR, // KRP vs KR
KRPPKRP, // KRPP vs KRP
KPsK, // King and pawns vs king
@@ -68,6 +69,7 @@ public:
EndgameFunctionBase(Color c) : strongerSide(c), weakerSide(opposite_color(c)) {}
virtual ~EndgameFunctionBase() {}
virtual T apply(const Position&) = 0;
Color color() const { return strongerSide; }
protected:
Color strongerSide, weakerSide;
@@ -81,42 +83,19 @@ typedef EndgameFunctionBase<ScaleFactor> EndgameScalingFunctionBase;
template<EndgameType>
struct EvaluationFunction : public EndgameEvaluationFunctionBase {
typedef EndgameEvaluationFunctionBase Base;
explicit EvaluationFunction(Color c): EndgameEvaluationFunctionBase(c) {}
Value apply(const Position&);
};
template<EndgameType>
struct ScalingFunction : public EndgameScalingFunctionBase {
typedef EndgameScalingFunctionBase Base;
explicit ScalingFunction(Color c) : EndgameScalingFunctionBase(c) {}
ScaleFactor apply(const Position&);
};
////
//// Constants and variables
////
extern EvaluationFunction<KXK> EvaluateKXK, EvaluateKKX; // Generic "mate lone king" eval
extern EvaluationFunction<KBNK> EvaluateKBNK, EvaluateKKBN; // KBN vs K
extern EvaluationFunction<KPK> EvaluateKPK, EvaluateKKP; // KP vs K
extern EvaluationFunction<KRKP> EvaluateKRKP, EvaluateKPKR; // KR vs KP
extern EvaluationFunction<KRKB> EvaluateKRKB, EvaluateKBKR; // KR vs KB
extern EvaluationFunction<KRKN> EvaluateKRKN, EvaluateKNKR; // KR vs KN
extern EvaluationFunction<KQKR> EvaluateKQKR, EvaluateKRKQ; // KQ vs KR
extern EvaluationFunction<KBBKN> EvaluateKBBKN, EvaluateKNKBB; // KBB vs KN
extern EvaluationFunction<KmmKm> EvaluateKmmKm; // K and two minors vs K and one or two minors:
extern ScalingFunction<KBPK> ScaleKBPK, ScaleKKBP; // KBP vs K
extern ScalingFunction<KQKRP> ScaleKQKRP, ScaleKRPKQ; // KQ vs KRP
extern ScalingFunction<KRPKR> ScaleKRPKR, ScaleKRKRP; // KRP vs KR
extern ScalingFunction<KRPPKRP> ScaleKRPPKRP, ScaleKRPKRPP; // KRPP vs KRP
extern ScalingFunction<KPsK> ScaleKPsK, ScaleKKPs; // King and pawns vs king
extern ScalingFunction<KBPKB> ScaleKBPKB, ScaleKBKBP; // KBP vs KB
extern ScalingFunction<KBPPKB> ScaleKBPPKB, ScaleKBKBPP; // KBPP vs KB
extern ScalingFunction<KBPKN> ScaleKBPKN, ScaleKNKBP; // KBP vs KN
extern ScalingFunction<KNPK> ScaleKNPK, ScaleKKNP; // KNP vs K
extern ScalingFunction<KPKP> ScaleKPKPw, ScaleKPKPb; // KP vs KP
////
//// Prototypes
////
+574 -580
View File
File diff suppressed because it is too large Load Diff
+9 -6
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -25,6 +25,8 @@
//// Includes
////
#include <iostream>
#include "material.h"
#include "pawns.h"
@@ -45,8 +47,10 @@ class Position;
struct EvalInfo {
EvalInfo() { futilityMargin[0] = futilityMargin[1] = Value(0); }
// Middle game and endgame evaluations
Value mgValue, egValue;
Score value;
// Pointers to material and pawn hash table entries
MaterialInfo* mi;
@@ -89,11 +93,11 @@ struct EvalInfo {
Move mateThreat[2];
// Middle game and endgame mobility scores.
Value mgMobility, egMobility;
Score mobility;
// Extra futility margin. This is added to the standard futility margin
// in the quiescence search.
Value futilityMargin;
// in the quiescence search. One for each color.
Value futilityMargin[2];
};
@@ -102,7 +106,6 @@ struct EvalInfo {
////
extern Value evaluate(const Position& pos, EvalInfo& ei, int threadID);
extern Value quick_evaluate(const Position& pos);
extern void init_eval(int threads);
extern void quit_eval();
extern void read_weights(Color sideToMove);
+25 -15
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -42,8 +42,7 @@ History::History() { clear(); }
void History::clear() {
memset(history, 0, 2 * 8 * 64 * sizeof(int));
memset(successCount, 0, 2 * 8 * 64 * sizeof(int));
memset(failureCount, 0, 2 * 8 * 64 * sizeof(int));
memset(maxStaticValueDelta, 0, 2 * 8 * 64 * sizeof(int));
}
@@ -58,13 +57,12 @@ void History::success(Piece p, Square to, Depth d) {
assert(square_is_ok(to));
history[p][to] += int(d) * int(d);
successCount[p][to]++;
// Prevent history overflow
if (history[p][to] >= HistoryMax)
for (int i = 0; i < 16; i++)
for (int j = 0; j < 64; j++)
history[i][j] /= 4;
history[i][j] /= 2;
}
@@ -72,12 +70,18 @@ void History::success(Piece p, Square to, Depth d) {
/// called for each non-capturing move which failed to produce a beta cutoff
/// at a node where a beta cutoff was finally found.
void History::failure(Piece p, Square to) {
void History::failure(Piece p, Square to, Depth d) {
assert(piece_is_ok(p));
assert(square_is_ok(to));
failureCount[p][to]++;
history[p][to] -= int(d) * int(d);
// Prevent history underflow
if (history[p][to] <= -HistoryMax)
for (int i = 0; i < 16; i++)
for (int j = 0; j < 64; j++)
history[i][j] /= 2;
}
@@ -93,13 +97,19 @@ int History::move_ordering_score(Piece p, Square to) const {
}
/// History::ok_to_prune() decides whether a move has been sufficiently
/// unsuccessful that it makes sense to prune it entirely.
/// History::set_gain() and History::gain() store and retrieve the
/// gain of a move given the delta of the static position evaluations
/// before and after the move.
bool History::ok_to_prune(Piece p, Square to, Depth d) const {
assert(piece_is_ok(p));
assert(square_is_ok(to));
return (int(d) * successCount[p][to] < failureCount[p][to]);
void History::set_gain(Piece p, Square to, Value delta)
{
if (delta >= maxStaticValueDelta[p][to])
maxStaticValueDelta[p][to] = delta;
else
maxStaticValueDelta[p][to]--;
}
Value History::gain(Piece p, Square to) const
{
return Value(maxStaticValueDelta[p][to]);
}
+7 -6
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -28,6 +28,7 @@
#include "depth.h"
#include "move.h"
#include "piece.h"
#include "value.h"
////
@@ -47,14 +48,14 @@ public:
History();
void clear();
void success(Piece p, Square to, Depth d);
void failure(Piece p, Square to);
void failure(Piece p, Square to, Depth d);
int move_ordering_score(Piece p, Square to) const;
bool ok_to_prune(Piece p, Square to, Depth d) const;
void set_gain(Piece p, Square to, Value delta);
Value gain(Piece p, Square to) const;
private:
int history[16][64]; // [piece][square]
int successCount[16][64];
int failureCount[16][64];
int maxStaticValueDelta[16][64]; // [piece][from_square][to_square]
};
@@ -70,7 +71,7 @@ private:
/// recently have a bigger importance for move ordering than the moves which
/// have been searched a long time ago.
const int HistoryMax = 25000 * OnePly;
const int HistoryMax = 50000 * OnePly;
#endif // !defined(HISTORY_H_INCLUDED)
+3 -3
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -86,8 +86,9 @@ typedef pthread_mutex_t Lock;
#else
#define WIN32_LEAN_AND_MEAN
#include <windows.h>
#undef WIN32_LEAN_AND_MEAN
typedef CRITICAL_SECTION Lock;
# define lock_init(x, y) InitializeCriticalSection(x)
@@ -98,5 +99,4 @@ typedef CRITICAL_SECTION Lock;
#endif
#endif // !defined(LOCK_H_INCLUDED)
+2 -2
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -63,7 +63,7 @@ int main(int argc, char *argv[]) {
if (string(argv[1]) != "bench" || argc < 4 || argc > 8)
cout << "Usage: stockfish bench <hash size> <threads> "
<< "[time = 60s] [fen positions file = default] "
<< "[time, depth or node limited = time] "
<< "[time, depth, perft or node limited = time] "
<< "[timing file name = none]" << endl;
else
{
+168 -130
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -28,7 +28,8 @@
#include "material.h"
using std::string;
using namespace std;
////
//// Local definitions
@@ -37,67 +38,95 @@ using std::string;
namespace {
// Values modified by Joona Kiiski
const Value BishopPairMidgameBonus = Value(109);
const Value BishopPairEndgameBonus = Value(97);
const Value MidgameLimit = Value(15581);
const Value EndgameLimit = Value(3998);
Key KNNKMaterialKey, KKNNMaterialKey;
// Polynomial material balance parameters
const Value RedundantQueenPenalty = Value(320);
const Value RedundantRookPenalty = Value(554);
const int LinearCoefficients[6] = { 1617, -162, -1172, -190, 105, 26 };
const int QuadraticCoefficientsSameColor[][6] = {
{ 7, 7, 7, 7, 7, 7 }, { 39, 2, 7, 7, 7, 7 }, { 35, 271, -4, 7, 7, 7 },
{ 7, 25, 4, 7, 7, 7 }, { -27, -2, 46, 100, 56, 7 }, { 58, 29, 83, 148, -3, -25 } };
const int QuadraticCoefficientsOppositeColor[][6] = {
{ 41, 41, 41, 41, 41, 41 }, { 37, 41, 41, 41, 41, 41 }, { 10, 62, 41, 41, 41, 41 },
{ 57, 64, 39, 41, 41, 41 }, { 50, 40, 23, -22, 41, 41 }, { 106, 101, 3, 151, 171, 41 } };
// Named endgame evaluation and scaling functions, these
// are accessed direcly and not through the function maps.
EvaluationFunction<KmmKm> EvaluateKmmKm(WHITE);
EvaluationFunction<KXK> EvaluateKXK(WHITE), EvaluateKKX(BLACK);
ScalingFunction<KBPsK> ScaleKBPsK(WHITE), ScaleKKBPs(BLACK);
ScalingFunction<KQKRPs> ScaleKQKRPs(WHITE), ScaleKRPsKQ(BLACK);
ScalingFunction<KPsK> ScaleKPsK(WHITE), ScaleKKPs(BLACK);
ScalingFunction<KPKP> ScaleKPKPw(WHITE), ScaleKPKPb(BLACK);
typedef EndgameEvaluationFunctionBase EF;
typedef EndgameScalingFunctionBase SF;
}
////
//// Classes
////
/// See header for a class description. It is declared here to avoid
/// to include <map> in the header file.
/// EndgameFunctions class stores endgame evaluation and scaling functions
/// in two std::map. Because STL library is not guaranteed to be thread
/// safe even for read access, the maps, although with identical content,
/// are replicated for each thread. This is faster then using locks.
class EndgameFunctions {
public:
EndgameFunctions();
EndgameEvaluationFunctionBase* getEEF(Key key) const;
EndgameScalingFunctionBase* getESF(Key key, Color* c) const;
~EndgameFunctions();
template<class T> T* get(Key key) const;
private:
void add(const string& keyCode, EndgameEvaluationFunctionBase* f);
void add(const string& keyCode, Color c, EndgameScalingFunctionBase* f);
Key buildKey(const string& keyCode);
template<class T> void add(const string& keyCode);
struct ScalingInfo
{
Color col;
EndgameScalingFunctionBase* fun;
};
static Key buildKey(const string& keyCode);
static const string swapColors(const string& keyCode);
std::map<Key, EndgameEvaluationFunctionBase*> EEFmap;
std::map<Key, ScalingInfo> ESFmap;
// Here we store two maps, for evaluate and scaling functions
pair<map<Key, EF*>, map<Key, SF*> > maps;
// Maps accessing functions returning const and non-const references
template<typename T> const map<Key, T*>& get() const { return maps.first; }
template<typename T> map<Key, T*>& get() { return maps.first; }
};
// Explicit specializations of a member function shall be declared in
// the namespace of which the class template is a member.
template<> const map<Key, SF*>&
EndgameFunctions::get<SF>() const { return maps.second; }
template<> map<Key, SF*>&
EndgameFunctions::get<SF>() { return maps.second; }
////
//// Functions
////
/// Constructor for the MaterialInfoTable class
/// MaterialInfoTable c'tor and d'tor, called once by each thread
MaterialInfoTable::MaterialInfoTable(unsigned int numOfEntries) {
size = numOfEntries;
entries = new MaterialInfo[size];
funcs = new EndgameFunctions();
if (!entries || !funcs)
{
std::cerr << "Failed to allocate " << (numOfEntries * sizeof(MaterialInfo))
<< " bytes for material hash table." << std::endl;
cerr << "Failed to allocate " << numOfEntries * sizeof(MaterialInfo)
<< " bytes for material hash table." << endl;
Application::exit_with_failure();
}
}
/// Destructor for the MaterialInfoTable class
MaterialInfoTable::~MaterialInfoTable() {
delete funcs;
@@ -105,6 +134,22 @@ MaterialInfoTable::~MaterialInfoTable() {
}
/// MaterialInfoTable::game_phase() calculates the phase given the current
/// position. Because the phase is strictly a function of the material, it
/// is stored in MaterialInfo.
Phase MaterialInfoTable::game_phase(const Position& pos) {
Value npm = pos.non_pawn_material(WHITE) + pos.non_pawn_material(BLACK);
if (npm >= MidgameLimit)
return PHASE_MIDGAME;
else if (npm <= EndgameLimit)
return PHASE_ENDGAME;
return Phase(((npm - EndgameLimit) * 128) / (MidgameLimit - EndgameLimit));
}
/// MaterialInfoTable::get_material_info() takes a position object as input,
/// computes or looks up a MaterialInfo object, and returns a pointer to it.
/// If the material configuration is not already present in the table, it
@@ -127,18 +172,13 @@ MaterialInfo* MaterialInfoTable::get_material_info(const Position& pos) {
mi->clear();
mi->key = key;
// A special case before looking for a specialized evaluation function
// KNN vs K is a draw.
if (key == KNNKMaterialKey || key == KKNNMaterialKey)
{
mi->factor[WHITE] = mi->factor[BLACK] = 0;
return mi;
}
// Store game phase
mi->gamePhase = MaterialInfoTable::game_phase(pos);
// Let's look if we have a specialized evaluation function for this
// particular material configuration. First we look for a fixed
// configuration one, then a generic one if previous search failed.
if ((mi->evaluationFunction = funcs->getEEF(key)) != NULL)
if ((mi->evaluationFunction = funcs->get<EF>(key)) != NULL)
return mi;
else if ( pos.non_pawn_material(BLACK) == Value(0)
@@ -155,14 +195,14 @@ MaterialInfo* MaterialInfoTable::get_material_info(const Position& pos) {
mi->evaluationFunction = &EvaluateKKX;
return mi;
}
else if ( pos.pawns() == EmptyBoardBB
&& pos.rooks() == EmptyBoardBB
&& pos.queens() == EmptyBoardBB)
else if ( pos.pieces(PAWN) == EmptyBoardBB
&& pos.pieces(ROOK) == EmptyBoardBB
&& pos.pieces(QUEEN) == EmptyBoardBB)
{
// Minor piece endgame with at least one minor piece per side,
// and no pawns.
assert(pos.knights(WHITE) | pos.bishops(WHITE));
assert(pos.knights(BLACK) | pos.bishops(BLACK));
// Minor piece endgame with at least one minor piece per side and
// no pawns. Note that the case KmmK is already handled by KXK.
assert((pos.pieces(KNIGHT, WHITE) | pos.pieces(BISHOP, WHITE)));
assert((pos.pieces(KNIGHT, BLACK) | pos.pieces(BISHOP, BLACK)));
if ( pos.piece_count(WHITE, BISHOP) + pos.piece_count(WHITE, KNIGHT) <= 2
&& pos.piece_count(BLACK, BISHOP) + pos.piece_count(BLACK, KNIGHT) <= 2)
@@ -176,41 +216,43 @@ MaterialInfo* MaterialInfoTable::get_material_info(const Position& pos) {
// material configuration. Is there a suitable scaling function?
//
// The code below is rather messy, and it could easily get worse later,
// if we decide to add more special cases. We face problems when there
// if we decide to add more special cases. We face problems when there
// are several conflicting applicable scaling functions and we need to
// decide which one to use.
Color c;
EndgameScalingFunctionBase* sf;
SF* sf;
if ((sf = funcs->getESF(key, &c)) != NULL)
if ((sf = funcs->get<SF>(key)) != NULL)
{
mi->scalingFunction[c] = sf;
mi->scalingFunction[sf->color()] = sf;
return mi;
}
// Generic scaling functions that refer to more then one material
// distribution. Should be probed after the specialized ones.
// Note that these ones don't return after setting the function.
if ( pos.non_pawn_material(WHITE) == BishopValueMidgame
&& pos.piece_count(WHITE, BISHOP) == 1
&& pos.piece_count(WHITE, PAWN) >= 1)
mi->scalingFunction[WHITE] = &ScaleKBPK;
mi->scalingFunction[WHITE] = &ScaleKBPsK;
if ( pos.non_pawn_material(BLACK) == BishopValueMidgame
&& pos.piece_count(BLACK, BISHOP) == 1
&& pos.piece_count(BLACK, PAWN) >= 1)
mi->scalingFunction[BLACK] = &ScaleKKBP;
mi->scalingFunction[BLACK] = &ScaleKKBPs;
if ( pos.piece_count(WHITE, PAWN) == 0
&& pos.non_pawn_material(WHITE) == QueenValueMidgame
&& pos.piece_count(WHITE, QUEEN) == 1
&& pos.piece_count(BLACK, ROOK) == 1
&& pos.piece_count(BLACK, PAWN) >= 1)
mi->scalingFunction[WHITE] = &ScaleKQKRP;
mi->scalingFunction[WHITE] = &ScaleKQKRPs;
else if ( pos.piece_count(BLACK, PAWN) == 0
&& pos.non_pawn_material(BLACK) == QueenValueMidgame
&& pos.piece_count(BLACK, QUEEN) == 1
&& pos.piece_count(WHITE, ROOK) == 1
&& pos.piece_count(WHITE, PAWN) >= 1)
mi->scalingFunction[BLACK] = &ScaleKRPKQ;
mi->scalingFunction[BLACK] = &ScaleKRPsKQ;
if (pos.non_pawn_material(WHITE) + pos.non_pawn_material(BLACK) == Value(0))
{
@@ -226,6 +268,8 @@ MaterialInfo* MaterialInfoTable::get_material_info(const Position& pos) {
}
else if (pos.piece_count(WHITE, PAWN) == 1 && pos.piece_count(BLACK, PAWN) == 1)
{
// This is a special case because we set scaling functions
// for both colors instead of only one.
mi->scalingFunction[WHITE] = &ScaleKPKPw;
mi->scalingFunction[BLACK] = &ScaleKPKPb;
}
@@ -244,10 +288,13 @@ MaterialInfo* MaterialInfoTable::get_material_info(const Position& pos) {
}
// Evaluate the material balance
int sign;
Value egValue = Value(0);
Value mgValue = Value(0);
const int pieceCount[2][6] = { { pos.piece_count(WHITE, BISHOP) > 1, pos.piece_count(WHITE, PAWN), pos.piece_count(WHITE, KNIGHT),
pos.piece_count(WHITE, BISHOP), pos.piece_count(WHITE, ROOK), pos.piece_count(WHITE, QUEEN) },
{ pos.piece_count(BLACK, BISHOP) > 1, pos.piece_count(BLACK, PAWN), pos.piece_count(BLACK, KNIGHT),
pos.piece_count(BLACK, BISHOP), pos.piece_count(BLACK, ROOK), pos.piece_count(BLACK, QUEEN) } };
Color c, them;
int sign, pt1, pt2, pc;
int v, vv, matValue = 0;
for (c = WHITE, sign = 1; c <= BLACK; c++, sign = -sign)
{
@@ -274,73 +321,68 @@ MaterialInfo* MaterialInfoTable::get_material_info(const Position& pos) {
}
}
// Bishop pair
if (pos.piece_count(c, BISHOP) >= 2)
{
mgValue += sign * BishopPairMidgameBonus;
egValue += sign * BishopPairEndgameBonus;
}
// Knights are stronger when there are many pawns on the board. The
// formula is taken from Larry Kaufman's paper "The Evaluation of Material
// Imbalances in Chess":
// Redundancy of major pieces, formula based on Kaufman's paper
// "The Evaluation of Material Imbalances in Chess"
// http://mywebpages.comcast.net/danheisman/Articles/evaluation_of_material_imbalance.htm
mgValue += sign * Value(pos.piece_count(c, KNIGHT)*(pos.piece_count(c, PAWN)-5)*16);
egValue += sign * Value(pos.piece_count(c, KNIGHT)*(pos.piece_count(c, PAWN)-5)*16);
if (pieceCount[c][ROOK] >= 1)
matValue -= sign * ((pieceCount[c][ROOK] - 1) * RedundantRookPenalty + pieceCount[c][QUEEN] * RedundantQueenPenalty);
// Redundancy of major pieces, again based on Kaufman's paper:
if (pos.piece_count(c, ROOK) >= 1)
them = opposite_color(c);
v = 0;
// Second-degree polynomial material imbalance by Tord Romstad
//
// We use NO_PIECE_TYPE as a place holder for the bishop pair "extended piece",
// this allow us to be more flexible in defining bishop pair bonuses.
for (pt1 = NO_PIECE_TYPE; pt1 <= QUEEN; pt1++)
{
Value v = Value((pos.piece_count(c, ROOK) - 1) * 32 + pos.piece_count(c, QUEEN) * 16);
mgValue -= sign * v;
egValue -= sign * v;
pc = pieceCount[c][pt1];
if (!pc)
continue;
vv = LinearCoefficients[pt1];
for (pt2 = NO_PIECE_TYPE; pt2 <= pt1; pt2++)
vv += pieceCount[c][pt2] * QuadraticCoefficientsSameColor[pt1][pt2]
+ pieceCount[them][pt2] * QuadraticCoefficientsOppositeColor[pt1][pt2];
v += pc * vv;
}
matValue += sign * v;
}
mi->mgValue = int16_t(mgValue);
mi->egValue = int16_t(egValue);
mi->value = int16_t(matValue / 16);
return mi;
}
/// EndgameFunctions member definitions. This class is used to store the maps
/// of end game and scaling functions that MaterialInfoTable will query for
/// each key. The maps are constant and are populated only at construction,
/// but are per-thread instead of globals to avoid expensive locks.
/// EndgameFunctions member definitions.
EndgameFunctions::EndgameFunctions() {
KNNKMaterialKey = buildKey("KNNK");
KKNNMaterialKey = buildKey("KKNN");
add<EvaluationFunction<KNNK> >("KNNK");
add<EvaluationFunction<KPK> >("KPK");
add<EvaluationFunction<KBNK> >("KBNK");
add<EvaluationFunction<KRKP> >("KRKP");
add<EvaluationFunction<KRKB> >("KRKB");
add<EvaluationFunction<KRKN> >("KRKN");
add<EvaluationFunction<KQKR> >("KQKR");
add<EvaluationFunction<KBBKN> >("KBBKN");
add("KPK", &EvaluateKPK);
add("KKP", &EvaluateKKP);
add("KBNK", &EvaluateKBNK);
add("KKBN", &EvaluateKKBN);
add("KRKP", &EvaluateKRKP);
add("KPKR", &EvaluateKPKR);
add("KRKB", &EvaluateKRKB);
add("KBKR", &EvaluateKBKR);
add("KRKN", &EvaluateKRKN);
add("KNKR", &EvaluateKNKR);
add("KQKR", &EvaluateKQKR);
add("KRKQ", &EvaluateKRKQ);
add("KBBKN", &EvaluateKBBKN);
add("KNKBB", &EvaluateKNKBB);
add<ScalingFunction<KNPK> >("KNPK");
add<ScalingFunction<KRPKR> >("KRPKR");
add<ScalingFunction<KBPKB> >("KBPKB");
add<ScalingFunction<KBPPKB> >("KBPPKB");
add<ScalingFunction<KBPKN> >("KBPKN");
add<ScalingFunction<KRPPKRP> >("KRPPKRP");
}
add("KNPK", WHITE, &ScaleKNPK);
add("KKNP", BLACK, &ScaleKKNP);
add("KRPKR", WHITE, &ScaleKRPKR);
add("KRKRP", BLACK, &ScaleKRKRP);
add("KBPKB", WHITE, &ScaleKBPKB);
add("KBKBP", BLACK, &ScaleKBKBP);
add("KBPPKB", WHITE, &ScaleKBPPKB);
add("KBKBPP", BLACK, &ScaleKBKBPP);
add("KBPKN", WHITE, &ScaleKBPKN);
add("KNKBP", BLACK, &ScaleKNKBP);
add("KRPPKRP", WHITE, &ScaleKRPPKRP);
add("KRPKRPP", BLACK, &ScaleKRPKRPP);
add("KRPPKRP", WHITE, &ScaleKRPPKRP);
add("KRPKRPP", BLACK, &ScaleKRPKRPP);
EndgameFunctions::~EndgameFunctions() {
for (map<Key, EF*>::iterator it = maps.first.begin(); it != maps.first.end(); ++it)
delete (*it).second;
for (map<Key, SF*>::iterator it = maps.second.begin(); it != maps.second.end(); ++it)
delete (*it).second;
}
Key EndgameFunctions::buildKey(const string& keyCode) {
@@ -348,11 +390,11 @@ Key EndgameFunctions::buildKey(const string& keyCode) {
assert(keyCode.length() > 0 && keyCode[0] == 'K');
assert(keyCode.length() < 8);
std::stringstream s;
stringstream s;
bool upcase = false;
// Build up a fen substring with the given pieces, note
// that the fen string could be of an illegal position.
// Build up a fen string with the given pieces, note that
// the fen string could be of an illegal position.
for (size_t i = 0; i < keyCode.length(); i++)
{
if (keyCode[i] == 'K')
@@ -364,29 +406,25 @@ Key EndgameFunctions::buildKey(const string& keyCode) {
return Position(s.str()).get_material_key();
}
void EndgameFunctions::add(const string& keyCode, EndgameEvaluationFunctionBase* f) {
const string EndgameFunctions::swapColors(const string& keyCode) {
EEFmap.insert(std::pair<Key, EndgameEvaluationFunctionBase*>(buildKey(keyCode), f));
// Build corresponding key for the opposite color: "KBPKN" -> "KNKBP"
size_t idx = keyCode.find("K", 1);
return keyCode.substr(idx) + keyCode.substr(0, idx);
}
void EndgameFunctions::add(const string& keyCode, Color c, EndgameScalingFunctionBase* f) {
template<class T>
void EndgameFunctions::add(const string& keyCode) {
ScalingInfo s = {c, f};
ESFmap.insert(std::pair<Key, ScalingInfo>(buildKey(keyCode), s));
typedef typename T::Base F;
get<F>().insert(pair<Key, F*>(buildKey(keyCode), new T(WHITE)));
get<F>().insert(pair<Key, F*>(buildKey(swapColors(keyCode)), new T(BLACK)));
}
EndgameEvaluationFunctionBase* EndgameFunctions::getEEF(Key key) const {
template<class T>
T* EndgameFunctions::get(Key key) const {
std::map<Key, EndgameEvaluationFunctionBase*>::const_iterator it(EEFmap.find(key));
return (it != EEFmap.end() ? it->second : NULL);
}
EndgameScalingFunctionBase* EndgameFunctions::getESF(Key key, Color* c) const {
std::map<Key, ScalingInfo>::const_iterator it(ESFmap.find(key));
if (it == ESFmap.end())
return NULL;
*c = it->second.col;
return it->second.fun;
typename map<Key, T*>::const_iterator it(get<T>().find(key));
return (it != get<T>().end() ? it->second : NULL);
}
+21 -23
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -51,10 +51,10 @@ class MaterialInfo {
public:
MaterialInfo() : key(0) { clear(); }
Value mg_value() const;
Value eg_value() const;
Score material_value() const;
ScaleFactor scale_factor(const Position& pos, Color c) const;
int space_weight() const;
Phase game_phase() const;
bool specialized_eval_exists() const;
Value evaluate(const Position& pos) const;
@@ -62,27 +62,19 @@ private:
inline void clear();
Key key;
int16_t mgValue;
int16_t egValue;
int16_t value;
uint8_t factor[2];
EndgameEvaluationFunctionBase* evaluationFunction;
EndgameScalingFunctionBase* scalingFunction[2];
int spaceWeight;
Phase gamePhase;
};
/// EndgameFunctions class stores the endgame evaluation functions std::map.
/// Because STL library is not thread safe even for read access, the maps,
/// although with identical content, are replicated for each thread. This
/// is faster then using locks with an unique set of global maps.
class EndgameFunctions;
/// The MaterialInfoTable class represents a pawn hash table. It is basically
/// just an array of MaterialInfo objects and a few methods for accessing these
/// objects. The most important method is get_material_info, which looks up a
/// position in the table and returns a pointer to a MaterialInfo object.
class EndgameFunctions;
class MaterialInfoTable {
@@ -91,6 +83,8 @@ public:
~MaterialInfoTable();
MaterialInfo* get_material_info(const Position& pos);
static Phase game_phase(const Position& pos);
private:
unsigned size;
MaterialInfo* entries;
@@ -102,17 +96,13 @@ private:
//// Inline functions
////
/// MaterialInfo::mg_value and MaterialInfo::eg_value simply returns the
/// material balance evaluation for the middle game and the endgame.
inline Value MaterialInfo::mg_value() const {
/// MaterialInfo::material_value simply returns the material balance
/// evaluation that is independent from game phase.
return Value(mgValue);
}
inline Score MaterialInfo::material_value() const {
inline Value MaterialInfo::eg_value() const {
return Value(egValue);
return make_score(value, value);
}
@@ -121,7 +111,7 @@ inline Value MaterialInfo::eg_value() const {
inline void MaterialInfo::clear() {
mgValue = egValue = 0;
value = 0;
factor[WHITE] = factor[BLACK] = uint8_t(SCALE_FACTOR_NORMAL);
evaluationFunction = NULL;
scalingFunction[WHITE] = scalingFunction[BLACK] = NULL;
@@ -156,6 +146,14 @@ inline int MaterialInfo::space_weight() const {
return spaceWeight;
}
/// MaterialInfo::game_phase() returns the game phase according
/// to this material configuration.
inline Phase MaterialInfo::game_phase() const {
return gamePhase;
}
/// MaterialInfo::specialized_eval_exists decides whether there is a
/// specialized evaluation function for the current material configuration,
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+77 -35
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -27,40 +27,21 @@
# include <sys/time.h>
# include <sys/types.h>
# include <unistd.h>
# if defined(__hpux)
# include <sys/pstat.h>
# endif
#else
/*
(c) Copyright 1992 Eric Backus
This software may be used freely so long as this copyright notice is
left intact. There is no warrantee on this software.
*/
# include <windows.h>
# include <time.h>
# include "dos.h"
static int gettimeofday(struct timeval* tp, struct timezone*)
{
SYSTEMTIME systime;
if (tp)
{
struct tm tmrec;
time_t theTime = time(NULL);
tmrec = *localtime(&theTime);
tp->tv_sec = mktime(&tmrec);
GetLocalTime(&systime); /* system time */
tp->tv_usec = systime.wMilliseconds * 1000;
}
return 0;
}
#define _CRT_SECURE_NO_DEPRECATE
#include <windows.h>
#include <sys/timeb.h>
#endif
#include <cassert>
#include <cstdio>
#include <cstring>
#include <iomanip>
#include <iostream>
#include <sstream>
@@ -73,7 +54,7 @@ using namespace std;
/// Version number. If this is left empty, the current date (in the format
/// YYMMDD) is used as a version number.
static const string EngineVersion = "1.4";
static const string EngineVersion = "1.7";
static const string AppName = "Stockfish";
static const string AppTag = "";
@@ -189,29 +170,45 @@ const string engine_name() {
/// milliseconds.
int get_system_time() {
struct timeval t;
gettimeofday(&t, NULL);
return t.tv_sec*1000 + t.tv_usec/1000;
#if defined(_MSC_VER)
struct _timeb t;
_ftime(&t);
return int(t.time*1000 + t.millitm);
#else
struct timeval t;
gettimeofday(&t, NULL);
return t.tv_sec*1000 + t.tv_usec/1000;
#endif
}
/// cpu_count() tries to detect the number of CPU cores.
/// builtin_cpu_count() tries to detect the number of CPU cores, if
/// hyper-threading is enabled this is the number of logical processors.
#if !defined(_MSC_VER)
# if defined(_SC_NPROCESSORS_ONLN)
int cpu_count() {
static int builtin_cpu_count() {
return Min(sysconf(_SC_NPROCESSORS_ONLN), 8);
}
# elif defined(__hpux)
static int builtin_cpu_count() {
struct pst_dynamic psd;
if (pstat_getdynamic(&psd, sizeof(psd), (size_t)1, 0) == -1)
return 1;
return Min(psd.psd_proc_cnt, 8);
}
# else
int cpu_count() {
static int builtin_cpu_count() {
return 1;
}
# endif
#else
int cpu_count() {
static int builtin_cpu_count() {
SYSTEM_INFO s;
GetSystemInfo(&s);
return Min(s.dwNumberOfProcessors, 8);
@@ -220,6 +217,51 @@ int cpu_count() {
#endif
/// HT_enabled() returns true if hyper-threading is enabled on current machine
static bool HT_enabled() {
char CPUString[0x20];
int CPUInfo[4] = {-1};
int nIds, nLogicalCPU, nCores;
// Detect CPU producer
__cpuid(CPUInfo, 0);
nIds = CPUInfo[0];
memset(CPUString, 0, sizeof(CPUString));
memcpy(&CPUString[0], &CPUInfo[1], sizeof(int));
memcpy(&CPUString[4], &CPUInfo[3], sizeof(int));
memcpy(&CPUString[8], &CPUInfo[2], sizeof(int));
// Not an Intel CPU or CPUID.4 not supported
if (strcmp(CPUString, "GenuineIntel") || nIds < 4)
return false;
// Detect if HT Technology is supported
__cpuid(CPUInfo, 1);
if (!((CPUInfo[3] >> 28) & 1))
return false;
nLogicalCPU = (CPUInfo[1] >> 16) & 0xFF;
// Detect number of cores
__cpuid(CPUInfo, 4);
nCores = 1 + ((CPUInfo[0] >> 26) & 0x3F);
return nLogicalCPU > nCores;
}
/// cpu_count() tries to detect the number of physical CPU cores taking
/// in account hyper-threading.
int cpu_count() {
return HT_enabled() ? builtin_cpu_count() / 2 : builtin_cpu_count();
}
/*
From Beowulf, from Olithink
*/
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+80 -2
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -62,9 +62,83 @@ struct MoveStack {
int score;
};
// Note that operator< is set up such that std::sort() will sort in descending order
// Note that operator< is set up such that sorting will be in descending order
inline bool operator<(const MoveStack& f, const MoveStack& s) { return s.score < f.score; }
// An helper insertion sort implementation
template<typename T>
inline void insertion_sort(T* firstMove, T* lastMove)
{
T value;
T *cur, *p, *d;
if (firstMove != lastMove)
for (cur = firstMove + 1; cur != lastMove; cur++)
{
p = d = cur;
value = *p--;
if (value < *p)
{
do *d = *p;
while (--d != firstMove && value < *--p);
*d = value;
}
}
}
// Our dedicated sort in range [firstMove, lastMove), first splits
// positive scores from ramining then order seaprately the two sets.
template<typename T>
inline void sort_moves(T* firstMove, T* lastMove, T** lastPositive)
{
T tmp;
T *p, *d;
d = lastMove;
p = firstMove - 1;
d->score = -1; // right guard
// Split positives vs non-positives
do {
while ((++p)->score > 0);
if (p != d)
{
while (--d != p && d->score <= 0);
tmp = *p;
*p = *d;
*d = tmp;
}
} while (p != d);
// Sort just positive scored moves, remaining only when we get there
insertion_sort<T>(firstMove, p);
*lastPositive = p;
}
// Picks up the best move in range [curMove, lastMove), one per cycle.
// It is faster then sorting all the moves in advance when moves are few,
// as normally are the possible captures. Note that is not a stable alghoritm.
template<typename T>
inline T pick_best(T* curMove, T* lastMove)
{
T bestMove, tmp;
bestMove = *curMove;
while (++curMove != lastMove)
{
if (*curMove < bestMove)
{
tmp = *curMove;
*curMove = bestMove;
bestMove = tmp;
}
}
return bestMove;
}
////
//// Inline functions
@@ -82,6 +156,10 @@ inline PieceType move_promotion_piece(Move m) {
return PieceType((int(m) >> 12) & 7);
}
inline int move_is_special(Move m) {
return m & (0x1F << 12);
}
inline int move_is_promotion(Move m) {
return m & (7 << 12);
}
+327 -600
View File
File diff suppressed because it is too large Load Diff
+7 -6
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -32,12 +32,13 @@
//// Prototypes
////
extern int generate_captures(const Position& pos, MoveStack* mlist);
extern int generate_noncaptures(const Position& pos, MoveStack* mlist);
extern int generate_non_capture_checks(const Position& pos, MoveStack* mlist, Bitboard dc);
extern int generate_evasions(const Position& pos, MoveStack* mlist, Bitboard pinned);
extern int generate_legal_moves(const Position& pos, MoveStack* mlist);
extern MoveStack* generate_captures(const Position& pos, MoveStack* mlist);
extern MoveStack* generate_noncaptures(const Position& pos, MoveStack* mlist);
extern MoveStack* generate_non_capture_checks(const Position& pos, MoveStack* mlist);
extern MoveStack* generate_evasions(const Position& pos, MoveStack* mlist);
extern MoveStack* generate_moves(const Position& pos, MoveStack* mlist, bool pseudoLegal = false);
extern bool move_is_legal(const Position& pos, const Move m, Bitboard pinned);
extern bool move_is_legal(const Position& pos, const Move m);
#endif // !defined(MOVEGEN_H_INCLUDED)
+238 -292
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -23,11 +23,9 @@
//// Includes
////
#include <algorithm>
#include <cassert>
#include "history.h"
#include "evaluate.h"
#include "movegen.h"
#include "movepick.h"
#include "search.h"
@@ -40,14 +38,23 @@
namespace {
/// Variables
MovePicker::MovegenPhase PhaseTable[32];
int MainSearchPhaseIndex;
int EvasionsPhaseIndex;
int QsearchWithChecksPhaseIndex;
int QsearchWithoutChecksPhaseIndex;
enum MovegenPhase {
PH_TT_MOVES, // Transposition table move and mate killer
PH_GOOD_CAPTURES, // Queen promotions and captures with SEE values >= 0
PH_KILLERS, // Killer moves from the current ply
PH_NONCAPTURES, // Non-captures and underpromotions
PH_BAD_CAPTURES, // Queen promotions and captures with SEE values < 0
PH_EVASIONS, // Check evasions
PH_QCAPTURES, // Captures in quiescence search
PH_QCHECKS, // Non-capture checks in quiescence search
PH_STOP
};
CACHE_LINE_ALIGNMENT
const uint8_t MainSearchPhaseTable[] = { PH_TT_MOVES, PH_GOOD_CAPTURES, PH_KILLERS, PH_NONCAPTURES, PH_BAD_CAPTURES, PH_STOP};
const uint8_t EvasionsPhaseTable[] = { PH_TT_MOVES, PH_EVASIONS, PH_STOP};
const uint8_t QsearchWithChecksPhaseTable[] = { PH_TT_MOVES, PH_QCAPTURES, PH_QCHECKS, PH_STOP};
const uint8_t QsearchWithoutChecksPhaseTable[] = { PH_TT_MOVES, PH_QCAPTURES, PH_STOP};
}
@@ -56,7 +63,7 @@ namespace {
////
/// Constructor for the MovePicker class. Apart from the position for which
/// Constructor for the MovePicker class. Apart from the position for which
/// it is asked to pick legal moves, MovePicker also wants some information
/// to help it to return the presumably good moves first, to decide which
/// moves to return (in the quiescence search, for instance, we only want to
@@ -64,203 +71,145 @@ namespace {
/// move ordering is at the current node.
MovePicker::MovePicker(const Position& p, Move ttm, Depth d,
const History& h, SearchStack* ss) : pos(p), H(h) {
ttMove = ttm;
if (ss)
{
mateKiller = (ss->mateKiller == ttm)? MOVE_NONE : ss->mateKiller;
killer1 = ss->killers[0];
killer2 = ss->killers[1];
} else
mateKiller = killer1 = killer2 = MOVE_NONE;
const History& h, SearchStack* ss, Value beta) : pos(p), H(h) {
int searchTT = ttm;
ttMoves[0].move = ttm;
lastBadCapture = badCaptures;
badCaptureThreshold = 0;
movesPicked = numOfMoves = numOfBadCaptures = 0;
checkKillers = checkLegal = finished = false;
pinned = p.pinned_pieces(pos.side_to_move());
if (ss && !p.is_check())
{
ttMoves[1].move = (ss->mateKiller == ttm)? MOVE_NONE : ss->mateKiller;
searchTT |= ttMoves[1].move;
killers[0].move = ss->killers[0];
killers[1].move = ss->killers[1];
} else
ttMoves[1].move = killers[0].move = killers[1].move = MOVE_NONE;
if (p.is_check())
phaseIndex = EvasionsPhaseIndex;
phasePtr = EvasionsPhaseTable;
else if (d > Depth(0))
phaseIndex = MainSearchPhaseIndex;
else if (d == Depth(0))
phaseIndex = QsearchWithChecksPhaseIndex;
{
// Consider sligtly negative captures as good if at low
// depth and far from beta.
if (ss && ss->eval < beta - PawnValueMidgame && d < 3 * OnePly)
badCaptureThreshold = -PawnValueMidgame;
phasePtr = MainSearchPhaseTable;
} else if (d == Depth(0))
phasePtr = QsearchWithChecksPhaseTable;
else
phaseIndex = QsearchWithoutChecksPhaseIndex;
{
phasePtr = QsearchWithoutChecksPhaseTable;
Color us = pos.side_to_move();
// Skip TT move if is not a capture or a promotion, this avoids
// qsearch tree explosion due to a possible perpetual check or
// similar rare cases when TT table is full.
if (ttm != MOVE_NONE && !pos.move_is_capture_or_promotion(ttm))
searchTT = ttMoves[0].move = MOVE_NONE;
}
dc = p.discovered_check_candidates(us);
pinned = p.pinned_pieces(us);
finished = false;
phasePtr += !searchTT - 1;
go_next_phase();
}
/// MovePicker::get_next_move() is the most important method of the MovePicker
/// class. It returns a new legal move every time it is called, until there
/// are no more moves left of the types we are interested in.
/// MovePicker::go_next_phase() generates, scores and sorts the next bunch
/// of moves when there are no more moves to try for the current phase.
Move MovePicker::get_next_move() {
void MovePicker::go_next_phase() {
Move move;
curMove = moves;
phase = *(++phasePtr);
switch (phase) {
while (true)
{
// If we already have a list of generated moves, pick the best move from
// the list, and return it.
move = pick_move_from_list();
if (move != MOVE_NONE)
{
assert(move_is_ok(move));
return move;
}
case PH_TT_MOVES:
curMove = ttMoves;
lastMove = curMove + 2;
return;
// Next phase
phaseIndex++;
switch (PhaseTable[phaseIndex]) {
case PH_GOOD_CAPTURES:
lastMove = generate_captures(pos, moves);
score_captures();
return;
case PH_TT_MOVE:
if (ttMove != MOVE_NONE)
{
assert(move_is_ok(ttMove));
if (move_is_legal(pos, ttMove, pinned))
return ttMove;
}
break;
case PH_KILLERS:
curMove = killers;
lastMove = curMove + 2;
return;
case PH_MATE_KILLER:
if (mateKiller != MOVE_NONE)
{
assert(move_is_ok(mateKiller));
if (move_is_legal(pos, mateKiller, pinned))
return mateKiller;
}
break;
case PH_NONCAPTURES:
lastMove = generate_noncaptures(pos, moves);
score_noncaptures();
sort_moves(moves, lastMove, &lastGoodNonCapture);
return;
case PH_GOOD_CAPTURES:
numOfMoves = generate_captures(pos, moves);
score_captures();
std::sort(moves, moves + numOfMoves);
movesPicked = 0;
checkLegal = true;
break;
case PH_BAD_CAPTURES:
// Bad captures SEE value is already calculated so just sort them
// to get SEE move ordering.
curMove = badCaptures;
lastMove = lastBadCapture;
return;
case PH_KILLERS:
movesPicked = numOfMoves = 0;
checkLegal = false;
if (killer1 != MOVE_NONE && move_is_legal(pos, killer1, pinned) && !pos.move_is_capture(killer1))
moves[numOfMoves++].move = killer1;
if (killer2 != MOVE_NONE && move_is_legal(pos, killer2, pinned) && !pos.move_is_capture(killer2) )
moves[numOfMoves++].move = killer2;
break;
case PH_EVASIONS:
assert(pos.is_check());
lastMove = generate_evasions(pos, moves);
score_evasions_or_checks();
return;
case PH_NONCAPTURES:
checkKillers = (numOfMoves != 0); // previous phase is PH_KILLERS
numOfMoves = generate_noncaptures(pos, moves);
score_noncaptures();
std::sort(moves, moves + numOfMoves);
movesPicked = 0;
checkLegal = true;
break;
case PH_QCAPTURES:
lastMove = generate_captures(pos, moves);
score_captures();
return;
case PH_BAD_CAPTURES:
// Bad captures SEE value is already calculated by score_captures()
// so just sort them to get SEE move ordering.
std::sort(badCaptures, badCaptures + numOfBadCaptures);
movesPicked = 0;
break;
case PH_QCHECKS:
lastMove = generate_non_capture_checks(pos, moves);
score_evasions_or_checks();
return;
case PH_EVASIONS:
assert(pos.is_check());
numOfMoves = generate_evasions(pos, moves, pinned);
score_evasions();
std::sort(moves, moves + numOfMoves);
movesPicked = 0;
break;
case PH_STOP:
lastMove = curMove + 1; // Avoids another go_next_phase() call
return;
case PH_QCAPTURES:
numOfMoves = generate_captures(pos, moves);
score_qcaptures();
std::sort(moves, moves + numOfMoves);
movesPicked = 0;
break;
case PH_QCHECKS:
// Perhaps we should order moves move here? FIXME
numOfMoves = generate_non_capture_checks(pos, moves, dc);
movesPicked = 0;
break;
case PH_STOP:
return MOVE_NONE;
default:
assert(false);
return MOVE_NONE;
}
default:
assert(false);
return;
}
}
/// A variant of get_next_move() which takes a lock as a parameter, used to
/// prevent multiple threads from picking the same move at a split point.
Move MovePicker::get_next_move(Lock &lock) {
lock_grab(&lock);
if (finished)
{
lock_release(&lock);
return MOVE_NONE;
}
Move m = get_next_move();
if (m == MOVE_NONE)
finished = true;
lock_release(&lock);
return m;
}
/// MovePicker::score_captures(), MovePicker::score_noncaptures(),
/// MovePicker::score_evasions() and MovePicker::score_qcaptures() assign a
/// numerical move ordering score to each move in a move list. The moves
/// with highest scores will be picked first by pick_move_from_list().
/// MovePicker::score_captures(), MovePicker::score_noncaptures() and
/// MovePicker::score_evasions() assign a numerical move ordering score
/// to each move in a move list. The moves with highest scores will be
/// picked first by get_next_move().
void MovePicker::score_captures() {
// Winning and equal captures in the main search are ordered by MVV/LVA.
// Suprisingly, this appears to perform slightly better than SEE based
// move ordering. The reason is probably that in a position with a winning
// move ordering. The reason is probably that in a position with a winning
// capture, capturing a more valuable (but sufficiently defended) piece
// first usually doesn't hurt. The opponent will have to recapture, and
// first usually doesn't hurt. The opponent will have to recapture, and
// the hanging piece will still be hanging (except in the unusual cases
// where it is possible to recapture with the hanging piece). Exchanging
// big pieces before capturing a hanging piece probably helps to reduce
// the subtree size.
// While scoring captures it moves all captures with negative SEE values
// to the badCaptures[] array.
// In main search we want to push captures with negative SEE values to
// badCaptures[] array, but instead of doing it now we delay till when
// the move has been picked up in pick_move_from_list(), this way we save
// some SEE calls in case we get a cutoff (idea from Pablo Vazquez).
Move m;
int seeValue;
for (int i = 0; i < numOfMoves; i++)
// Use MVV/LVA ordering
for (MoveStack* cur = moves; cur != lastMove; cur++)
{
m = moves[i].move;
seeValue = pos.see(m);
if (seeValue >= 0)
{
if (move_is_promotion(m))
moves[i].score = QueenValueMidgame;
else
moves[i].score = int(pos.midgame_value_of_piece_on(move_to(m)))
-int(pos.type_of_piece_on(move_from(m)));
}
m = cur->move;
if (move_is_promotion(m))
cur->score = QueenValueMidgame;
else
{
// Losing capture, move it to the badCaptures[] array
assert(numOfBadCaptures < 63);
moves[i].score = seeValue;
badCaptures[numOfBadCaptures++] = moves[i];
moves[i--] = moves[--numOfMoves];
}
cur->score = pos.midgame_value_of_piece_on(move_to(m))
- pos.type_of_piece_on(move_from(m));
}
}
@@ -268,154 +217,151 @@ void MovePicker::score_noncaptures() {
// First score by history, when no history is available then use
// piece/square tables values. This seems to be better then a
// random choice when we don't have an history for any move.
Move m;
Piece piece;
Square from, to;
int hs;
for (int i = 0; i < numOfMoves; i++)
for (MoveStack* cur = moves; cur != lastMove; cur++)
{
from = move_from(moves[i].move);
to = move_to(moves[i].move);
m = cur->move;
from = move_from(m);
to = move_to(m);
piece = pos.piece_on(from);
hs = H.move_ordering_score(piece, to);
// Ensure history is always preferred to pst
// Ensure history has always highest priority
if (hs > 0)
hs += 1000;
hs += 10000;
// pst based scoring
moves[i].score = hs + pos.pst_delta<Position::MidGame>(piece, from, to);
// Gain table based scoring
cur->score = hs + 16 * H.gain(piece, to);
}
}
void MovePicker::score_evasions() {
void MovePicker::score_evasions_or_checks() {
// Try good captures ordered by MVV/LVA, then non-captures if
// destination square is not under attack, ordered by history
// value, and at the end bad-captures and non-captures with a
// negative SEE. This last group is ordered by the SEE score.
Move m;
int seeScore;
for (int i = 0; i < numOfMoves; i++)
// Skip if we don't have at least two moves to order
if (lastMove < moves + 2)
return;
for (MoveStack* cur = moves; cur != lastMove; cur++)
{
Move m = moves[i].move;
if (m == ttMove)
moves[i].score = 2*HistoryMax;
else if (!pos.square_is_empty(move_to(m)))
{
int seeScore = pos.see(m);
moves[i].score = (seeScore >= 0)? seeScore + HistoryMax : seeScore;
} else
moves[i].score = H.move_ordering_score(pos.piece_on(move_from(m)), move_to(m));
}
}
void MovePicker::score_qcaptures() {
// Use MVV/LVA ordering
for (int i = 0; i < numOfMoves; i++)
{
Move m = moves[i].move;
if (move_is_promotion(m))
moves[i].score = QueenValueMidgame;
m = cur->move;
if ((seeScore = pos.see_sign(m)) < 0)
cur->score = seeScore - HistoryMax; // Be sure are at the bottom
else if (pos.move_is_capture(m))
cur->score = pos.midgame_value_of_piece_on(move_to(m))
- pos.type_of_piece_on(move_from(m)) + HistoryMax;
else
moves[i].score = int(pos.midgame_value_of_piece_on(move_to(m)))
-int(pos.type_of_piece_on(move_from(m)));
cur->score = H.move_ordering_score(pos.piece_on(move_from(m)), move_to(m));
}
}
/// MovePicker::get_next_move() is the most important method of the MovePicker
/// class. It returns a new legal move every time it is called, until there
/// are no more moves left.
/// It picks the move with the biggest score from a list of generated moves taking
/// care not to return the tt move if has already been searched previously.
/// Note that this function is not thread safe so should be lock protected by
/// caller when accessed through a shared MovePicker object.
/// MovePicker::pick_move_from_list() picks the move with the biggest score
/// from a list of generated moves (moves[] or badCaptures[], depending on
/// the current move generation phase). It takes care not to return the
/// transposition table move if that has already been serched previously.
Move MovePicker::get_next_move() {
Move MovePicker::pick_move_from_list() {
Move move;
assert(movesPicked >= 0);
assert(!pos.is_check() || PhaseTable[phaseIndex] == PH_EVASIONS || PhaseTable[phaseIndex] == PH_STOP);
assert( pos.is_check() || PhaseTable[phaseIndex] != PH_EVASIONS);
switch (PhaseTable[phaseIndex]) {
case PH_GOOD_CAPTURES:
case PH_KILLERS:
case PH_NONCAPTURES:
while (movesPicked < numOfMoves)
while (true)
{
while (curMove != lastMove)
{
Move move = moves[movesPicked++].move;
if ( move != ttMove
&& move != mateKiller
&& (!checkKillers || (move != killer1 && move != killer2))
&& (!checkLegal || pos.pl_move_is_legal(move, pinned)))
switch (phase) {
case PH_TT_MOVES:
move = (curMove++)->move;
if ( move != MOVE_NONE
&& move_is_legal(pos, move, pinned))
return move;
break;
case PH_GOOD_CAPTURES:
move = pick_best(curMove++, lastMove).move;
if ( move != ttMoves[0].move
&& move != ttMoves[1].move
&& pos.pl_move_is_legal(move, pinned))
{
// Check for a non negative SEE now
int seeValue = pos.see_sign(move);
if (seeValue >= badCaptureThreshold)
return move;
// Losing capture, move it to the badCaptures[] array, note
// that move has now been already checked for legality.
assert(int(lastBadCapture - badCaptures) < 63);
lastBadCapture->move = move;
lastBadCapture->score = seeValue;
lastBadCapture++;
}
break;
case PH_KILLERS:
move = (curMove++)->move;
if ( move != MOVE_NONE
&& move != ttMoves[0].move
&& move != ttMoves[1].move
&& move_is_legal(pos, move, pinned)
&& !pos.move_is_capture(move))
return move;
break;
case PH_NONCAPTURES:
// Sort negative scored moves only when we get there
if (curMove == lastGoodNonCapture)
insertion_sort(lastGoodNonCapture, lastMove);
move = (curMove++)->move;
if ( move != ttMoves[0].move
&& move != ttMoves[1].move
&& move != killers[0].move
&& move != killers[1].move
&& pos.pl_move_is_legal(move, pinned))
return move;
break;
case PH_BAD_CAPTURES:
move = pick_best(curMove++, lastMove).move;
return move;
case PH_EVASIONS:
case PH_QCAPTURES:
move = pick_best(curMove++, lastMove).move;
if ( move != ttMoves[0].move
&& pos.pl_move_is_legal(move, pinned))
return move;
break;
case PH_QCHECKS:
move = (curMove++)->move;
if ( move != ttMoves[0].move
&& pos.pl_move_is_legal(move, pinned))
return move;
break;
case PH_STOP:
return MOVE_NONE;
default:
assert(false);
break;
}
}
break;
case PH_EVASIONS:
if (movesPicked < numOfMoves)
return moves[movesPicked++].move;
break;
case PH_BAD_CAPTURES:
while (movesPicked < numOfBadCaptures)
{
Move move = badCaptures[movesPicked++].move;
if ( move != ttMove
&& move != mateKiller
&& pos.pl_move_is_legal(move, pinned))
return move;
}
break;
case PH_QCAPTURES:
case PH_QCHECKS:
while (movesPicked < numOfMoves)
{
Move move = moves[movesPicked++].move;
// Maybe postpone the legality check until after futility pruning?
if ( move != ttMove
&& pos.pl_move_is_legal(move, pinned))
return move;
}
break;
default:
break;
go_next_phase();
}
return MOVE_NONE;
}
/// MovePicker::init_phase_table() initializes the PhaseTable[],
/// MainSearchPhaseIndex, EvasionPhaseIndex, QsearchWithChecksPhaseIndex
/// and QsearchWithoutChecksPhaseIndex. It is only called once during
/// program startup, and never again while the program is running.
void MovePicker::init_phase_table() {
int i = 0;
// Main search
MainSearchPhaseIndex = i - 1;
PhaseTable[i++] = PH_TT_MOVE;
PhaseTable[i++] = PH_MATE_KILLER;
PhaseTable[i++] = PH_GOOD_CAPTURES;
PhaseTable[i++] = PH_KILLERS;
PhaseTable[i++] = PH_NONCAPTURES;
PhaseTable[i++] = PH_BAD_CAPTURES;
PhaseTable[i++] = PH_STOP;
// Check evasions
EvasionsPhaseIndex = i - 1;
PhaseTable[i++] = PH_EVASIONS;
PhaseTable[i++] = PH_STOP;
// Quiescence search with checks
QsearchWithChecksPhaseIndex = i - 1;
PhaseTable[i++] = PH_TT_MOVE;
PhaseTable[i++] = PH_QCAPTURES;
PhaseTable[i++] = PH_QCHECKS;
PhaseTable[i++] = PH_STOP;
// Quiescence search without checks
QsearchWithoutChecksPhaseIndex = i - 1;
PhaseTable[i++] = PH_TT_MOVE;
PhaseTable[i++] = PH_QCAPTURES;
PhaseTable[i++] = PH_STOP;
}
+17 -47
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -27,7 +27,6 @@
#include "depth.h"
#include "history.h"
#include "lock.h"
#include "position.h"
@@ -35,7 +34,6 @@
//// Types
////
struct EvalInfo;
struct SearchStack;
/// MovePicker is a class which is used to pick one legal move at a time from
@@ -51,45 +49,24 @@ class MovePicker {
MovePicker& operator=(const MovePicker&); // silence a warning under MSVC
public:
enum MovegenPhase {
PH_TT_MOVE, // Transposition table move
PH_MATE_KILLER, // Mate killer from the current ply
PH_GOOD_CAPTURES, // Queen promotions and captures with SEE values >= 0
PH_KILLERS, // Killer moves from the current ply
PH_NONCAPTURES, // Non-captures and underpromotions
PH_BAD_CAPTURES, // Queen promotions and captures with SEE values < 0
PH_EVASIONS, // Check evasions
PH_QCAPTURES, // Captures in quiescence search
PH_QCHECKS, // Non-capture checks in quiescence search
PH_STOP
};
MovePicker(const Position& p, Move ttm, Depth d, const History& h, SearchStack* ss = NULL);
MovePicker(const Position& p, Move ttm, Depth d, const History& h, SearchStack* ss = NULL, Value beta = -VALUE_INFINITE);
Move get_next_move();
Move get_next_move(Lock& lock);
int number_of_moves() const;
Bitboard discovered_check_candidates() const;
static void init_phase_table();
int number_of_evasions() const;
private:
void score_captures();
void score_noncaptures();
void score_evasions();
void score_qcaptures();
Move pick_move_from_list();
void score_evasions_or_checks();
void go_next_phase();
const Position& pos;
const History& H;
Move ttMove, mateKiller, killer1, killer2;
Bitboard pinned, dc;
MoveStack ttMoves[2], killers[2];
int badCaptureThreshold, phase;
const uint8_t* phasePtr;
MoveStack *curMove, *lastMove, *lastGoodNonCapture, *lastBadCapture;
Bitboard pinned;
MoveStack moves[256], badCaptures[64];
int phaseIndex;
int numOfMoves, numOfBadCaptures;
int movesPicked;
bool checkKillers, checkLegal;
bool finished;
};
@@ -97,21 +74,14 @@ private:
//// Inline functions
////
/// MovePicker::number_of_moves() simply returns the numOfMoves member
/// variable. It is intended to be used in positions where the side to move
/// is in check, for detecting checkmates or situations where there is only
/// a single reply to check.
/// MovePicker::number_of_evasions() simply returns the number of moves in
/// evasions phase. It is intended to be used in positions where the side to
/// move is in check, for detecting checkmates or situations where there is
/// only a single reply to check.
/// WARNING: It works as long as PH_EVASIONS is the _only_ phase for evasions.
inline int MovePicker::number_of_moves() const {
return numOfMoves;
}
/// MovePicker::discovered_check_candidates() returns a bitboard containing
/// all pieces which can possibly give discovered check. This bitboard is
/// computed by the constructor function.
inline Bitboard MovePicker::discovered_check_candidates() const {
return dc;
inline int MovePicker::number_of_evasions() const {
return int(lastMove - moves);
}
#endif // !defined(MOVEPICK_H_INCLUDED)
+231 -260
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -23,6 +23,7 @@
////
#include <cassert>
#include <cstring>
#include "bitcount.h"
#include "pawns.h"
@@ -37,67 +38,39 @@ namespace {
/// Constants and variables
// Doubled pawn penalty by file, middle game.
const Value DoubledPawnMidgamePenalty[8] = {
Value(13), Value(20), Value(23), Value(23),
Value(23), Value(23), Value(20), Value(13)
#define S(mg, eg) make_score(mg, eg)
// Doubled pawn penalty by file
const Score DoubledPawnPenalty[8] = {
S(13, 43), S(20, 48), S(23, 48), S(23, 48),
S(23, 48), S(23, 48), S(20, 48), S(13, 43)
};
// Doubled pawn penalty by file, endgame.
const Value DoubledPawnEndgamePenalty[8] = {
Value(43), Value(48), Value(48), Value(48),
Value(48), Value(48), Value(48), Value(43)
// Isolated pawn penalty by file
const Score IsolatedPawnPenalty[8] = {
S(25, 30), S(36, 35), S(40, 35), S(40, 35),
S(40, 35), S(40, 35), S(36, 35), S(25, 30)
};
// Isolated pawn penalty by file, middle game.
const Value IsolatedPawnMidgamePenalty[8] = {
Value(25), Value(36), Value(40), Value(40),
Value(40), Value(40), Value(36), Value(25)
// Backward pawn penalty by file
const Score BackwardPawnPenalty[8] = {
S(20, 28), S(29, 31), S(33, 31), S(33, 31),
S(33, 31), S(33, 31), S(29, 31), S(20, 28)
};
// Isolated pawn penalty by file, endgame.
const Value IsolatedPawnEndgamePenalty[8] = {
Value(30), Value(35), Value(35), Value(35),
Value(35), Value(35), Value(35), Value(30)
// Pawn chain membership bonus by file
const Score ChainBonus[8] = {
S(11,-1), S(13,-1), S(13,-1), S(14,-1),
S(14,-1), S(13,-1), S(13,-1), S(11,-1)
};
// Backward pawn penalty by file, middle game.
const Value BackwardPawnMidgamePenalty[8] = {
Value(20), Value(29), Value(33), Value(33),
Value(33), Value(33), Value(29), Value(20)
// Candidate passed pawn bonus by rank
const Score CandidateBonus[8] = {
S( 0, 0), S( 6, 13), S(6,13), S(14,29),
S(34,68), S(83,166), S(0, 0), S( 0, 0)
};
// Backward pawn penalty by file, endgame.
const Value BackwardPawnEndgamePenalty[8] = {
Value(28), Value(31), Value(31), Value(31),
Value(31), Value(31), Value(31), Value(28)
};
// Pawn chain membership bonus by file, middle game.
const Value ChainMidgameBonus[8] = {
Value(11), Value(13), Value(13), Value(14),
Value(14), Value(13), Value(13), Value(11)
};
// Pawn chain membership bonus by file, endgame.
const Value ChainEndgameBonus[8] = {
Value(-1), Value(-1), Value(-1), Value(-1),
Value(-1), Value(-1), Value(-1), Value(-1)
};
// Candidate passed pawn bonus by rank, middle game.
const Value CandidateMidgameBonus[8] = {
Value( 0), Value( 6), Value(6), Value(14),
Value(34), Value(83), Value(0), Value( 0)
};
// Candidate passed pawn bonus by rank, endgame.
const Value CandidateEndgameBonus[8] = {
Value( 0), Value( 13), Value(13), Value(29),
Value(68), Value(166), Value( 0), Value( 0)
};
// Pawn storm tables for positions with opposite castling:
// Pawn storm tables for positions with opposite castling
const int QStormTable[64] = {
0, 0, 0, 0, 0, 0, 0, 0,
-22,-22,-22,-14,-6, 0, 0, 0,
@@ -140,7 +113,7 @@ PawnInfoTable::PawnInfoTable(unsigned numOfEntries) {
size = numOfEntries;
entries = new PawnInfo[size];
if (entries == NULL)
if (!entries)
{
std::cerr << "Failed to allocate " << (numOfEntries * sizeof(PawnInfo))
<< " bytes for pawn hash table." << std::endl;
@@ -156,22 +129,32 @@ PawnInfoTable::~PawnInfoTable() {
}
/// PawnInfo::clear() resets to zero the PawnInfo entry. Note that
/// kingSquares[] is initialized to SQ_NONE instead.
void PawnInfo::clear() {
memset(this, 0, sizeof(PawnInfo));
kingSquares[WHITE] = kingSquares[BLACK] = SQ_NONE;
}
/// PawnInfoTable::get_pawn_info() takes a position object as input, computes
/// a PawnInfo object, and returns a pointer to it. The result is also
/// stored in a hash table, so we don't have to recompute everything when
/// the same pawn structure occurs again.
PawnInfo *PawnInfoTable::get_pawn_info(const Position &pos) {
PawnInfo* PawnInfoTable::get_pawn_info(const Position& pos) {
assert(pos.is_ok());
Key key = pos.get_pawn_key();
int index = int(key & (size - 1));
PawnInfo *pi = entries + index;
PawnInfo* pi = entries + index;
// If pi->key matches the position's pawn hash key, it means that we
// have analysed this pawn structure before, and we can simply return the
// information we found the last time instead of recomputing it
// have analysed this pawn structure before, and we can simply return
// the information we found the last time instead of recomputing it.
if (pi->key == key)
return pi;
@@ -179,209 +162,197 @@ PawnInfo *PawnInfoTable::get_pawn_info(const Position &pos) {
pi->clear();
pi->key = key;
Value mgValue[2] = {Value(0), Value(0)};
Value egValue[2] = {Value(0), Value(0)};
// Calculate pawn attacks
Bitboard whitePawns = pos.pieces(PAWN, WHITE);
Bitboard blackPawns = pos.pieces(PAWN, BLACK);
pi->pawnAttacks[WHITE] = ((whitePawns << 9) & ~FileABB) | ((whitePawns << 7) & ~FileHBB);
pi->pawnAttacks[BLACK] = ((blackPawns >> 7) & ~FileABB) | ((blackPawns >> 9) & ~FileHBB);
// Loop through the pawns for both colors
for (Color us = WHITE; us <= BLACK; us++)
{
Color them = opposite_color(us);
Bitboard ourPawns = pos.pawns(us);
Bitboard theirPawns = pos.pawns(them);
Bitboard pawns = ourPawns;
int bonus;
// Initialize pawn storm scores by giving bonuses for open files
for (File f = FILE_A; f <= FILE_H; f++)
if (Position::file_is_half_open(ourPawns, f))
{
pi->ksStormValue[us] += KStormOpenFileBonus[f];
pi->qsStormValue[us] += QStormOpenFileBonus[f];
}
// Loop through all pawns of the current color and score each pawn
while (pawns)
{
bool passed, doubled, isolated, backward, chain, candidate;
Square s = pop_1st_bit(&pawns);
File f = square_file(s);
Rank r = square_rank(s);
assert(pos.piece_on(s) == piece_of_color_and_type(us, PAWN));
// The file containing the pawn is not half open
pi->halfOpenFiles[us] &= ~(1 << f);
// Passed, isolated or doubled pawn?
passed = Position::pawn_is_passed(theirPawns, us, s);
isolated = Position::pawn_is_isolated(ourPawns, s);
doubled = Position::pawn_is_doubled(ourPawns, us, s);
// We calculate kingside and queenside pawn storm
// scores for both colors. These are used when evaluating
// middle game positions with opposite side castling.
//
// Each pawn is given a base score given by a piece square table
// (KStormTable[] or QStormTable[]). Pawns which seem to have good
// chances of creating an open file by exchanging itself against an
// enemy pawn on an adjacent file gets an additional bonus.
// Kingside pawn storms
bonus = KStormTable[relative_square(us, s)];
if (f >= FILE_F)
{
Bitboard b = outpost_mask(us, s) & theirPawns & (FileFBB | FileGBB | FileHBB);
while (b)
{
Square s2 = pop_1st_bit(&b);
if (!(theirPawns & neighboring_files_bb(s2) & rank_bb(s2)))
{
// The enemy pawn has no pawn beside itself, which makes it
// particularly vulnerable. Big bonus, especially against a
// weakness on the rook file.
if (square_file(s2) == FILE_H)
bonus += 4*StormLeverBonus[f] - 8*square_distance(s, s2);
else
bonus += 2*StormLeverBonus[f] - 4*square_distance(s, s2);
} else
// There is at least one enemy pawn beside the enemy pawn we look
// at, which means that the pawn has somewhat better chances of
// defending itself by advancing. Smaller bonus.
bonus += StormLeverBonus[f] - 2*square_distance(s, s2);
}
}
pi->ksStormValue[us] += bonus;
// Queenside pawn storms
bonus = QStormTable[relative_square(us, s)];
if (f <= FILE_C)
{
Bitboard b = outpost_mask(us, s) & theirPawns & (FileABB | FileBBB | FileCBB);
while (b)
{
Square s2 = pop_1st_bit(&b);
if (!(theirPawns & neighboring_files_bb(s2) & rank_bb(s2)))
{
// The enemy pawn has no pawn beside itself, which makes it
// particularly vulnerable. Big bonus, especially against a
// weakness on the rook file.
if (square_file(s2) == FILE_A)
bonus += 4*StormLeverBonus[f] - 16*square_distance(s, s2);
else
bonus += 2*StormLeverBonus[f] - 8*square_distance(s, s2);
} else
// There is at least one enemy pawn beside the enemy pawn we look
// at, which means that the pawn has somewhat better chances of
// defending itself by advancing. Smaller bonus.
bonus += StormLeverBonus[f] - 4*square_distance(s, s2);
}
}
pi->qsStormValue[us] += bonus;
// Member of a pawn chain (but not the backward one)? We could speed up
// the test a little by introducing an array of masks indexed by color
// and square for doing the test, but because everything is hashed,
// it probably won't make any noticable difference.
chain = ourPawns
& neighboring_files_bb(f)
& (rank_bb(r) | rank_bb(r - (us == WHITE ? 1 : -1)));
// Test for backward pawn
//
// If the pawn is passed, isolated, or member of a pawn chain
// it cannot be backward. If can capture an enemy pawn or if
// there are friendly pawns behind on neighboring files it cannot
// be backward either.
if ( passed
|| isolated
|| chain
|| (pos.pawn_attacks(us, s) & theirPawns)
|| (ourPawns & behind_bb(us, r) & neighboring_files_bb(f)))
backward = false;
else
{
// We now know that there are no friendly pawns beside or behind this
// pawn on neighboring files. We now check whether the pawn is
// backward by looking in the forward direction on the neighboring
// files, and seeing whether we meet a friendly or an enemy pawn first.
Bitboard b;
if (us == WHITE)
{
for (b = pos.pawn_attacks(us, s); !(b & (ourPawns | theirPawns)); b <<= 8);
backward = (b | (b << 8)) & theirPawns;
}
else
{
for (b = pos.pawn_attacks(us, s); !(b & (ourPawns | theirPawns)); b >>= 8);
backward = (b | (b >> 8)) & theirPawns;
}
}
// Test for candidate passed pawn
candidate = !passed
&& Position::file_is_half_open(theirPawns, f)
&& ( count_1s_max_15(neighboring_files_bb(f) & (behind_bb(us, r) | rank_bb(r)) & ourPawns)
- count_1s_max_15(neighboring_files_bb(f) & in_front_bb(us, r) & theirPawns)
>= 0);
// In order to prevent doubled passed pawns from receiving a too big
// bonus, only the frontmost passed pawn on each file is considered as
// a true passed pawn.
if (passed && (ourPawns & squares_in_front_of(us, s)))
{
// candidate = true;
passed = false;
}
// Score this pawn
Value mv = Value(0), ev = Value(0);
if (isolated)
{
mv -= IsolatedPawnMidgamePenalty[f];
ev -= IsolatedPawnEndgamePenalty[f];
if (Position::file_is_half_open(theirPawns, f))
{
mv -= IsolatedPawnMidgamePenalty[f] / 2;
ev -= IsolatedPawnEndgamePenalty[f] / 2;
}
}
if (doubled)
{
mv -= DoubledPawnMidgamePenalty[f];
ev -= DoubledPawnEndgamePenalty[f];
}
if (backward)
{
mv -= BackwardPawnMidgamePenalty[f];
ev -= BackwardPawnEndgamePenalty[f];
if (Position::file_is_half_open(theirPawns, f))
{
mv -= BackwardPawnMidgamePenalty[f] / 2;
ev -= BackwardPawnEndgamePenalty[f] / 2;
}
}
if (chain)
{
mv += ChainMidgameBonus[f];
ev += ChainEndgameBonus[f];
}
if (candidate)
{
mv += CandidateMidgameBonus[relative_rank(us, s)];
ev += CandidateEndgameBonus[relative_rank(us, s)];
}
mgValue[us] += mv;
egValue[us] += ev;
// If the pawn is passed, set the square of the pawn in the passedPawns
// bitboard
if (passed)
set_bit(&(pi->passedPawns), s);
} // while(pawns)
} // for(colors)
pi->mgValue = int16_t(mgValue[WHITE] - mgValue[BLACK]);
pi->egValue = int16_t(egValue[WHITE] - egValue[BLACK]);
// Evaluate pawns for both colors
pi->value = evaluate_pawns<WHITE>(pos, whitePawns, blackPawns, pi)
- evaluate_pawns<BLACK>(pos, blackPawns, whitePawns, pi);
return pi;
}
/// PawnInfoTable::evaluate_pawns() evaluates each pawn of the given color
template<Color Us>
Score PawnInfoTable::evaluate_pawns(const Position& pos, Bitboard ourPawns,
Bitboard theirPawns, PawnInfo* pi) {
Square s;
File f;
Rank r;
bool passed, isolated, doubled, chain, backward, candidate;
int bonus;
Score value = make_score(0, 0);
const Square* ptr = pos.piece_list_begin(Us, PAWN);
// Initialize pawn storm scores by giving bonuses for open files
for (f = FILE_A; f <= FILE_H; f++)
if (!(ourPawns & file_bb(f)))
{
pi->ksStormValue[Us] += KStormOpenFileBonus[f];
pi->qsStormValue[Us] += QStormOpenFileBonus[f];
pi->halfOpenFiles[Us] |= (1 << f);
}
// Loop through all pawns of the current color and score each pawn
while ((s = *ptr++) != SQ_NONE)
{
f = square_file(s);
r = square_rank(s);
assert(pos.piece_on(s) == piece_of_color_and_type(Us, PAWN));
// Passed, isolated or doubled pawn?
passed = Position::pawn_is_passed(theirPawns, Us, s);
isolated = Position::pawn_is_isolated(ourPawns, s);
doubled = Position::pawn_is_doubled(ourPawns, Us, s);
// We calculate kingside and queenside pawn storm
// scores for both colors. These are used when evaluating
// middle game positions with opposite side castling.
//
// Each pawn is given a base score given by a piece square table
// (KStormTable[] or QStormTable[]). Pawns which seem to have good
// chances of creating an open file by exchanging itself against an
// enemy pawn on an adjacent file gets an additional bonus.
// Kingside pawn storms
bonus = KStormTable[relative_square(Us, s)];
if (f >= FILE_F)
{
Bitboard b = outpost_mask(Us, s) & theirPawns & (FileFBB | FileGBB | FileHBB);
while (b)
{
// Give a bonus according to the distance of the nearest enemy pawn
Square s2 = pop_1st_bit(&b);
int v = StormLeverBonus[f] - 2 * square_distance(s, s2);
// If enemy pawn has no pawn beside itself is particularly vulnerable.
// Big bonus, especially against a weakness on the rook file
if (!(theirPawns & neighboring_files_bb(s2) & rank_bb(s2)))
v *= (square_file(s2) == FILE_H ? 4 : 2);
bonus += v;
}
}
pi->ksStormValue[Us] += bonus;
// Queenside pawn storms
bonus = QStormTable[relative_square(Us, s)];
if (f <= FILE_C)
{
Bitboard b = outpost_mask(Us, s) & theirPawns & (FileABB | FileBBB | FileCBB);
while (b)
{
// Give a bonus according to the distance of the nearest enemy pawn
Square s2 = pop_1st_bit(&b);
int v = StormLeverBonus[f] - 4 * square_distance(s, s2);
// If enemy pawn has no pawn beside itself is particularly vulnerable.
// Big bonus, especially against a weakness on the rook file
if (!(theirPawns & neighboring_files_bb(s2) & rank_bb(s2)))
v *= (square_file(s2) == FILE_A ? 4 : 2);
bonus += v;
}
}
pi->qsStormValue[Us] += bonus;
// Member of a pawn chain (but not the backward one)? We could speed up
// the test a little by introducing an array of masks indexed by color
// and square for doing the test, but because everything is hashed,
// it probably won't make any noticable difference.
chain = ourPawns
& neighboring_files_bb(f)
& (rank_bb(r) | rank_bb(r - (Us == WHITE ? 1 : -1)));
// Test for backward pawn
//
// If the pawn is passed, isolated, or member of a pawn chain
// it cannot be backward. If can capture an enemy pawn or if
// there are friendly pawns behind on neighboring files it cannot
// be backward either.
if ( (passed | isolated | chain)
|| (ourPawns & behind_bb(Us, r) & neighboring_files_bb(f))
|| (pos.attacks_from<PAWN>(s, Us) & theirPawns))
backward = false;
else
{
// We now know that there are no friendly pawns beside or behind this
// pawn on neighboring files. We now check whether the pawn is
// backward by looking in the forward direction on the neighboring
// files, and seeing whether we meet a friendly or an enemy pawn first.
Bitboard b = pos.attacks_from<PAWN>(s, Us);
// Note that we are sure to find something because pawn is not passed
// nor isolated, so loop is potentially infinite, but it isn't.
while (!(b & (ourPawns | theirPawns)))
Us == WHITE ? b <<= 8 : b >>= 8;
// The friendly pawn needs to be at least two ranks closer than the enemy
// pawn in order to help the potentially backward pawn advance.
backward = (b | (Us == WHITE ? b << 8 : b >> 8)) & theirPawns;
}
// Test for candidate passed pawn
candidate = !passed
&& !(theirPawns & file_bb(f))
&& ( count_1s_max_15(neighboring_files_bb(f) & (behind_bb(Us, r) | rank_bb(r)) & ourPawns)
- count_1s_max_15(neighboring_files_bb(f) & in_front_bb(Us, r) & theirPawns)
>= 0);
// In order to prevent doubled passed pawns from receiving a too big
// bonus, only the frontmost passed pawn on each file is considered as
// a true passed pawn.
if (passed && (ourPawns & squares_in_front_of(Us, s)))
passed = false;
// Score this pawn
if (passed)
set_bit(&(pi->passedPawns), s);
if (isolated)
{
value -= IsolatedPawnPenalty[f];
if (!(theirPawns & file_bb(f)))
value -= IsolatedPawnPenalty[f] / 2;
}
if (doubled)
value -= DoubledPawnPenalty[f];
if (backward)
{
value -= BackwardPawnPenalty[f];
if (!(theirPawns & file_bb(f)))
value -= BackwardPawnPenalty[f] / 2;
}
if (chain)
value += ChainBonus[f];
if (candidate)
value += CandidateBonus[relative_rank(Us, s)];
}
return value;
}
/// PawnInfo::updateShelter calculates and caches king shelter. It is called
/// only when king square changes, about 20% of total get_king_shelter() calls.
int PawnInfo::updateShelter(const Position& pos, Color c, Square ksq) {
unsigned shelter = 0;
Bitboard pawns = pos.pieces(PAWN, c) & this_and_neighboring_files_bb(ksq);
unsigned r = ksq & (7 << 3);
for (int i = 1, k = (c ? -8 : 8); i < 4; i++)
{
r += k;
shelter += BitCount8Bit[(pawns >> r) & 0xFF] * (128 >> i);
}
kingSquares[c] = ksq;
kingShelters[c] = shelter;
return shelter;
}
+21 -33
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -45,29 +45,30 @@ class PawnInfo {
friend class PawnInfoTable;
public:
PawnInfo() : key(0) { clear(); }
PawnInfo() { clear(); }
Value mg_value() const;
Value eg_value() const;
Score pawns_value() const;
Value kingside_storm_value(Color c) const;
Value queenside_storm_value(Color c) const;
Bitboard pawn_attacks(Color c) const;
Bitboard passed_pawns() const;
int file_is_half_open(Color c, File f) const;
int has_open_file_to_left(Color c, File f) const;
int has_open_file_to_right(Color c, File f) const;
int kingShelter(Color c, Square ksq) const;
void setKingShelter(Color c, Square ksq, int value);
int get_king_shelter(const Position& pos, Color c, Square ksq);
private:
inline void clear();
void clear();
int updateShelter(const Position& pos, Color c, Square ksq);
Key key;
Bitboard passedPawns;
int16_t mgValue, egValue;
Bitboard pawnAttacks[2];
Score value;
int16_t ksStormValue[2], qsStormValue[2];
uint8_t halfOpenFiles[2];
Square kingSquares[2];
int16_t kingShelters[2];
uint8_t kingShelters[2];
};
/// The PawnInfoTable class represents a pawn hash table. It is basically
@@ -83,6 +84,9 @@ public:
PawnInfo* get_pawn_info(const Position& pos);
private:
template<Color Us>
Score evaluate_pawns(const Position& pos, Bitboard ourPawns, Bitboard theirPawns, PawnInfo* pi);
unsigned size;
PawnInfo* entries;
};
@@ -92,18 +96,18 @@ private:
//// Inline functions
////
inline Value PawnInfo::mg_value() const {
return Value(mgValue);
}
inline Value PawnInfo::eg_value() const {
return Value(egValue);
inline Score PawnInfo::pawns_value() const {
return value;
}
inline Bitboard PawnInfo::passed_pawns() const {
return passedPawns;
}
inline Bitboard PawnInfo::pawn_attacks(Color c) const {
return pawnAttacks[c];
}
inline Value PawnInfo::kingside_storm_value(Color c) const {
return Value(ksStormValue[c]);
}
@@ -124,24 +128,8 @@ inline int PawnInfo::has_open_file_to_right(Color c, File f) const {
return halfOpenFiles[c] & ~((1 << int(f+1)) - 1);
}
inline int PawnInfo::kingShelter(Color c, Square ksq) const {
return (kingSquares[c] == ksq ? kingShelters[c] : -1);
inline int PawnInfo::get_king_shelter(const Position& pos, Color c, Square ksq) {
return (kingSquares[c] == ksq ? kingShelters[c] : updateShelter(pos, c, ksq));
}
inline void PawnInfo::setKingShelter(Color c, Square ksq, int value) {
kingSquares[c] = ksq;
kingShelters[c] = (int16_t)value;
}
inline void PawnInfo::clear() {
passedPawns = EmptyBoardBB;
mgValue = egValue = 0;
ksStormValue[WHITE] = ksStormValue[BLACK] = 0;
qsStormValue[WHITE] = qsStormValue[BLACK] = 0;
halfOpenFiles[WHITE] = halfOpenFiles[BLACK] = 0xFF;
kingSquares[WHITE] = kingSquares[BLACK] = SQ_NONE;
}
#endif // !defined(PAWNS_H_INCLUDED)
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+1 -5
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -83,10 +83,6 @@ inline int piece_is_slider(Piece p) {
return SlidingArray[int(p)];
}
inline int piece_is_slider(PieceType pt) {
return SlidingArray[int(pt)];
}
inline SquareDelta pawn_push(Color c) {
return (c == WHITE ? DELTA_N : DELTA_S);
}
+660 -822
View File
File diff suppressed because it is too large Load Diff
+98 -245
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -63,6 +63,18 @@ const int MaxGameLength = 220;
//// Types
////
/// struct checkInfo is initialized at c'tor time and keeps
/// info used to detect if a move gives check.
struct CheckInfo {
CheckInfo(const Position&);
Square ksq;
Bitboard dcCandidates;
Bitboard checkSq[8];
};
/// Castle rights, encoded as bit fields
enum CastleRights {
@@ -87,12 +99,13 @@ enum Phase {
/// must be passed as a parameter.
struct StateInfo {
Key key, pawnKey, materialKey;
int castleRights, rule50;
Key pawnKey, materialKey;
int castleRights, rule50, pliesFromNull;
Square epSquare;
Value mgValue, egValue;
Score value;
Value npMaterial[2];
Key key;
PieceType capture;
Bitboard checkersBB;
StateInfo* previous;
@@ -133,9 +146,9 @@ public:
};
// Constructors
Position() {};
Position(const Position& pos);
Position(const std::string& fen);
Position();
explicit Position(const Position& pos);
explicit Position(const std::string& fen);
// Text input/output
void from_fen(const std::string& fen);
@@ -143,7 +156,6 @@ public:
void print(Move m = MOVE_NONE) const;
// Copying
void copy(const Position& pos);
void flipped_copy(const Position& pos);
// The piece on a given square
@@ -162,26 +174,10 @@ public:
Bitboard empty_squares() const;
Bitboard occupied_squares() const;
Bitboard pieces_of_color(Color c) const;
Bitboard pieces_of_type(PieceType pt) const;
Bitboard pieces_of_color_and_type(Color c, PieceType pt) const;
Bitboard pawns() const;
Bitboard knights() const;
Bitboard bishops() const;
Bitboard rooks() const;
Bitboard queens() const;
Bitboard kings() const;
Bitboard rooks_and_queens() const;
Bitboard bishops_and_queens() const;
Bitboard sliders() const;
Bitboard pawns(Color c) const;
Bitboard knights(Color c) const;
Bitboard bishops(Color c) const;
Bitboard rooks(Color c) const;
Bitboard queens(Color c) const;
Bitboard kings(Color c) const;
Bitboard rooks_and_queens(Color c) const;
Bitboard bishops_and_queens(Color c) const;
Bitboard sliders_of_color(Color c) const;
Bitboard pieces(PieceType pt) const;
Bitboard pieces(PieceType pt, Color c) const;
Bitboard pieces(PieceType pt1, PieceType pt2) const;
Bitboard pieces(PieceType pt1, PieceType pt2, Color c) const;
// Number of pieces of each color and type
int piece_count(Color c, PieceType pt) const;
@@ -199,66 +195,50 @@ public:
Square initial_kr_square(Color c) const;
Square initial_qr_square(Color c) const;
// Attack bitboards
Bitboard sliding_attacks(Square s, Direction d) const;
Bitboard ray_attacks(Square s, SignedDirection d) const;
Bitboard pawn_attacks(Color c, Square s) const;
template<PieceType>
Bitboard piece_attacks(Square s) const;
// Bitboards for pinned pieces and discovered check candidates
Bitboard discovered_check_candidates(Color c) const;
Bitboard pinned_pieces(Color c, Bitboard& p) const;
Bitboard pinned_pieces(Color c) const;
// Checking pieces
// Checking pieces and under check information
Bitboard checkers() const;
bool is_check() const;
// Piece lists
Square piece_list(Color c, PieceType pt, int index) const;
const Square* piece_list_begin(Color c, PieceType pt) const;
// Attack information for a given square
bool square_is_attacked(Square s, Color c) const;
Bitboard attacks_to(Square s) const;
Bitboard attacks_to(Square s, Color c) const;
bool is_check() const;
bool pawn_attacks_square(Color c, Square f, Square t) const;
template<PieceType>
Bitboard piece_attacks_square(Square f, Square t) const; // Dispatch at compile-time
bool piece_attacks_square(Piece p, Square f, Square t) const; // Dispatch at run-time
// Information about attacks to or from a given square
Bitboard attackers_to(Square s) const;
Bitboard attacks_from(Piece p, Square s) const;
template<PieceType> Bitboard attacks_from(Square s) const;
template<PieceType> Bitboard attacks_from(Square s, Color c) const;
// Properties of moves
bool pl_move_is_legal(Move m) const;
bool pl_move_is_legal(Move m, Bitboard pinned) const;
bool pl_move_is_evasion(Move m, Bitboard pinned) const;
bool move_is_check(Move m) const;
bool move_is_check(Move m, Bitboard dcCandidates) const;
bool move_is_check(Move m, const CheckInfo& ci) const;
bool move_is_capture(Move m) const;
bool move_is_deep_pawn_push(Move m) const;
bool move_is_pawn_push_to_7th(Move m) const;
bool move_is_capture_or_promotion(Move m) const;
bool move_is_passed_pawn_push(Move m) const;
bool move_was_passed_pawn_push(Move m) const;
bool move_attacks_square(Move m, Square s) const;
// Piece captured with previous moves
PieceType captured_piece() const;
// Information about pawns
bool pawn_is_passed(Color c, Square s) const;
static bool pawn_is_passed(Bitboard theirPawns, Color c, Square s);
static bool pawn_is_isolated(Bitboard ourPawns, Square s);
static bool pawn_is_doubled(Bitboard ourPawns, Color c, Square s);
// Open and half-open files
static bool file_is_open(Bitboard pawns, File f);
static bool file_is_half_open(Bitboard pawns, File f);
// Weak squares
bool square_is_weak(Square s, Color c) const;
// Doing and undoing moves
void saveState();
void detach();
void do_move(Move m, StateInfo& st);
void do_move(Move m, StateInfo& st, Bitboard dcCandidates);
void do_move(Move m, StateInfo& st, const CheckInfo& ci, bool moveIsCheck);
void undo_move(Move m);
void do_null_move(StateInfo& st);
void undo_null_move();
@@ -267,18 +247,18 @@ public:
int see(Square from, Square to) const;
int see(Move m) const;
int see(Square to) const;
int see_sign(Move m) const;
// Accessing hash keys
Key get_key() const;
Key get_exclusion_key() const;
Key get_pawn_key() const;
Key get_material_key() const;
// Incremental evaluation
Value mg_value() const;
Value eg_value() const;
Score value() const;
Value non_pawn_material(Color c) const;
Phase game_phase() const;
template<GamePhase> Value pst_delta(Piece piece, Square from, Square to) const;
Score pst_delta(Piece piece, Square from, Square to) const;
// Game termination checks
bool is_mate() const;
@@ -300,7 +280,7 @@ public:
// Position consistency check, for debugging
bool is_ok(int* failedStep = NULL) const;
// Static member functions:
// Static member functions
static void init_zobrist();
static void init_piece_square_tables();
@@ -313,18 +293,11 @@ private:
void allow_ooo(Color c);
// Helper functions for doing and undoing moves
void do_capture_move(PieceType capture, Color them, Square to);
void do_capture_move(Bitboard& key, PieceType capture, Color them, Square to, bool ep);
void do_castle_move(Move m);
void do_promotion_move(Move m);
void do_ep_move(Move m);
void undo_castle_move(Move m);
void undo_promotion_move(Move m);
void undo_ep_move(Move m);
void find_checkers();
template<PieceType Piece>
void update_checkers(Bitboard* pCheckersBB, Square ksq, Square from, Square to, Bitboard dcCandidates);
template<bool FindPinned>
Bitboard hidden_checkers(Color c) const;
@@ -334,41 +307,40 @@ private:
Key compute_material_key() const;
// Computing incremental evaluation scores and material counts
template<GamePhase> Value pst(Color c, PieceType pt, Square s) const;
template<GamePhase> Value compute_value() const;
Score pst(Color c, PieceType pt, Square s) const;
Score compute_value() const;
Value compute_non_pawn_material(Color c) const;
// Bitboards
Bitboard byColorBB[2], byTypeBB[8];
// Board
Piece board[64];
// Bitboards
Bitboard byTypeBB[8], byColorBB[2];
// Piece counts
int pieceCount[2][8]; // [color][pieceType]
// Piece lists
Square pieceList[2][8][16]; // [color][pieceType][index]
int index[64];
int index[64]; // [square]
// Other info
Square kingSquare[2];
Color sideToMove;
int gamePly;
Key history[MaxGameLength];
int castleRightsMask[64];
File initialKFile, initialKRFile, initialQRFile;
StateInfo startState;
StateInfo* st;
// Static variables
static int castleRightsMask[64];
static Key zobrist[2][8][64];
static Key zobEp[64];
static Key zobCastle[16];
static Key zobMaterial[2][8][16];
static Key zobSideToMove;
static Value MgPieceSquareTable[16][64];
static Value EgPieceSquareTable[16][64];
static Score PieceSquareTable[16][64];
static Key zobExclusion;
};
@@ -420,80 +392,20 @@ inline Bitboard Position::pieces_of_color(Color c) const {
return byColorBB[c];
}
inline Bitboard Position::pieces_of_type(PieceType pt) const {
inline Bitboard Position::pieces(PieceType pt) const {
return byTypeBB[pt];
}
inline Bitboard Position::pieces_of_color_and_type(Color c, PieceType pt) const {
return pieces_of_color(c) & pieces_of_type(pt);
inline Bitboard Position::pieces(PieceType pt, Color c) const {
return byTypeBB[pt] & byColorBB[c];
}
inline Bitboard Position::pawns() const {
return pieces_of_type(PAWN);
inline Bitboard Position::pieces(PieceType pt1, PieceType pt2) const {
return byTypeBB[pt1] | byTypeBB[pt2];
}
inline Bitboard Position::knights() const {
return pieces_of_type(KNIGHT);
}
inline Bitboard Position::bishops() const {
return pieces_of_type(BISHOP);
}
inline Bitboard Position::rooks() const {
return pieces_of_type(ROOK);
}
inline Bitboard Position::queens() const {
return pieces_of_type(QUEEN);
}
inline Bitboard Position::kings() const {
return pieces_of_type(KING);
}
inline Bitboard Position::rooks_and_queens() const {
return rooks() | queens();
}
inline Bitboard Position::bishops_and_queens() const {
return bishops() | queens();
}
inline Bitboard Position::sliders() const {
return bishops() | queens() | rooks();
}
inline Bitboard Position::pawns(Color c) const {
return pieces_of_color_and_type(c, PAWN);
}
inline Bitboard Position::knights(Color c) const {
return pieces_of_color_and_type(c, KNIGHT);
}
inline Bitboard Position::bishops(Color c) const {
return pieces_of_color_and_type(c, BISHOP);
}
inline Bitboard Position::rooks(Color c) const {
return pieces_of_color_and_type(c, ROOK);
}
inline Bitboard Position::queens(Color c) const {
return pieces_of_color_and_type(c, QUEEN);
}
inline Bitboard Position::kings(Color c) const {
return pieces_of_color_and_type(c, KING);
}
inline Bitboard Position::rooks_and_queens(Color c) const {
return rooks_and_queens() & pieces_of_color(c);
}
inline Bitboard Position::bishops_and_queens(Color c) const {
return bishops_and_queens() & pieces_of_color(c);
inline Bitboard Position::pieces(PieceType pt1, PieceType pt2, Color c) const {
return (byTypeBB[pt1] | byTypeBB[pt2]) & byColorBB[c];
}
inline int Position::piece_count(Color c, PieceType pt) const {
@@ -504,12 +416,16 @@ inline Square Position::piece_list(Color c, PieceType pt, int index) const {
return pieceList[c][pt][index];
}
inline const Square* Position::piece_list_begin(Color c, PieceType pt) const {
return pieceList[c][pt];
}
inline Square Position::ep_square() const {
return st->epSquare;
}
inline Square Position::king_square(Color c) const {
return kingSquare[c];
return pieceList[c][KING][0];
}
inline bool Position::can_castle_kingside(Color side) const {
@@ -532,33 +448,29 @@ inline Square Position::initial_qr_square(Color c) const {
return relative_square(c, make_square(initialQRFile, RANK_1));
}
inline Bitboard Position::pawn_attacks(Color c, Square s) const {
template<>
inline Bitboard Position::attacks_from<PAWN>(Square s, Color c) const {
return StepAttackBB[piece_of_color_and_type(c, PAWN)][s];
}
template<PieceType Piece> // Knight and King
inline Bitboard Position::piece_attacks(Square s) const {
template<PieceType Piece> // Knight and King and white pawns
inline Bitboard Position::attacks_from(Square s) const {
return StepAttackBB[Piece][s];
}
template<>
inline Bitboard Position::piece_attacks<PAWN>(Square s) const {
return StepAttackBB[piece_of_color_and_type(opposite_color(sideToMove), PAWN)][s];
}
template<>
inline Bitboard Position::piece_attacks<BISHOP>(Square s) const {
inline Bitboard Position::attacks_from<BISHOP>(Square s) const {
return bishop_attacks_bb(s, occupied_squares());
}
template<>
inline Bitboard Position::piece_attacks<ROOK>(Square s) const {
inline Bitboard Position::attacks_from<ROOK>(Square s) const {
return rook_attacks_bb(s, occupied_squares());
}
template<>
inline Bitboard Position::piece_attacks<QUEEN>(Square s) const {
return piece_attacks<ROOK>(s) | piece_attacks<BISHOP>(s);
inline Bitboard Position::attacks_from<QUEEN>(Square s) const {
return attacks_from<ROOK>(s) | attacks_from<BISHOP>(s);
}
inline Bitboard Position::checkers() const {
@@ -569,27 +481,8 @@ inline bool Position::is_check() const {
return st->checkersBB != EmptyBoardBB;
}
inline bool Position::pawn_attacks_square(Color c, Square f, Square t) const {
return bit_is_set(pawn_attacks(c, f), t);
}
template<PieceType Piece>
Bitboard Position::piece_attacks_square(Square f, Square t) const {
return bit_is_set(piece_attacks<Piece>(f), t);
}
inline Bitboard Position::attacks_to(Square s, Color c) const {
return attacks_to(s) & pieces_of_color(c);
}
inline bool Position::square_is_attacked(Square s, Color c) const {
return attacks_to(s, c) != EmptyBoardBB;
}
inline bool Position::pawn_is_passed(Color c, Square s) const {
return !(pawns(opposite_color(c)) & passed_pawn_mask(c, s));
return !(pieces(PAWN, opposite_color(c)) & passed_pawn_mask(c, s));
}
inline bool Position::pawn_is_passed(Bitboard theirPawns, Color c, Square s) {
@@ -604,22 +497,18 @@ inline bool Position::pawn_is_doubled(Bitboard ourPawns, Color c, Square s) {
return ourPawns & squares_behind(c, s);
}
inline bool Position::file_is_open(Bitboard pawns, File f) {
return !(pawns & file_bb(f));
}
inline bool Position::file_is_half_open(Bitboard pawns, File f) {
return !(pawns & file_bb(f));
}
inline bool Position::square_is_weak(Square s, Color c) const {
return !(pawns(c) & outpost_mask(opposite_color(c), s));
return !(pieces(PAWN, c) & outpost_mask(opposite_color(c), s));
}
inline Key Position::get_key() const {
return st->key;
}
inline Key Position::get_exclusion_key() const {
return st->key ^ zobExclusion;
}
inline Key Position::get_pawn_key() const {
return st->pawnKey;
}
@@ -628,60 +517,22 @@ inline Key Position::get_material_key() const {
return st->materialKey;
}
template<Position::GamePhase Ph>
inline Value Position::pst(Color c, PieceType pt, Square s) const {
return (Ph == MidGame ? MgPieceSquareTable[piece_of_color_and_type(c, pt)][s]
: EgPieceSquareTable[piece_of_color_and_type(c, pt)][s]);
inline Score Position::pst(Color c, PieceType pt, Square s) const {
return PieceSquareTable[piece_of_color_and_type(c, pt)][s];
}
template<Position::GamePhase Ph>
inline Value Position::pst_delta(Piece piece, Square from, Square to) const {
return (Ph == MidGame ? MgPieceSquareTable[piece][to] - MgPieceSquareTable[piece][from]
: EgPieceSquareTable[piece][to] - EgPieceSquareTable[piece][from]);
inline Score Position::pst_delta(Piece piece, Square from, Square to) const {
return PieceSquareTable[piece][to] - PieceSquareTable[piece][from];
}
inline Value Position::mg_value() const {
return st->mgValue;
}
inline Value Position::eg_value() const {
return st->egValue;
inline Score Position::value() const {
return st->value;
}
inline Value Position::non_pawn_material(Color c) const {
return st->npMaterial[c];
}
inline Phase Position::game_phase() const {
// Values modified by Joona Kiiski
static const Value MidgameLimit = Value(15713);
static const Value EndgameLimit = Value(4428);
Value npm = non_pawn_material(WHITE) + non_pawn_material(BLACK);
if (npm >= MidgameLimit)
return PHASE_MIDGAME;
else if(npm <= EndgameLimit)
return PHASE_ENDGAME;
else
return Phase(((npm - EndgameLimit) * 128) / (MidgameLimit - EndgameLimit));
}
inline bool Position::move_is_deep_pawn_push(Move m) const {
Color c = side_to_move();
return piece_on(move_from(m)) == piece_of_color_and_type(c, PAWN)
&& relative_rank(c, move_to(m)) > RANK_4;
}
inline bool Position::move_is_pawn_push_to_7th(Move m) const {
Color c = side_to_move();
return piece_on(move_from(m)) == piece_of_color_and_type(c, PAWN)
&& relative_rank(c, move_to(m)) == RANK_7;
}
inline bool Position::move_is_passed_pawn_push(Move m) const {
Color c = side_to_move();
@@ -689,13 +540,6 @@ inline bool Position::move_is_passed_pawn_push(Move m) const {
&& pawn_is_passed(c, move_to(m));
}
inline bool Position::move_was_passed_pawn_push(Move m) const {
Color c = opposite_color(side_to_move());
return piece_on(move_to(m)) == piece_of_color_and_type(c, PAWN)
&& pawn_is_passed(c, move_to(m));
}
inline int Position::rule_50_counter() const {
return st->rule50;
@@ -710,14 +554,23 @@ inline bool Position::opposite_colored_bishops() const {
inline bool Position::has_pawn_on_7th(Color c) const {
return pawns(c) & relative_rank_bb(c, RANK_7);
return pieces(PAWN, c) & relative_rank_bb(c, RANK_7);
}
inline bool Position::move_is_capture(Move m) const {
// Move must not be MOVE_NONE !
return (m & (3 << 15)) ? !move_is_castle(m) : !square_is_empty(move_to(m));
}
return (!square_is_empty(move_to(m)) && !move_is_castle(m)) || move_is_ep(m);
inline bool Position::move_is_capture_or_promotion(Move m) const {
// Move must not be MOVE_NONE !
return (m & (0x1F << 12)) ? !move_is_castle(m) : !square_is_empty(move_to(m));
}
inline PieceType Position::captured_piece() const {
return st->capture;
}
#endif // !defined(POSITION_H_INCLUDED)
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+14 -11
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -63,7 +63,7 @@ namespace {
/// that the move is a legal move from the position. The return value is
/// a string containing the move in short algebraic notation.
const string move_to_san(const Position& pos, Move m) {
const string move_to_san(Position& pos, Move m) {
assert(pos.is_ok());
assert(move_is_ok(m));
@@ -123,10 +123,10 @@ const string move_to_san(const Position& pos, Move m) {
// Position::move_is_check doesn't detect all checks (not castling moves,
// promotions and en passant captures).
StateInfo st;
Position p(pos);
p.do_move(m, st);
if (p.is_check())
san += p.is_mate()? "#" : "+";
pos.do_move(m, st);
if (pos.is_check())
san += pos.is_mate() ? "#" : "+";
pos.undo_move(m);
return san;
}
@@ -142,13 +142,14 @@ Move move_from_san(const Position& pos, const string& movestr) {
assert(pos.is_ok());
MovePicker mp = MovePicker(pos, MOVE_NONE, OnePly, H);
Bitboard pinned = pos.pinned_pieces(pos.side_to_move());
// Castling moves
if (movestr == "O-O-O" || movestr == "O-O-O+")
{
Move m;
while ((m = mp.get_next_move()) != MOVE_NONE)
if (move_is_long_castle(m) && pos.pl_move_is_legal(m))
if (move_is_long_castle(m) && pos.pl_move_is_legal(m, pinned))
return m;
return MOVE_NONE;
@@ -157,7 +158,7 @@ Move move_from_san(const Position& pos, const string& movestr) {
{
Move m;
while ((m = mp.get_next_move()) != MOVE_NONE)
if (move_is_short_castle(m) && pos.pl_move_is_legal(m))
if (move_is_short_castle(m) && pos.pl_move_is_legal(m, pinned))
return m;
return MOVE_NONE;
@@ -325,14 +326,15 @@ const string line_to_san(const Position& pos, Move line[], int startColumn, bool
/// when the UCI parameter "Use Search Log" is "true").
const string pretty_pv(const Position& pos, int time, int depth,
uint64_t nodes, Value score, Move pv[]) {
uint64_t nodes, Value score, ValueType type, Move pv[]) {
std::stringstream s;
// Depth
s << std::setw(2) << depth << " ";
// Score
s << std::setw(8) << score_string(score);
s << ((type == VALUE_TYPE_LOWER)? ">" : ((type == VALUE_TYPE_UPPER)? "<" : " "));
s << std::setw(7) << score_string(score);
// Time
s << std::setw(8) << time_string(time) << " ";
@@ -366,11 +368,12 @@ namespace {
return AMBIGUITY_NONE;
MovePicker mp = MovePicker(pos, MOVE_NONE, OnePly, H);
Bitboard pinned = pos.pinned_pieces(pos.side_to_move());
Move mv, moveList[8];
int n = 0;
while ((mv = mp.get_next_move()) != MOVE_NONE)
if (move_to(mv) == to && pos.piece_on(move_from(mv)) == pc && pos.pl_move_is_legal(mv))
if (move_to(mv) == to && pos.piece_on(move_from(mv)) == pc && pos.pl_move_is_legal(mv, pinned))
moveList[n++] = mv;
if (n == 1)
+3 -3
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -36,9 +36,9 @@
//// Prototypes
////
extern const std::string move_to_san(const Position& pos, Move m);
extern const std::string move_to_san(Position& pos, Move m);
extern Move move_from_san(const Position& pos, const std::string& str);
extern const std::string line_to_san(const Position& pos, Move line[], int startColumn, bool breakLines);
extern const std::string pretty_pv(const Position& pos, int time, int depth, uint64_t nodes, Value score, Move pv[]);
extern const std::string pretty_pv(const Position& pos, int time, int depth, uint64_t nodes, Value score, ValueType type, Move pv[]);
#endif // !defined(SAN_H_INCLUDED)
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+1814 -1543
View File
File diff suppressed because it is too large Load Diff
+8 -3
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -27,6 +27,7 @@
#include "depth.h"
#include "move.h"
#include "value.h"
////
@@ -46,14 +47,16 @@ const int KILLER_MAX = 2;
/// from nodes shallower and deeper in the tree during the search. Each
/// search thread has its own array of SearchStack objects, indexed by the
/// current ply.
struct EvalInfo;
struct SearchStack {
Move pv[PLY_MAX];
Move pv[PLY_MAX_PLUS_2];
Move currentMove;
Move mateKiller;
Move threatMove;
Move killers[KILLER_MAX];
Depth reduction;
Value eval;
void init(int ply);
void initKillers();
@@ -64,11 +67,13 @@ struct SearchStack {
//// Prototypes
////
extern void init_search();
extern void init_threads();
extern void stop_threads();
extern void exit_threads();
extern bool think(const Position &pos, bool infinite, bool ponder, int side_to_move,
int time[], int increment[], int movesToGo, int maxDepth,
int maxNodes, int maxTime, Move searchMoves[]);
extern int perft(Position &pos, Depth depth);
extern int64_t nodes_searched();
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+37 -21
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -26,6 +26,8 @@
//// Includes
////
#include <cstring>
#include "lock.h"
#include "movepick.h"
#include "position.h"
@@ -36,7 +38,8 @@
//// Constants and variables
////
const int THREAD_MAX = 8;
const int MAX_THREADS = 8;
const int ACTIVE_SPLIT_POINTS_MAX = 8;
////
@@ -44,35 +47,48 @@ const int THREAD_MAX = 8;
////
struct SplitPoint {
SplitPoint *parent;
Position pos;
SearchStack sstack[THREAD_MAX][PLY_MAX];
SearchStack *parentSstack;
int ply;
Depth depth;
volatile Value alpha, beta, bestValue;
// Const data after splitPoint has been setup
SplitPoint* parent;
const Position* pos;
bool pvNode;
Bitboard dcCandidates;
int master, slaves[THREAD_MAX];
Depth depth;
bool mateThreat;
Value beta;
int ply, master, slaves[MAX_THREADS];
SearchStack sstack[MAX_THREADS][PLY_MAX_PLUS_2];
// Const pointers to shared data
MovePicker* mp;
SearchStack* parentSstack;
// Shared data
Lock lock;
MovePicker *mp;
volatile Value alpha;
volatile Value bestValue;
volatile int moves;
volatile int cpus;
bool finished;
volatile bool stopRequest;
};
// ThreadState type is used to represent thread's current state
enum ThreadState
{
THREAD_SEARCHING, // thread is performing work
THREAD_AVAILABLE, // thread is polling for work
THREAD_SLEEPING, // we are not thinking, so thread is sleeping
THREAD_BOOKED, // other thread (master) has booked us as a slave
THREAD_WORKISWAITING, // master has ordered us to start
THREAD_TERMINATED // we are quitting and thread is terminated
};
struct Thread {
SplitPoint *splitPoint;
int activeSplitPoints;
SplitPoint* splitPoint;
volatile int activeSplitPoints;
uint64_t nodes;
uint64_t betaCutOffs[2];
bool failHighPly1;
volatile bool stop;
volatile bool running;
volatile bool idle;
volatile bool workIsWaiting;
volatile bool printCurrentLine;
volatile ThreadState state;
unsigned char pad[64]; // set some distance among local data for each thread
};
+92 -25
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -25,9 +25,15 @@
#include <cassert>
#include <cmath>
#include <cstring>
#if !(defined(__hpux) || defined(__ppc__) || defined(__ppc64__) || defined(__arm__))
# include <xmmintrin.h>
#endif
#include "movegen.h"
#include "tt.h"
// The main transposition table
TranspositionTable TT;
////
//// Functions
@@ -49,22 +55,20 @@ TranspositionTable::~TranspositionTable() {
/// TranspositionTable::set_size sets the size of the transposition table,
/// measured in megabytes.
void TranspositionTable::set_size(unsigned mbSize) {
void TranspositionTable::set_size(size_t mbSize) {
assert(mbSize >= 4 && mbSize <= 4096);
size_t newSize = 1024;
unsigned newSize = 1024;
// We store a cluster of 4 TTEntry for each position and newSize is
// the maximum number of storable positions
while ((2 * newSize) * 4 * (sizeof(TTEntry)) <= (mbSize << 20))
// We store a cluster of ClusterSize number of TTEntry for each position
// and newSize is the maximum number of storable positions.
while ((2 * newSize) * sizeof(TTCluster) <= (mbSize << 20))
newSize *= 2;
if (newSize != size)
{
size = newSize;
delete [] entries;
entries = new TTEntry[size * 4];
entries = new TTCluster[size];
if (!entries)
{
std::cerr << "Failed to allocate " << mbSize
@@ -83,7 +87,17 @@ void TranspositionTable::set_size(unsigned mbSize) {
void TranspositionTable::clear() {
memset(entries, 0, size * 4 * sizeof(TTEntry));
memset(entries, 0, size * sizeof(TTCluster));
}
/// TranspositionTable::first_entry returns a pointer to the first
/// entry of a cluster given a position. The low 32 bits of the key
/// are used to get the index in the table.
inline TTEntry* TranspositionTable::first_entry(const Key posKey) const {
return entries[uint32_t(posKey) & (size - 1)].data;
}
@@ -100,20 +114,22 @@ void TranspositionTable::clear() {
void TranspositionTable::store(const Key posKey, Value v, ValueType t, Depth d, Move m) {
TTEntry *tte, *replace;
uint32_t posKey32 = posKey >> 32; // Use the high 32 bits as key
tte = replace = first_entry(posKey);
for (int i = 0; i < 4; i++, tte++)
for (int i = 0; i < ClusterSize; i++, tte++)
{
if (!tte->key() || tte->key() == posKey) // empty or overwrite old
if (!tte->key() || tte->key() == posKey32) // empty or overwrite old
{
// Do not overwrite when new type is VALUE_TYPE_EVAL
if (tte->key() && t == VALUE_TYPE_EVAL)
// Do not overwrite when new type is VALUE_TYPE_EV_LO
if (tte->key() && t == VALUE_TYPE_EV_LO)
return;
// Preserve any exsisting ttMove
if (m == MOVE_NONE)
m = tte->move();
*tte = TTEntry(posKey, v, t, d, m, generation);
*tte = TTEntry(posKey32, v, t, d, m, generation);
return;
}
else if (i == 0) // replace would be a no-op in this common case
@@ -126,7 +142,7 @@ void TranspositionTable::store(const Key posKey, Value v, ValueType t, Depth d,
if (c1 + c2 + c3 > 0)
replace = tte;
}
*replace = TTEntry(posKey, v, t, d, m, generation);
*replace = TTEntry(posKey32, v, t, d, m, generation);
writes++;
}
@@ -137,24 +153,42 @@ void TranspositionTable::store(const Key posKey, Value v, ValueType t, Depth d,
TTEntry* TranspositionTable::retrieve(const Key posKey) const {
TTEntry *tte = first_entry(posKey);
uint32_t posKey32 = posKey >> 32;
TTEntry* tte = first_entry(posKey);
for (int i = 0; i < 4; i++, tte++)
if (tte->key() == posKey)
for (int i = 0; i < ClusterSize; i++, tte++)
if (tte->key() == posKey32)
return tte;
return NULL;
}
/// TranspositionTable::first_entry returns a pointer to the first
/// entry of a cluster given a position.
/// TranspositionTable::prefetch looks up the current position in the
/// transposition table and load it in L1/L2 cache. This is a non
/// blocking function and do not stalls the CPU waiting for data
/// to be loaded from RAM, that can be very slow. When we will
/// subsequently call retrieve() the TT data will be already
/// quickly accessible in L1/L2 CPU cache.
#if defined(__hpux) || defined(__ppc__) || defined(__ppc64__) || defined(__arm__)
void TranspositionTable::prefetch(const Key) const {} // Not supported on HP UX
#else
inline TTEntry* TranspositionTable::first_entry(const Key posKey) const {
void TranspositionTable::prefetch(const Key posKey) const {
return entries + (int(posKey & (size - 1)) << 2);
#if defined(__INTEL_COMPILER) || defined(__ICL)
// This hack prevents prefetches to be optimized away by
// Intel compiler. Both MSVC and gcc seems not affected.
__asm__ ("");
#endif
char const* addr = (char*)first_entry(posKey);
_mm_prefetch(addr, _MM_HINT_T2);
_mm_prefetch(addr+64, _MM_HINT_T2); // 64 bytes ahead
}
#endif
/// TranspositionTable::new_search() is called at the beginning of every new
/// search. It increments the "generation" variable, which is used to
/// distinguish transposition table entries from previous searches from
@@ -179,18 +213,51 @@ void TranspositionTable::insert_pv(const Position& pos, Move pv[]) {
for (int i = 0; pv[i] != MOVE_NONE; i++)
{
store(p.get_key(), VALUE_NONE, VALUE_TYPE_NONE, Depth(-127*OnePly), pv[i]);
TTEntry *tte = retrieve(p.get_key());
if (!tte || tte->move() != pv[i])
store(p.get_key(), VALUE_NONE, VALUE_TYPE_NONE, Depth(-127*OnePly), pv[i]);
p.do_move(pv[i], st);
}
}
/// TranspositionTable::extract_pv() extends a PV by adding moves from the
/// transposition table at the end. This should ensure that the PV is almost
/// always at least two plies long, which is important, because otherwise we
/// will often get single-move PVs when the search stops while failing high,
/// and a single-move PV means that we don't have a ponder move.
void TranspositionTable::extract_pv(const Position& pos, Move pv[], const int PLY_MAX) {
const TTEntry* tte;
StateInfo st;
Position p(pos);
int ply = 0;
// Update position to the end of current PV
while (pv[ply] != MOVE_NONE)
p.do_move(pv[ply++], st);
// Try to add moves from TT while possible
while ( (tte = retrieve(p.get_key())) != NULL
&& tte->move() != MOVE_NONE
&& move_is_legal(p, tte->move())
&& (!p.is_draw() || ply < 2)
&& ply < PLY_MAX)
{
pv[ply] = tte->move();
p.do_move(pv[ply++], st);
}
pv[ply] = MOVE_NONE;
}
/// TranspositionTable::full() returns the permill of all transposition table
/// entries which have received at least one write during the current search.
/// It is used to display the "info hashfull ..." information in UCI.
int TranspositionTable::full() const {
double N = double(size) * 4.0;
double N = double(size) * ClusterSize;
return int(1000 * (1 - exp(writes * log(1.0 - 1.0/N))));
}
+32 -14
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -36,12 +36,12 @@
/// The TTEntry class is the class of transposition table entries
///
/// A TTEntry needs 128 bits to be stored
/// A TTEntry needs 96 bits to be stored
///
/// bit 0-63: key
/// bit 64-95: data
/// bit 96-111: value
/// bit 112-127: depth
/// bit 0-31: key
/// bit 32-63: data
/// bit 64-79: value
/// bit 80-95: depth
///
/// the 32 bits of the data field are so defined
///
@@ -54,11 +54,11 @@ class TTEntry {
public:
TTEntry() {}
TTEntry(Key k, Value v, ValueType t, Depth d, Move m, int generation)
TTEntry(uint32_t k, Value v, ValueType t, Depth d, Move m, int generation)
: key_ (k), data((m & 0x1FFFF) | (t << 20) | (generation << 23)),
value_(int16_t(v)), depth_(int16_t(d)) {}
Key key() const { return key_; }
uint32_t key() const { return key_; }
Depth depth() const { return Depth(depth_); }
Move move() const { return Move(data & 0x1FFFF); }
Value value() const { return Value(value_); }
@@ -66,13 +66,27 @@ public:
int generation() const { return (data >> 23); }
private:
Key key_;
uint32_t key_;
uint32_t data;
int16_t value_;
int16_t depth_;
};
/// The transposition table class. This is basically just a huge array
/// This is the number of TTEntry slots for each position
const int ClusterSize = 5;
/// Each group of ClusterSize number of TTEntry form a TTCluster
/// that is indexed by a single position key. Cluster is padded
/// to a cache line size so to guarantee always aligned accesses.
struct TTCluster {
TTEntry data[ClusterSize];
char cache_line_padding[64 - sizeof(TTEntry[ClusterSize])];
};
/// The transposition table class. This is basically just a huge array
/// containing TTEntry objects, and a few methods for writing new entries
/// and reading new ones.
@@ -81,26 +95,30 @@ class TranspositionTable {
public:
TranspositionTable();
~TranspositionTable();
void set_size(unsigned mbSize);
void set_size(size_t mbSize);
void clear();
void store(const Key posKey, Value v, ValueType type, Depth d, Move m);
TTEntry* retrieve(const Key posKey) const;
void prefetch(const Key posKey) const;
void new_search();
void insert_pv(const Position& pos, Move pv[]);
void extract_pv(const Position& pos, Move pv[], const int PLY_MAX);
int full() const;
private:
inline TTEntry* first_entry(const Key posKey) const;
// Be sure 'writes' is at least one cacheline away
// Be sure 'writes' is at least one cache line away
// from read only variables.
unsigned char pad_before[64 - sizeof(unsigned)];
unsigned writes; // heavy SMP read/write access here
unsigned char pad_after[64];
unsigned size;
TTEntry* entries;
size_t size;
TTCluster* entries;
uint8_t generation;
};
extern TranspositionTable TT;
#endif // !defined(TT_H_INCLUDED)
+35 -4
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -31,9 +31,9 @@ typedef __int8 int8_t;
typedef unsigned __int8 uint8_t;
typedef __int16 int16;
typedef unsigned __int16 uint16_t;
typedef __int32 int32;
typedef __int32 int32_t;
typedef unsigned __int32 uint32_t;
typedef __int64 int64;
typedef __int64 int64_t;
typedef unsigned __int64 uint64_t;
typedef __int16 int16_t;
@@ -62,8 +62,39 @@ typedef uint64_t Bitboard;
#define IS_64BIT
#endif
#if defined(IS_64BIT) && !defined(_WIN64) && (defined(__GNUC__) || defined(__INTEL_COMPILER))
#if defined(IS_64BIT) && (defined(__GNUC__) || defined(__INTEL_COMPILER))
#define USE_BSFQ
#endif
// Cache line alignment specification
#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
#define CACHE_LINE_ALIGNMENT __declspec(align(64))
#else
#define CACHE_LINE_ALIGNMENT __attribute__ ((aligned(64)))
#endif
// Define a __cpuid() function for gcc compilers, for Intel and MSVC
// is already available as an intrinsic.
#if defined(_MSC_VER)
#include <intrin.h>
#elif defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))
inline void __cpuid(int CPUInfo[4], int InfoType)
{
int* eax = CPUInfo + 0;
int* ebx = CPUInfo + 1;
int* ecx = CPUInfo + 2;
int* edx = CPUInfo + 3;
*eax = InfoType;
*ecx = 0;
__asm__("cpuid" : "=a" (*eax), "=b" (*ebx), "=c" (*ecx), "=d" (*edx)
: "0" (*eax), "2" (*ecx));
}
#else
inline void __cpuid(int CPUInfo[4], int)
{
CPUInfo[0] = CPUInfo[1] = CPUInfo[2] = CPUInfo[3] = 0;
}
#endif
#endif // !defined(TYPES_H_INCLUDED)
+50 -43
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -61,6 +61,7 @@ namespace {
void set_option(UCIInputParser& uip);
void set_position(UCIInputParser& uip);
bool go(UCIInputParser& uip);
void perft(UCIInputParser& uip);
}
@@ -106,7 +107,8 @@ namespace {
UCIInputParser uip(command);
string token;
uip >> token; // operator>>() skips any whitespace
if (!(uip >> token)) // operator>>() skips any whitespace
return true;
if (token == "quit")
return false;
@@ -123,7 +125,7 @@ namespace {
}
else if (token == "ucinewgame")
{
push_button("Clear Hash");
push_button("New Game");
Position::init_piece_square_tables();
RootPosition.from_fen(StartPosition);
}
@@ -147,23 +149,19 @@ namespace {
else if (token == "eval")
{
EvalInfo ei;
cout << "Incremental mg: " << RootPosition.mg_value()
<< "\nIncremental eg: " << RootPosition.eg_value()
cout << "Incremental mg: " << mg_value(RootPosition.value())
<< "\nIncremental eg: " << eg_value(RootPosition.value())
<< "\nFull eval: " << evaluate(RootPosition, ei, 0) << endl;
}
else if (token == "key")
cout << "key: " << hex << RootPosition.get_key()
<< "\nmaterial key: " << RootPosition.get_material_key()
<< "\npawn key: " << RootPosition.get_pawn_key() << endl;
else if (token == "perft")
perft(uip);
else
{
cout << "Unknown command: " << command << endl;
while (!uip.eof())
{
uip >> token;
cout << token << endl;
}
}
return true;
}
@@ -178,33 +176,33 @@ namespace {
string token;
uip >> token; // operator>>() skips any whitespace
if (!(uip >> token)) // operator>>() skips any whitespace
return;
if (token == "startpos")
RootPosition.from_fen(StartPosition);
else if (token == "fen")
{
string fen;
while (token != "moves" && !uip.eof())
while (uip >> token && token != "moves")
{
uip >> token;
fen += token;
fen += ' ';
}
RootPosition.from_fen(fen);
}
if (!uip.eof())
if (uip.good())
{
if (token != "moves")
uip >> token;
if (token == "moves")
{
Move move;
StateInfo st;
while (!uip.eof())
while (uip >> token)
{
uip >> token;
move = move_from_string(RootPosition, token);
RootPosition.do_move(move, st);
if (RootPosition.rule_50_counter() == 0)
@@ -212,7 +210,7 @@ namespace {
}
// Our StateInfo st is about going out of scope so copy
// its content inside RootPosition before they disappear.
RootPosition.saveState();
RootPosition.detach();
}
}
}
@@ -226,24 +224,22 @@ namespace {
void set_option(UCIInputParser& uip) {
string token, name;
string token, name, value;
uip >> token;
if (token == "name")
if (!(uip >> token)) // operator>>() skips any whitespace
return;
if (token == "name" && uip >> name)
{
uip >> name;
while (!uip.eof())
{
uip >> token;
if (token == "value")
break;
while (uip >> token && token != "value")
name += (" " + token);
}
if (token == "value")
if (token == "value" && uip >> value)
{
getline(uip, token); // reads until end of line
set_option_value(name, token);
while (uip >> token)
value += (" " + token);
set_option_value(name, value);
} else
push_button(name);
}
@@ -270,10 +266,8 @@ namespace {
searchMoves[0] = MOVE_NONE;
while (!uip.eof())
while (uip >> token)
{
uip >> token;
if (token == "infinite")
infinite = true;
else if (token == "ponder")
@@ -297,22 +291,35 @@ namespace {
else if (token == "searchmoves")
{
int numOfMoves = 0;
while (!uip.eof())
{
uip >> token;
while (uip >> token)
searchMoves[numOfMoves++] = move_from_string(RootPosition, token);
}
searchMoves[numOfMoves] = MOVE_NONE;
}
}
if (moveTime)
infinite = true; // HACK
assert(RootPosition.is_ok());
return think(RootPosition, infinite, ponder, RootPosition.side_to_move(),
time, inc, movesToGo, depth, nodes, moveTime, searchMoves);
}
void perft(UCIInputParser& uip) {
string token;
int depth, tm, n;
Position pos(RootPosition);
if (!(uip >> depth))
return;
tm = get_system_time();
n = perft(pos, depth * OnePly);
tm = get_system_time() - tm;
std::cout << "\nNodes " << n
<< "\nTime (ms) " << tm
<< "\nNodes/second " << (int)(n/(tm/1000.0)) << std::endl;
}
}
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+49 -46
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -88,26 +88,10 @@ namespace {
o["Space"] = Option(100, 0, 200);
o["Aggressiveness"] = Option(100, 0, 200);
o["Cowardice"] = Option(100, 0, 200);
o["King Safety Curve"] = Option("Quadratic", COMBO);
o["King Safety Curve"].comboValues.push_back("Quadratic");
o["King Safety Curve"].comboValues.push_back("Linear"); /*, "From File"*/
o["King Safety Coefficient"] = Option(40, 1, 100);
o["King Safety X Intercept"] = Option(0, 0, 20);
o["King Safety Max Slope"] = Option(30, 10, 100);
o["King Safety Max Value"] = Option(500, 100, 1000);
o["Queen Contact Check Bonus"] = Option(3, 0, 8);
o["Queen Check Bonus"] = Option(2, 0, 4);
o["Rook Check Bonus"] = Option(1, 0, 4);
o["Bishop Check Bonus"] = Option(1, 0, 4);
o["Knight Check Bonus"] = Option(1, 0, 4);
o["Discovered Check Bonus"] = Option(3, 0, 8);
o["Mate Threat Bonus"] = Option(3, 0, 8);
o["Check Extension (PV nodes)"] = Option(2, 0, 2);
o["Check Extension (non-PV nodes)"] = Option(1, 0, 2);
o["Single Reply Extension (PV nodes)"] = Option(2, 0, 2);
o["Single Reply Extension (non-PV nodes)"] = Option(2, 0, 2);
o["Single Evasion Extension (PV nodes)"] = Option(2, 0, 2);
o["Single Evasion Extension (non-PV nodes)"] = Option(2, 0, 2);
o["Mate Threat Extension (PV nodes)"] = Option(0, 0, 2);
o["Mate Threat Extension (non-PV nodes)"] = Option(0, 0, 2);
o["Pawn Push to 7th Extension (PV nodes)"] = Option(1, 0, 2);
@@ -116,23 +100,18 @@ namespace {
o["Passed Pawn Extension (non-PV nodes)"] = Option(0, 0, 2);
o["Pawn Endgame Extension (PV nodes)"] = Option(2, 0, 2);
o["Pawn Endgame Extension (non-PV nodes)"] = Option(2, 0, 2);
o["Full Depth Moves (PV nodes)"] = Option(14, 1, 100);
o["Full Depth Moves (non-PV nodes)"] = Option(3, 1, 100);
o["Threat Depth"] = Option(5, 0, 100);
o["LSN filtering"] = Option(false);
o["LSN Time Margin (sec)"] = Option(4, 1, 10);
o["LSN Value Margin"] = Option(200, 100, 600);
o["Randomness"] = Option(0, 0, 10);
o["Minimum Split Depth"] = Option(4, 4, 7);
o["Maximum Number of Threads per Split Point"] = Option(5, 4, 8);
o["Threads"] = Option(1, 1, 8);
o["Hash"] = Option(32, 4, 4096);
o["Threads"] = Option(1, 1, MAX_THREADS);
o["Hash"] = Option(32, 4, 8192);
o["Clear Hash"] = Option(false, BUTTON);
o["New Game"] = Option(false, BUTTON);
o["Ponder"] = Option(true);
o["OwnBook"] = Option(true);
o["MultiPV"] = Option(1, 1, 500);
o["UCI_ShowCurrLine"] = Option(false);
o["UCI_Chess960"] = Option(false);
o["UCI_AnalyseMode"] = Option(false);
// Any option should know its name so to be easily printed
for (Options::iterator it = o.begin(); it != o.end(); ++it)
@@ -171,6 +150,18 @@ namespace {
return ret;
}
// Specialization for std::string where instruction 'ss >> ret;'
// would erroneusly tokenize a string with spaces.
template<>
string get_option_value<string>(const string& optionName) {
if (options.find(optionName) == options.end())
return string();
return options[optionName].currentValue;
}
}
////
@@ -185,24 +176,18 @@ void init_uci_options() {
load_defaults(options);
// Limit the default value of "Threads" to 7 even if we have 8 CPU cores.
// According to Ken Dail's tests, Glaurung plays much better with 7 than
// with 8 threads. This is weird, but it is probably difficult to find out
// why before I have a 8-core computer to experiment with myself.
// Set optimal value for parameter "Minimum Split Depth"
// according to number of available cores.
assert(options.find("Threads") != options.end());
assert(options.find("Minimum Split Depth") != options.end());
options["Threads"].defaultValue = stringify(Min(cpu_count(), 7));
options["Threads"].currentValue = stringify(Min(cpu_count(), 7));
Option& thr = options["Threads"];
Option& msd = options["Minimum Split Depth"];
// Increase the minimum split depth when the number of CPUs is big.
// It would probably be better to let this depend on the number of threads
// instead.
if (cpu_count() > 4)
{
options["Minimum Split Depth"].defaultValue = "6";
options["Minimum Split Depth"].currentValue = "6";
}
thr.defaultValue = thr.currentValue = stringify(cpu_count());
if (cpu_count() >= 8)
msd.defaultValue = msd.currentValue = stringify(7);
}
@@ -290,10 +275,28 @@ void set_option_value(const string& name, const string& value) {
else if (v == "false")
v = "0";
if (options.find(name) != options.end())
options[name].currentValue = v;
else
if (options.find(name) == options.end())
{
std::cout << "No such option: " << name << std::endl;
return;
}
// Normally it's up to the GUI to check for option's limits,
// but we could receive the new value directly from the user
// by teminal window. So let's check the bounds anyway.
Option& opt = options[name];
if (opt.type == CHECK && v != "0" && v != "1")
return;
else if (opt.type == SPIN)
{
int val = atoi(v.c_str());
if (val < opt.minValue || val > opt.maxValue)
return;
}
opt.currentValue = v;
}
@@ -313,7 +316,7 @@ void push_button(const string& buttonName) {
bool button_was_pressed(const string& buttonName) {
if (!get_option_value<bool>(buttonName))
return false;
return false;
set_option_value(buttonName, "false");
return true;
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+1 -1
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
+45 -5
View File
@@ -1,7 +1,7 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
Copyright (C) 2008-2009 Marco Costalba
Copyright (C) 2008-2010 Marco Costalba, Joona Kiiski, Tord Romstad
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -37,7 +37,9 @@ enum ValueType {
VALUE_TYPE_UPPER = 1, // Upper bound
VALUE_TYPE_LOWER = 2, // Lower bound
VALUE_TYPE_EXACT = 3, // Exact score
VALUE_TYPE_EVAL = 4 // Evaluation cache
VALUE_TYPE_EVAL = 4, // Evaluation cache
VALUE_TYPE_EV_UP = 5, // Evaluation cache for upper bound
VALUE_TYPE_EV_LO = 6 // Evaluation cache for lower bound
};
@@ -46,10 +48,49 @@ enum Value {
VALUE_KNOWN_WIN = 15000,
VALUE_MATE = 30000,
VALUE_INFINITE = 30001,
VALUE_NONE = 30002
VALUE_NONE = 30002,
VALUE_ENSURE_SIGNED = -1
};
/// Score enum keeps a midgame and an endgame value in a single
/// integer (enum), first LSB 16 bits are used to store endgame
/// value, while upper bits are used for midgame value.
// Compiler is free to choose the enum type as long as can keep
// its data, so ensure Score to be an integer type.
enum Score { ENSURE_32_BITS_SIZE_P = (1 << 16), ENSURE_32_BITS_SIZE_N = -(1 << 16)};
// Extracting the _signed_ lower and upper 16 bits it not so trivial
// because according to the standard a simple cast to short is
// implementation defined and so is a right shift of a signed integer.
inline Value mg_value(Score s) { return Value(((int(s) + 32768) & ~0xffff) / 0x10000); }
// Unfortunatly on Intel 64 bit we have a small speed regression, so use a faster code in
// this case, although not 100% standard compliant it seems to work for Intel and MSVC.
#if defined(IS_64BIT) && (!defined(__GNUC__) || defined(__INTEL_COMPILER))
inline Value eg_value(Score s) { return Value(int16_t(s & 0xffff)); }
#else
inline Value eg_value(Score s) { return Value((int)(unsigned(s) & 0x7fffu) - (int)(unsigned(s) & 0x8000u)); }
#endif
inline Score make_score(int mg, int eg) { return Score((mg << 16) + eg); }
inline Score operator-(Score s) { return Score(-int(s)); }
inline Score operator+(Score s1, Score s2) { return Score(int(s1) + int(s2)); }
inline Score operator-(Score s1, Score s2) { return Score(int(s1) - int(s2)); }
inline void operator+=(Score& s1, Score s2) { s1 = Score(int(s1) + int(s2)); }
inline void operator-=(Score& s1, Score s2) { s1 = Score(int(s1) - int(s2)); }
inline Score operator*(int i, Score s) { return Score(i * int(s)); }
// Division must be handled separately for each term
inline Score operator/(Score s, int i) { return make_score(mg_value(s) / i, eg_value(s) / i); }
// Only declared but not defined. We don't want to multiply two scores due to
// a very high risk of overflow. So user should explicitly convert to integer.
inline Score operator*(Score s1, Score s2);
////
//// Constants and variables
////
@@ -95,8 +136,7 @@ const Value PieceValueEndgame[17] = {
/// Bonus for having the side to move (modified by Joona Kiiski)
const Value TempoValueMidgame = Value(48);
const Value TempoValueEndgame = Value(21);
const Score TempoValue = make_score(48, 22);
////