Compare commits

..

289 Commits

Author SHA1 Message Date
Marco Costalba c5bb9b9da9 Stockfish DD
Stockfish bench signature is: 8596156
2013-11-29 10:23:14 +01:00
Kelly Wilson 67e5581e37 Add support for PPC 64bit on Linux
In particular Debian Linux-3.9.8-1- PPC64

No functional change.
2013-11-29 10:06:54 +01:00
Joona Kiiski 3ede7daab3 Generate Qsearch checks only at depth 0
An old idea retested at SPRT(0, 3) with 60+0.05 TC:
LLR: 2.95 (-2.94,2.94) [0.00,3.00]
Total: 98872 W: 15549 L: 15123 D: 68200

This is a very small elo increase patch so it really
stresses the limits of fishtest.

bench: 8596156
2013-11-24 10:26:49 +01:00
Marco Costalba dbd6156fce Revert previous fix
It seems to intorduce a regression when tested
with 3 threads at 15+0.05:

ELO: -2.26 +-2.2 (95%) LOS: 2.4%
Total: 30000 W: 4813 L: 5008 D: 20179

bench: 8331357
2013-11-19 07:20:50 +01:00
Hongzhi Cheng 691ed425ba Get correct excluded moves for split nodes
Tested setting FakeSplit to true and running

./stockfish bench 128 2

There is a different signature with and without
the patch so it affects functionality but
only in SMP case.

bench: 8331357
2013-11-18 16:41:49 +01:00
Marco Costalba c376ffce0f Revert previous patch
It seems a regression at 15+0.05:
ELO: -4.82 +-2.1 (95%) LOS: 0.0%
Total: 40000 W: 7181 L: 7736 D: 25083

bench: 8331357
2013-11-17 23:47:18 +01:00
Marco Costalba 917944e9c5 Fix an assert in SMP case
SMP case is very tricky and raises an assert in stage_moves():

assert(stage == KILLERS_S1 || stage == QUIETS_1_S1 || stage == QUIETS_2_S1)

So rewrite the code to just return moves[] when we are sure
we are in quiet moves stages.

Also rename stage_moves to quiet_moves to reflect that.

No functional change (but needs testing in SMP case)
2013-11-17 10:24:25 +01:00
Marco Costalba d9c7cad630 Retire quietsSearched[]
Use MovePicker moves[] to access already tried
quiet moves. A bit of care shall be taken
to avoid calling stage_moves() when we are still
at ttMove stage, because moves are yet to be
generated. Actually our staging move generation
makes this code a bit more tricky than what I'd
like, but removing an ausiliary redundant
array like quietsSearched[] is a good thing.

Idea by DiscoCheck

bench: 9355734
2013-11-17 09:51:04 +01:00
Marco Costalba 9763c69fa5 Simplify generate<EVASIONS>
Use the newly introduced LineBB[] to simplify this
super hot-path function.

Verified with perft we don't have any speed regression, although
the number of squares removed is less than before in case of
contact check.

Insipred by DiscoCheck implementation.

Perft numbers are the same, but we have an harmless functional
change due to reorder of moves, because now some illegal moves
are no more detected at generation time, but in the search.

bench: 8331357
2013-11-11 19:53:19 +01:00
Gregor Cramer 555d9a8711 Faster castling in Chess960 case
Only rook attackers has to be considered, all other attackers are
already handled in the lines above.

No functional change.
2013-11-11 15:55:08 +01:00
Joona Kiiski b9768b8bc5 Reintroduce gains
This seems a die hard idea :-)

Passed both short TC
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 17485 W: 3307 L: 3156 D: 11022

And long TC
LLR: 2.97 (-2.94,2.94) [0.00,6.00]
Total: 38181 W: 6002 L: 5729 D: 26450

bench: 8659830
2013-11-11 08:46:11 +01:00
Jörg Oster 4ef6b2c32a Remove opposed flag for doubled pawns
Actually, it is not used, as both arrays have the
same values. Some local tests in either direction
showed no improvement.

Also some minor corrections in the comments.

No functional change.
2013-11-10 17:21:59 +01:00
Marco Costalba 1d18647e73 Rename squares_aligned()
Rename to the shorter but still
clear aligned()

No functional change.
2013-11-10 17:14:46 +01:00
Marco Costalba a518d5d3ad Simplify squares_aligned()
Use newly introduced LineBB[]

No functional change.
2013-11-10 12:05:19 +01:00
Chris Caino 091aff0445 Evaluate mobility of pinned pieces exactly
Previously some squares could be "incorrectly" awarded
to a pinned piece.

e.g. in 3k4/1q6/3b4/3Q4/8/5K2/B7/8 b - - 0 1 the black
bishop get 4 squares too many and the white queen gets 6.

Passed both short TC.
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 4871 W: 934 L: 817 D: 3120

And long TC:
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 38968 W: 6113 L: 5837 D: 27018

bench: 9282549
2013-11-10 11:52:38 +01:00
Chris Caino 3ed86ed3f9 Remove RedundantMajor
But compensate by reducing rook and queen
value by 53 = (160 / 3)

Material imbalances are affected as follows:

       Red. Major   Rook  Queen  Total
QRR      +160      -2*53    -53     +1
QR       +160        -53    -53    +54
RR       +160      -2*53      0    +54
R           0        -53      0    -53
Q           0          0    -53    -53

so that the imbalance changes by at most 54 + 53 = 107 units.
This corresponds to appromximately 3.5cp in the final evaluation.

Verified with fixed number 40000 games at both short
and long TC it does not regress.

Short TC 15+0.05
ELO: 1.93 +-2.1 (95%) LOS: 96.6%
Total: 40000 W: 7520 L: 7298 D: 25182

Long TC 60+0.05
ELO: -0.33 +-1.9 (95%) LOS: 36.5%
Total: 39663 W: 6067 L: 6105 D: 27491

bench: 6703846
2013-11-10 09:48:49 +01:00
Marco Costalba e4d34e1815 Fix printing of incorrect PV in some cases
As, Gary (that analyzed the bug) says:

SF does not print a PV when the original best move fails low,
we hit our time allowance, and stop the search.  The output from
the SF search is below.  It was failing low on Ne1 at depth 34.
Then, we get bestmove Qd3, but no PV change.

info depth 34 seldepth 45 score cp 38 upperbound nodes 483484489 nps 15464575 time 31264 multipv 1 pv f3e1 h5h4 e1d3 h4g3 f2g3 a6f6 f1f6 e7f6 d1a4 f6e7 a1f1 d8f8 a4b3 b7b6 b3c2 f7f6 c2a4 h3g5 b2b3 g5f7 a4c6 f7d6 h1g2 f6f5 e4f5 d6f5
info depth 34 seldepth 45 score cp 38 upperbound nodes 483484489 nps 15464575 time 31264 multipv 1 pv f3e1 h5h4 e1d3 h4g3 f2g3 a6f6 f1f6 e7f6 d1a4 f6e7 a1f1 d8f8 a4b3 b7b6 b3c2 f7f6 c2a4 h3g5 b2b3 g5f7 a4c6 f7d6 h1g2 f6f5 e4f5 d6f5
info depth 34 seldepth 47 score cp 30 upperbound nodes 2112334132 nps 17255517 time 122415 multipv 1 pv f3e1 h5h4 d1a4 a6f6 e1d3 d8f8 a4c2 h4g3 f2g3 f6f1 a1f1 h7g8 b2b3 f7f6 a3a4 b7b6
info depth 34 seldepth 47 score cp 30 upperbound nodes 2112334132 nps 17255517 time 122415 multipv 1 pv f3e1 h5h4 d1a4 a6f6 e1d3 d8f8 a4c2 h4g3 f2g3 f6f1 a1f1 h7g8 b2b3 f7f6 a3a4 b7b6
info nodes 18235667001 time 969824
bestmove e2d3 ponder c8d7

Looking at the code, if we hit Signals.stop, we return from id_loop
before printing any PV.  It is possible for us to have resorted the
RootMove list though, which will change the move that is actually
played.

No functional change.
2013-11-09 19:05:43 +01:00
Marco Costalba 42caebfaa5 Fix compile in debug mode
No functional change.
2013-11-09 18:41:51 +01:00
Lucas Braesch eed508b444 Futility pruning simplification
1/ eval margin and gains removed:
16bit are now free on TT entries, due to the removal of eval margin. may be useful
in the future :) gains removed: use instead by Value(128). search() and qsearch()
are now consistent in this regard.

2/ futility_margin()
linear formula instead of complex (log(depth), movecount) formula.

3/ unify pre & post futility pruning
pre futility pruning used depth < 7 plies, while post futility pruning used
depth < 4 plies. Now it's always depth < 7.

Tested with fixed number of games both at short TC:
ELO: 0.82 +-2.1 (95%) LOS: 77.3%
Total: 40000 W: 7939 L: 7845 D: 24216

And long TC
ELO: 0.59 +-2.0 (95%) LOS: 71.9%
Total: 40000 W: 6876 L: 6808 D: 26316

bench 7243575
2013-11-09 10:17:27 +01:00
Marco Costalba 343544f3f7 Revert "Retire eval margin and gains"
This reverts commit ecd07e51d0.

Patch was incorrect and partial. It will be reapplied in
the correct form.

bench: 9189063
2013-11-07 22:32:13 +01:00
Gary Linscott 13d1f0ae43 Restrict mobility of pinned pieces
Passed both short TC:
LLR: 3.00 (-2.94,2.94) [-1.50,4.50]
Total: 54342 W: 10950 L: 10692 D: 32700

And long TC:
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 61976 W: 10654 L: 10251 D: 41071

This patch introduces a slowdown of 3.5 % !!!!!

bench: 7911558
2013-11-07 22:26:03 +01:00
Lucas Braesch ecd07e51d0 Retire eval margin and gains
1/ eval margin and gains removed:
 - gains removed by Value(128): search() and qsearch() now behave consistently!

2/ futility_margin()
 - testing showed that there is no added value in this weird (log(depth), movecount)
   formula, and a much simpler linear formula is just as good. In fact, it is most
   likely better, as it is not yet optimally tuned.
 - the new simplified formula also means we get rid of FutilityMargins[], its
   initialization code, and more importantly ss->futilityMoveCount, and the hacky
   code that updates it throughout the search().
 - the current formula gives negative futility margins, and there is a hidden interaction
   between the move coutn pruning formula and the futility margin one: what happens is
   that MCP is supposed to be triggered before we use the non-sensical negative futility
   margins.

3/ unify pre & post futility pruning
 - pre futility pruning (what SF calls value based pruning) used depth < 7 plies,
   while post futility pruning (what SF calls static null move pruning) used depth < 4 plies.
 - also the condition depth < 7 in pre futility pruning was not obvious, and it seemd
   to be depth < 16 (futility_margin() returns an infinite value when depth >= 7).

Tested with fixed number of games both at short TC:
ELO: 0.82 +-2.1 (95%) LOS: 77.3%
Total: 40000 W: 7939 L: 7845 D: 24216

And long TC
ELO: 0.59 +-2.0 (95%) LOS: 71.9%
Total: 40000 W: 6876 L: 6808 D: 26316

bench: 10206576
2013-11-07 19:46:51 +01:00
Chris Caino 52ae0efccf Two more parameters eliminated
RedundantRook and RedundantQueen replaced by simple
variable RedundantMajor. Also the SameColor coefficient
for Queen<->Queen has been set by definition to 0.

The remaining 5 parameters:

LinearCoefficients[ROOK]
LinearCoefficients[QUEEN]
QuadraticCoefficientsSameColor[ROOK][ROOK]
QuadraticCoefficientsSameColor[QUEEN][ROOK]
RedundantMajor

are sufficient to equate the material imbalances for the
5 common material configurations of R, RR, Q, QR and QRR
to any desired values simultaneously.

With the chosen parameters there should be no functional
change unless one side has more than 2 rooks or more
than 1 queen. For example bench from the start position
using the commands:

./stockfish
go depth 16

produces identical output except for one extra node
in the last iteration.

bench: 8198094
2013-11-07 19:20:24 +01:00
Chris Caino 53c04c0429 Zero more redundant coefficients
Coefficients for Bishop<->BishopPair and Bishop<->Bishop
are also pretty much redundant. By altering the values
in LinearCoefficients[] these coefficients can be zeroed
without changing the imbalance calculations in any position
with less than 3 bishops for one side.

bench: 7995098
2013-11-05 20:08:39 +01:00
Chris Caino 1064288b38 Zero redundant material imbalance terms
First coefficient in the SameColor array does an
equivalent job when folded into the LinearCoefficients
array.

All of the diagonal terms in the OppositeColor array
are redundant due to cancellation.

No functional change.
2013-11-05 20:08:39 +01:00
Joona Kiiski d34bb889b1 Test Easy Move if no BestMoveChanges
In case we find a very good move after a
troubled start, we don't return immediately
anymore.

Tested directly at long TC where it passed:
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 13910 W: 2397 L: 2228 D: 9285

bench: 7995098
2013-11-02 11:34:42 +01:00
Marco Costalba a3a0df92a3 Set timer to a fixed interval
And remove a complex (and broken) formula.

Indeed previous code was broken in case of TC with big
time increments where available_time() was too similar
to total time yielding to many time losses, so for instance:

go wtime 2600 winc 2600
info nodes 4432770 time 2601 <-- time forfeit!

maximum search time = 2530 ms
available_time = 2300 ms

For a reference and further details see:

https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/dCPAvQDcm2E

Speed tested with bench disabling timer alltogheter vs timer set at
max resolution, showed we have no speed regressions both in single
core and when using all physical cores.

No functional change.
2013-11-01 08:56:15 +01:00
Ralph Ster e8f9447b11 Use a formula for chain membership bonus
Passed both short TC:
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 5087 W: 1072 L: 951 D: 3064

And long TC:
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 28620 W: 5042 L: 4798 D: 18780

bench: 7995098
2013-10-31 06:13:30 +01:00
Marco Costalba 3cc47edf62 Tweak bishop pair and knight weight
A combo of two patches that failed SPRT with score
higher than 50% but togheter they succeed:

SPRT at 60+0.05
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 7312 W: 1276 L: 1139 D: 4897

bench: 8029334
2013-10-28 19:31:25 +01:00
Matthew Sullivan d454cd4216 Fix divide by zero bug in late game
If the game got late enough that move_importance(currentPly) * slowMover / 100
rounds to 0, then we ended up dividing 0 by 0 when only looking 1 move ahead.

This apparently caused the search to almost immediately abort and Stockfish
would blunder in long games. So convert thisMoveImportance to a double.

No functional change.
2013-10-27 08:03:58 +01:00
Marco Costalba 48f38f3092 Retire mirror()
Inline the only caller site.

No functional change.
2013-10-24 20:40:26 +02:00
Marco Costalba 281472e50e Prefer file_bb() to FileBB[]
No functional change.
2013-10-24 20:34:23 +02:00
Jörg Oster f011a5af11 Penalty for Knight when enemy pawns are few
This seems more a material imbalance topic,
anyhow test is good and so patch is applied
as is.

Passed both short TC:
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 17391 W: 3548 L: 3393 D: 10450

And long TC:
LLR: 3.00 (-2.94,2.94) [0.00,6.00]
Total: 34660 W: 5972 L: 5700 D: 22988

bench: 8291883
2013-10-24 20:11:33 +02:00
Marco Costalba 67b0da83da Further smplify pawn endgames
Dumb down a bit the code and trade some possible
speed (but this is far from hot path anyhow) for
some added readability for the layman.

No functional change.
2013-10-23 19:56:35 +02:00
Chris Caino 3674f18b97 Use flip_sq idea in endgame.cpp
The normalising transformation is computed all at
once by the helper function get_flip_sq and then
applied immediately to the relevant squares as soon
as they are loaded from the position class.

bench: 8350690
2013-10-23 16:00:49 +02:00
Chris Caino 72f7282ad4 Simplify futility move count formula
Simpler formula but introduces some slight changes if d >= 10

Original code grows like  0.225 * d^1.8
New code grows like       0.222 * d^1.8

Full list of values:

d old new diff
--------------
0 2 2 0
1 2 2 0
2 3 3 0
3 4 4 0
4 5 5 0
5 6 6 0
6 7 7 0
7 9 9 0
8 11 11 0
9 13 13 0
10 15 16 1
11 18 19 1
12 21 21 0
13 24 24 0
14 27 28 1
15 31 31 0
16 35 35 0
17 39 38 -1
18 42 42 0
19 47 46 -1
20 51 51 0
21 55 55 0
22 60 60 0
23 65 65 0
24 70 70 0
25 75 75 0
26 81 80 -1
27 87 86 -1
28 92 91 -1
29 98 97 -1
30 104 103 -1
31 111 109 -2

Test code:

int main() {

  for(int d=0; d<32; d++)
  {
     int a = int(3 + 0.3 * pow(double(d), 1.8)) * 3/4 + (2 < d && d < 5);
     int b = int(2.4 + 0.222 * pow(d + 0.0, 1.8));

     std::cout << d << " " << a << " " << b << " " << b-a << std::endl;
  }

  return 0;
}

bench: 8350690
2013-10-22 23:09:40 +02:00
Chris Caino fbfce2132a Simplify futility margins formula
New formula mathces the old formula until d = 45

Test code:

int main() {

  for(int d=1; d<=45; d++)
  {
     int a = int(log(double(d * d) / 2) / log(2.0) + 1.001);
     int b = int(2.9 * log(double(d)));

     if (a != b) std::cout << d << std::endl;
  }

  return 0;
}

bench: 8455956
2013-10-22 23:06:06 +02:00
Marco Costalba 2c825294ec Tweak again chain pawn bonus
This is the first chain bonus version
from Ralph that also passed both

Short TC:
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 23460 W: 4727 L: 4556 D: 14177

And long TC:
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 31858 W: 5497 L: 5240 D: 21121

And performed better against current
committed version, always at 60secs:

LLR: -2.94 (-2.94,2.94) [-3.00,3.00]
Total: 26301 W: 4477 L: 4580 D: 17244

This test was done by Leonid.

bench: 8455956
2013-10-22 17:47:16 +02:00
Marco Costalba f86d2aee29 Re-add "Further increase safe checks bonus"
After 40K games at 60 secs, result is still
not clear, but not a regression against SF 4

After
ELO: 50.11 +-2.1 (95%) LOS: 100.0%
Total: 40000 W: 10547 L: 4817 D: 24636

Before
ELO: 49.51 +-2.1 (95%) LOS: 100.0%
Total: 40000 W: 10483 L: 4821 D: 24696

So re-apply the patch to avoid to
special-case this one.

bench: 7403882
2013-10-22 17:33:11 +02:00
Marco Costalba 35ea39bed2 Restore behaviour after count<ALL_PIECES> fix
Because pos.count<ALL_PIECES>(Us) was always zero,
rewrite the formula as if this would still be
the case.

bench: 8510004
2013-10-22 17:27:58 +02:00
Ralph Stößer 97015afce8 Further improve chain pawn evaluation
Passed both short TC:
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 28299 W: 5854 L: 5667 D: 16778

And long TC:
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 9738 W: 1797 L: 1644 D: 6297

bench: 9294116
2013-10-20 23:41:40 +02:00
Marco Costalba f22a63ce67 Fix pos.count<ALL_PIECES>()
It was never updated !

Currently it only affects evaluate_passed_pawns()
and in particularly the rule to increase the bonus
if we have more non-pawn pieces. We could simply use
popcount() instead and avoid the little slowdown
in put_piece() and remove_piece(), but this would
leave a very subtle and tricky hole where people
are forced to remember that pos.count<ALL_PIECES>()
does not work. This is not obvious and so dangerous.

Thanks to Ronald de Man for spotting this.

bench: 7931424
2013-10-20 23:36:46 +02:00
Marco Costalba c08e7419a0 Fix build on Intel compiler
Due to a strange issue (bug?) the ternary
operator does not return a BitCountType for
icc, so revert to the expression.

The same patch was already applied in
9749f1f14c

Thanks to NssY Wanyonyi for pointing out
this.

No functional change.
2013-10-20 23:19:08 +02:00
Marco Costalba 67f91bc5ea Revert "Further increase safe checks bonus"
This reverts commit 4bc2374450 for
two reasons.

First regression testing shows almost equal
score:

Before the patch:
ELO: 49.75 +-2.5 (95%) LOS: 100.0%
Total: 27205 W: 7113 L: 3244 D: 16848

After the patch:
ELO: 48.87 +-2.9 (95%) LOS: 100.0%
Total: 20860 W: 5478 L: 2563 D: 12819

Second, and more sensible to me, this patch
increases safe check bonuses to 4 times their
original value (!) and considering:

- Values were already well tuned

- Values are highly critical

- King safety is highly critical, very TC
  dependent and very difficult to test

- Our testing coverage is partial (self-testing,
  blitz times)

I think is better to be safe than sorry and so
I revert the patch.

bench: 8440524
2013-10-20 10:04:43 +02:00
Ralph Stößer 4bc2374450 Further increase safe checks bonus
Passed both short TC:
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 10466 W: 2087 L: 1953 D: 6426

And long TC:
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 26334 W: 4540 L: 4310 D: 17484

And also proved stronger than a slightly
different patch, also succesful against master:

https://github.com/mcostalba/Stockfish/commit/dc6830a3b4ed12

But losing against current one in a match
at 60secs with SPRT [-3, 3]:

LLR: -2.96 (-2.94,2.94) [-3.00,3.00]
Total: 44484 W: 7360 L: 7463 D: 29661

bench: 9160831
2013-10-19 12:19:36 +02:00
Marco Costalba f5e872a0e3 Some evaluation code reshuffle
No functional change.
2013-10-18 09:49:38 -07:00
Jörg Oster 25cb851f8a Score chain pawn also by rank
Use the (rescaled) CandidatePassed[] table
that is already rank based.

Passed both short TC
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 11048 W: 2272 L: 2135 D: 6641

And long TC
LLR: 2.97 (-2.94,2.94) [0.00,6.00]
Total: 4116 W: 769 L: 645 D: 2702

bench: 8440524
2013-10-18 10:27:55 +02:00
Chris Caino 3cddb0c076 Simplification of KPsK function
Also the drawing criteria has been slightly loosened.
It now detects a draw if the king is ahead of all the
pawns and on the same file or the adjacent file.

bench: 7700683
2013-10-15 07:36:01 +02:00
Chris Caino 2bf18bfc63 Bug fix for KQKRPs endgame
This lost position 8/8/3q4/8/5k2/2P1R3/2K2P2/8 w - - 0 1
was previously evaluated as a draw.

The king and rook need to be correctly placed with
respect to the _same_ pawn.

(Note also that the check for the pawn being on RANK_2
in the old version is redundant: it must be on RANK_2 if
it hopes to protect a rook on RANK_3)

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-10-15 07:34:29 +02:00
Ralph Stößer 5aeb907fa1 Double king safety weights
Good both at short TC:
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 5448 W: 1133 L: 1012 D: 3303

And at long TC:
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 40509 W: 6836 L: 6541 D: 27132

bench: 7700683
2013-10-14 23:24:29 +02:00
Chris Caino 0c68971c13 Remove a drawing rule from KBPsK function
The rule can be incorrect if the attacking king is
well placed e.g. 8/6K1/8/8/7k/1B6/7P/8 w - - 0 1

bench: 8279065
2013-10-14 19:55:07 +02:00
Marco Costalba d9be00342c Massive stronger/weaker renming
No functional change.
2013-10-14 19:38:08 +02:00
Chris Caino 027d85e82a Add helper function verify_material
Allows to remove a lot of assert code in endgames.

No functional change.
2013-10-14 19:23:57 +02:00
ceebo 3bc3c069f1 Add some knowledge for KRPKB endgame
bench: 8279065
2013-10-14 07:44:02 +02:00
ceebo e9366fa155 Improve KBPsK endgame
Better endgame with bishop and blocked g-pawn

bench: 8279065
2013-10-14 07:44:01 +02:00
Marco Costalba 549b5c478f Remove unuseful optimization in RKISS
Don't need a struct here. Speed test shows
result is teh same. Moreover RKISS is used
mainly at startup to compute magics, so
prefer to keep it simple...RKISS ;-)

Also some assorted triviality while there.

No functional change.
2013-10-13 03:35:17 -07:00
Joona Kiiski b15e148b5e Smoother transition for LMR
Passed both short TC:
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 12376 W: 2596 L: 2454 D: 7326

And long TC:
LLR: 2.97 (-2.94,2.94) [0.00,6.00]
Total: 14798 W: 2584 L: 2409 D: 9805

bench: 8279065
2013-10-09 19:13:41 +02:00
Uri Blass bb83a417cb Increase slowmover and reduce instability
These two changes go in opposite directions and it
seems that the combination is stronger than original.

Here are the positive tests at various TC:

15+0.05
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 24561 W: 4946 L: 4772 D: 14843

60+0.05
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 15259 W: 2598 L: 2423 D: 10238

40/30
LLR: 2.96 (-2.94,2.94) [-3.00,3.00]
Total: 2570 W: 527 L: 422 D: 1621

Unfortunately there is also a bad result
with one sec time increment that needs
to be further investigated:

12+1
LLR: -2.97 (-2.94,2.94) [-3.00,3.00]
Total: 2694 W: 438 L: 543 D: 1713

bench: 8340585
2013-10-08 21:24:21 +02:00
Lucas Braesch 984ee9d05b Use TT refined value to stand pat
Passed both short TC:
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17811 W: 3520 L: 3366 D: 10925

And long TC:
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 30255 W: 5070 L: 4825 D: 20360

bench: 8340585
2013-10-08 19:51:08 +02:00
Lucas Braesch a0cc15ccbc Use double everywhere
Rationale:

- Speed of double and float is about the same (not on the hot path anyway)

- Double makes code prettier (no need to write 1.0f, just 1.0)

- Only practical advantage of float is to use less memory, but since we never
  store large arrays of double, we don't care.

No functional change.
2013-10-05 18:12:52 +02:00
Lucas Braesch 7f142d6817 Use prefix operators wherever possible
No functional change.
2013-10-05 18:10:43 +02:00
Marco Costalba bd1c3ed7e3 Add more depth/positions to bench
Increase bench default depth from 12 to 13 and
add 15 new endgame positions to have broader
coverage and also more reliable nps calulcation
used for fishtest framework.

Due to the new endgame positions, where nps is higher,
the total nps is increased of about 15%.

Thanks to Lucas and Jörg for the suggestions.

No functional change, but bench number is now:

bench: 8336338
2013-09-29 09:43:10 +02:00
Marco Costalba cca34e234c Drop 'is' prefix from query functions
Most but not all.

No functional change.
2013-09-28 06:47:59 -07:00
Marco Costalba ed95ad1c0e Fix build on Mac OS X
For some users -stack_size,0x4000 does not work,
so revert for now.

osX 10.6.8
gcc version 4.7.3 (MacPorts gcc47 4.7.3_2)

g++: error: unrecognized command line option '-stack_size,0x4000'
make[2]: *** [stockfish] Error 1
make[1]: *** [gcc-profile-make] Error 2
make: *** [profile-build] Error 2

No functional change.
2013-09-28 04:16:16 -07:00
Marco Costalba c65d67feb5 Revert "Use a per-thread array"
This reverts commit 800410eef1 and instead increases
stack size.

I went through the old emails with Daylen that reported the
crash issue on Mac OS X and was fixed by 0049d3f337.

It was reported default stack size for a thread in Mac OS X is 8
megabytes while the patch that we are reverting allows to reduce
stack size at max of about 217KB, so the reason for the crash was
only marginal in MAX_MOVES value. On those emails Daylen also
hinted how to increase stack size for Mac OS X to 16MB.

So prefer to increase stack size to 16MB instad of re-inventing
the wheel and do our home grown stack as we did with the patch
that we are now reverting (it will remain anyhow in git history
for documentation purposes).

No functional change.
2013-09-28 10:10:51 +02:00
Lucas Braesch bc6faf633e Simplify extensions
Unify extensions between PV and not PV nodes
and remove all but check extensions.

This is a simplification so tested at fixed number
of games where proved to not regress.

About 45k games at 15+0.05
ELO: 1.23 +-2.0 (95%) LOS: 88.5%
Total: 45643 W: 9107 L: 8946 D: 27590

About 45k games at 60+0.05
ELO: 1.07 +-1.8 (95%) LOS: 87.8%
Total: 46786 W: 7728 L: 7584 D: 31474

bench: 3172206
2013-09-28 09:54:22 +02:00
Reuven Peleg 8d1c1074d5 Simplify tte use condition
No functional change.
2013-09-27 09:40:48 +02:00
Raminder Singh e654209211 Fix best move lookup bug
If the uci option 'Best Book Move' is set to true the lookup still
returns a move at random instead of the move with the highest
weight.

No functional change.
2013-09-27 09:04:24 +02:00
Marco Costalba b742a3f29a Increase MAX_MOVES to 256
This should be enough for any legal position, even
the handcrafted ones, like the one presented by Reuven:

1Q5R/4Q1K1/B1Q5/B4Q2/N2Q4/pQ4Q1/pn2Q3/krQ4R w - -

Where currently we crash. This reverts the patch
0049d3f337 of 8/4/2012 where stack
was shrinked due to crashes while in deep analysys.

No functional change.
2013-09-27 08:59:03 +02:00
Marco Costalba 800410eef1 Use a per-thread array for generated moves
This greately reduces stack usage and is a
prerequisite for next patch.

Verified with 40K games both in single and SMP
case that there are no regressions.

No functional change.
2013-09-27 08:44:36 +02:00
Jean-Francois Romang 7b2cda95d9 Update disabled warnings for Intel compiler
No functional change.
2013-09-23 08:11:43 +02:00
Jean-Francois Romang c2cefa6de0 Replace -O3 with -fast for intel compiler
No functional change.
2013-09-23 08:11:14 +02:00
Ralph Stößer d7f5f15d69 Reduce negative quiets by ONE_PLY / 2
Passed both short TC:
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 3402 W: 708 L: 593 D: 2101

And long TC:
LLR: 2.93 (-2.94,2.94) [0.00,6.00]
Total: 23379 W: 3972 L: 3759 D: 15648

bench: 3529630
2013-09-23 08:08:52 +02:00
Marco Costalba 84906b83ec Final time management setup
This is an even safer setup proposed and tested
by Alexandre Meirelles.

Regression testing of 40K games at 10+0.05 show
result is stable both against current master:

ELO: -0.29 +-2.2 (95%) LOS: 39.7%
Total: 40000 W: 8010 L: 8043 D: 23947

and again original master (the one with smallest
time parameters):

ELO: 1.71 +-2.2 (95%) LOS: 93.8%
Total: 40000 W: 8325 L: 8128 D: 23547

Alexandre verified with LittleBlitzer time losses are
greately reduced with this setup:

Games Completed = 2100 of 3000 (Avg game length = 35.745 sec)

Settings = RR/128MB/15000ms+50ms/M 1000cp for 12 moves, D 150 moves/
Time = 39200 sec elapsed, 16800 sec remaining
 1.  Stockfish 190913             1091.5/2100    803-720-577      (L: m=313 t=1 i=0 a=406)    (D: r=278 i=91 f=136 s=8 a=64)    (tpm=212.5 d=14.75 nps=925427)
 2.  Houdini 2.0 w32              1008.5/2100    720-803-577      (L: m=250 t=299 i=0 a=254)    (D: r=278 i=91 f=136 s=8 a=64)    (tpm=204.1 d=12.04 nps=1326351)

No functional change.
2013-09-23 07:59:51 +02:00
Marco Costalba 274079990a Increase Emergency Move Time to 20
Goes in the direction of avoiding time losses and seems
equivalent after almost 40K games at super fast TC of 10+0.05

ELO: 2.61 +-2.2 (95%) LOS: 99.1%
Total: 39869 W: 8258 L: 7959 D: 23652

No functional change.
2013-09-19 07:26:36 +02:00
Marco Costalba 10cb19d534 Increase Emergency Move Time to 10
Goes in the direction of avoiding time losses and seems
equivalent after almost 40K games at super fast TC of 10+0.05

ELO: 2.41 +-2.3 (95%) LOS: 98.1%
Total: 37222 W: 7843 L: 7585 D: 21794

No functional change.
2013-09-17 16:32:39 +02:00
Marco Costalba d50b33cacd Fix a silly unstoppable eval bug
The logic is broken for black side because we get more bonus
for pawn in 7th rank than for pawn in 2nd rank!

Spotted by Reuven Peleg

bench:3884409
2013-09-16 23:59:37 +02:00
Joona Kiiski 77b5ee0117 Fix time parameters for blitz games
The ideal setting for super-blitz might be something like:

    "Emergency Base Time" = 50
    "Emergency Move Time" = 5

This would give a total emergency time buffer of:

    50 + 40 * 5 = 250 ms

This setup replaces the previous half cooked hack
"Don't blunder under extreme time pressure".

Test results are very good at super blitz, but keep good even
at 60 secs.

At 5+0.05
ELO: 24.30 +-2.4 (95%) LOS: 100.0%
Total: 37802 W: 10060 L: 7420 D: 20322

At 15+0.05
ELO: 13.41 +-2.9 (95%) LOS: 100.0%
Total: 22271 W: 4853 L: 3994 D: 13424

At 60+0.05
ELO: 5.30 +-3.2 (95%) LOS: 99.9%
Total: 16000 W: 2897 L: 2653 D: 10450

No functional change.
2013-09-16 09:07:47 +02:00
Marco Costalba af750bd2ef Rewrite unstoppable pawns evaluation
Instead of current code, give a bonus according to the frontmost
square among candidate + passed pawns.

This is a big simplification that removes a lot of accurate code
substituting it with a statistically based one using the common
'bonus' scheme, leaving to the search to sort out the details.

Results are equivalent but code is much less and, as an added bonus,
we now store candidates bitboard in pawns hash and allow this
info to be used in evaluation. This paves the way to possible
candidate pawns evaluations together with all the other pieces,
as we do for passed.

Patch passed short TC
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 16927 W: 3462 L: 3308 D: 10157

Then failed (quite quickly) at long TC
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 8451 W: 1386 L: 1448 D: 5617

But when ran with a conclusive 40K fixed games at 60 secs it proved
almost equivalent to original one.

ELO: 1.08 +-2.0 (95%) LOS: 85.8%
Total: 40000 W: 6739 L: 6615 D: 26646

bench: 3884003
2013-09-16 08:57:37 +02:00
Reuven Peleg 21cbfafc03 Code style at passed pawn eval
No functional change.
2013-09-15 21:49:06 +02:00
Reuven Peleg d3947b2f3e Nicer operator declerations
No functional change.
2013-09-15 21:45:18 +02:00
Kojirion a71209868b Use pre-increment also for native types
Now that we use pre-increment on enums, it
make sense, for code style uniformity, to
swith to pre-increment also for native types,
although there is no speed difference.

No functional change.
2013-09-15 09:17:21 +02:00
Marco Costalba 7a1ff6d8ff Fix operator++ definition
ENABLE_OPERATORS_ON has incorrect definitions of
post-increment and post-decrement operators.

In particularly the returned value is the variable
already incremented/decremented, while instead they
should return the variable _before_ inc/dec.

This has no real effect because are only used in loops
and where the returned value is never used, neverthless
it is wrong. The fix would be to copy the variable to a
dummy, then inc/dec the variable, then return the dummy.

So instead, rename to pre-increment that can be implemented
without the dummy, actually the current implementation
it is already the correct pre-increment, with the only change
to return a reference (an l-value) and not a copy, so
to properly mimic the pre-increment on native integers.

Spotted by Kojirion.

No functional change.
2013-09-15 09:09:06 +02:00
Marco Costalba 82f6779c2e Don't blunder under extreme time pressure
We always attempt to keep at least this emergencyBaseTime
at clock. But if available time is very low it means that
we will force ourself to play immediately to satisfy the
emergencyBaseTime constrain and so leading to blunders.

Patch is good at short and very short TC (15secs and 5secs respectively)
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 26590 W: 5426 L: 5245 D: 15919

LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 5767 W: 1397 L: 1268 D: 3102

Instead seems has no influence at longer TC (60 secs)
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 79862 W: 13623 L: 13339 D: 52900

So it is committed to have a broader testing but is
to be consider still EXPERIMENTAL and can be reverted
easily.

No functional change.
2013-09-15 07:59:09 +02:00
Marco Costalba 3abccdc82d Move classify_leaf() to c'tor in bitbases
No functional change.
2013-09-14 13:08:37 +02:00
Marco Costalba 8d6d0223bf Small touches to bitbase.cpp
Inspired by Lucas's code:

https://github.com/lucasart/chess/blob/master/src/kpk.cc

No functional change.
2013-09-14 11:19:12 +02:00
Reuven Peleg fc17d0de77 Increase passed bonus for having more pieces
Passed both short TC:
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 36463 W: 7575 L: 7365 D: 21523

And long TC:
LLR: 2.97 (-2.94,2.94) [0.00,6.00]
Total: 2953 W: 564 L: 446 D: 1943

bench: 3846852
2013-09-13 18:08:22 +02:00
Marco Costalba 27f2ce8f6e Revert "Move draw by material check"
Possible regression

bench: 4554579
2013-09-12 08:44:11 +02:00
Marco Costalba 45b0aea875 Revert "Fix random moves when time < 10ms"
Possible regression.

No functional change.
2013-09-12 08:38:19 +02:00
Marco Costalba 4803d5772c Extend checks more when below alpha
Passed both short TC:
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 8739 W: 1830 L: 1698 D: 5211

And long TC:
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 6716 W: 1238 L: 1101 D: 4377

bench: 4554576
2013-09-11 19:15:28 +02:00
Uri Blass 738c5595ad Extend checks more in losing positions
Passed both short TC:
LLR: 2.98 (-2.94,2.94) [-1.50,4.50]
Total: 3974 W: 860 L: 741 D: 2373

And long TC:
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 16807 W: 2917 L: 2733 D: 11157

bench: 3767999
2013-09-11 09:15:47 +02:00
Reuven Peleg bebd6e16f6 Simplify unstoppable pawns evaluation
No functional change
2013-09-10 23:02:05 +02:00
Marco Costalba 49e110c52b Fix random moves when time < 10ms
In case we have less then 10ms to think as soon as
we wake up the timer, it immediately fires and calls
check_time() where due to condition:

elapsed > TimeMgr.maximum_time() - 2 * TimerResolution

the stop flag is set and search returns immediately, without
actually search anything.

Here the somewhat hacky fix is to start the timer after
at least one iteration as been completed.

No functional change.
2013-09-10 21:23:20 +02:00
Reuven Peleg 4d90aeb0ab More readable space mask
No functional change.
2013-09-10 19:18:10 +02:00
homoSapiensSapiens 03cd049c68 Change condition to use relative rank
No functional change
2013-09-10 00:47:31 +02:00
Marco Costalba 6ab8b9b6c6 Fix some comments in position.cpp
No functional change.
2013-09-08 06:28:53 -07:00
Marco Costalba 490f67a3f8 Move draw by material check
It is more natural to test this case among
others material distributions.

No functional change.
2013-09-08 06:11:35 -07:00
Marco Costalba 0515ad0fb0 Remove unreachable values in mobility table
The possible maximum mobility cardinality (plus one in case of
zero squares available) is:

- Knights: max. 8  squares -> max. 9  entries
- Bishops: max. 13 squares -> max. 14 entries
- Rooks:   max. 14 squares -> max. 15 entries
- Queen:   max. 27 squares -> max. 28 entries

So remove the extra entries in the table.

Spotted by Dariusz Orzechowski.

No functional change.
2013-09-07 18:25:24 +02:00
Marco Costalba bf51db2526 Fix warning: double to float truncation
MSVC 2013 says:
warning C4305: '*=' : truncation from 'double' to 'float'

No functional change.
2013-09-07 12:30:44 +02:00
Lucas Braesch 59702aca0d Singular extension at 8 plies also for PV nodes
Passed both short TC:
LLR: 2.96 (-2.94,2.94)
Total: 11451 W: 2455 L: 2282 D: 6714

And long TC
LLR: 2.96 (-2.94,2.94)
Total: 15813 W: 2907 L: 2723 D: 10183

bench: 3864419
2013-09-07 09:34:22 +02:00
Lucas Braesch c86eee3918 Union of 2 changes
Union of

- LMR >= 3 plies from Gary tests.stockfishchess.org/tests/view/522522960ebc595d328fcafd

- allows() tweak from Reuven tests.stockfishchess.org/tests/view/5225fa1c0ebc595d328fcb53

Both passed Step I and failed Step II.

Instead this union passed both short TC:
LLR: 2.95 (-2.94,2.94)
Total: 14525 W: 3063 L: 2874 D: 8588

And long TC
LLR: 2.94 (-2.94,2.94)
Total: 31075 W: 5566 L: 5308 D: 20201

bench: 4238160
2013-09-07 09:25:45 +02:00
Lucas Braesch 10b53e1c5e Do not prune useless checks in QS
Passed both SPRT tests in "simplification mode", so with
elo0: -3.00 alpha: 0.05 elo1: 3.00 beta: 0.05

Short TC:
LLR: 2.96 (-2.94,2.94)
Total: 6243 W: 1302 L: 1195 D: 3746

Long TC
LLR: 2.96 (-2.94,2.94)
Total: 22972 W: 4124 L: 4020 D: 14828

bench: 4633330
2013-09-05 18:50:16 +02:00
Marco Costalba a30d3571ca Revert "Fix check for bishop pair in material imbalance"
Idea is sound but implementation is partial. Ryan and Joona noticed that
    we leave an hole in material table. Also we got another report by an user
    of an odd behaviour. Namely, if you start stockfish and from the prompt
    give 'bench' you get 3453941, then if you run again bench you get 3453940.

    The reason is that two different positions with the same number of pieces,
    but one with a bishop pair and another without have the same material key.
    But after Eelco patch also different material imbalance and this yields
    to this issue.

    Restesting at long TC shows the patch does not really contribute at
    ELO improvement. Actually patch failed at long TC.

    LLR: -2.97 (-2.94,2.94)
    Total: 23109 W: 4104 L: 4092 D: 14913

    So revert.

    bench: 3453945
2013-09-05 06:34:48 +02:00
Reuven Peleg 457ac26de5 Rewrite backward pawn detection
Use the new backmost_sq() instead of a loop.

No functional change.
2013-09-03 20:11:00 +02:00
Eelco de Groot 679c2ea227 Fix check for bishop pair in material imbalance
Prefer pos.bishop_pair() to pos.count<BISHOP>(WHITE) > 1
because the first checks that the two bishops are on
different color squares.

Although the change seems to kick in only in very rare cases,
quite surprisingly it was able to pass SPRT test at short TC.

LLR: 2.95 (-2.94,2.94)
Total: 39818 W: 8174 L: 7956 D: 23688

bench: 3453941
2013-09-03 19:40:34 +02:00
Marco Costalba 9ff594c3a9 Rewrite KBBKN endgame
This was thought to be a draw but the bishops generally win. However,
it takes up to 66 moves. The position in the diagram was thought to be
a draw for over one hundred years, but tablebases show that White wins
in 45 moves. All of the long wins go through this type of semi-fortress
position. It takes several moves to force Black out of the temporary
fortress in the corner; then precise play with the bishops prevents Black
from forming the temporary fortress in another corner (Nunn 1995:265ff).

Before computer analysis, Speelman listed this position as unresolved,
but "probably a draw" (Speelman 1981:109).

bench: 3453945
2013-09-02 11:03:01 +02:00
Marco Costalba 849b089a63 Don't use lpthread for Android
Thanks to Peter Osterlund for the feedback.

No functional change.
2013-09-01 13:48:09 -07:00
Marco Costalba aee404f532 Improve ARM compatibility
STANDALONE-TOOLCHAIN.html in Android NDK says:

It is recommended to use the -mthumb compiler flag to force the generation
of 16-bit Thumb-1 instructions (the default being 32-bit ARM ones).

If you want to target the 'armeabi-v7a' ABI, you will need ensure that the
following two flags are being used:

  CFLAGS='-march=armv7-a -mfloat-abi=softfp'

Note: The first flag enables Thumb-2 instructions, and the second one
      enables H/W FPU instructions while ensuring that floating-point
      parameters are passed in core registers, which is critical for
      ABI compatibility. Do *not* use these flags separately!

Thanks to Peter Osterlund for pointout this doc and for showing me
an example Makefile to follow.

No functional change.
2013-09-01 09:18:37 -07:00
Uri Blass 0915f85895 Union of 2 changes that failed with good score
This is a union of 2 changes:

A tweak of recaptures limit from Joona Kiiski
http://tests.stockfishchess.org/tests/view/52166d7c0ebc59319a242400

and a tweak of move count pruning from Leonid Pechenik
http://tests.stockfishchess.org/tests/view/5217c7e60ebc59319a242456

The set passed both short TC at 30+0.05
LLR: 2.96 (-2.94,2.94)
Total: 18936 W: 3723 L: 3566 D: 11647

And the usual long TC at 60+0.05
LLR: 2.95 (-2.94,2.94)
Total: 48962 W: 8837 L: 8487 D: 31638

bench: 3453945
2013-09-01 08:07:21 -07:00
Marco Costalba 5e8bc6ac2a Assorted clean up in endgames
No functional change.
2013-09-01 07:39:04 -07:00
Marco Costalba 3e4dcaa06e Fix a bogus assert in allows()
Becuase castle is coded as "king captures the rook"
the to_sq(move), A1/8 or H1/8 is empty after the move,
leading to assert assert(p != NO_PIECE) in color_of().

Teach allows() asserts about castle and fix the crash.

Bug reported by Ryan Takker and tracked down by Tom Vijlbrief.

No functional change.
2013-08-30 16:42:18 +02:00
Marco Costalba 14f47c8ac6 Use frontmost_sq() and backmost_sq helpers
Should easier to read than the lsb() / msb() low
level functions.

No functional change.
2013-08-30 16:22:22 +02:00
Gary Linscott 5d90c149b5 Enable LMR for dangerous moves
Passed both short TC
LLR: 2.96 (-2.94,2.94)
Total: 5598 W: 1250 L: 1125 D: 3223

And long TC
LLR: 2.97 (-2.94,2.94)
Total: 16441 W: 3102 L: 2912 D: 10427

bench: 4620975
2013-08-29 23:02:18 +02:00
homoSapiensSapiens 4b9e338541 Bonus for rook behind a passed
If our rook is behind a passed pawn, all
squares are defended.

One of the longest tests to pass !

Passed both short TC
LLR: 2.97 (-2.94,2.94)
Total: 44560 W: 9518 L: 9281 D: 25761

And long TC
LLR: 2.96 (-2.94,2.94)
Total: 61348 W: 11618 L: 11192 D: 38538

bench: 3787694
2013-08-29 22:53:21 +02:00
homoSapiensSapiens a0cf424cfc Replace hardcoded 128 by constant
No functional change.
2013-08-29 13:59:49 +02:00
Gary Linscott aecdbfc4a0 Add lsb() overload
Helper to find least significant bit relative to
the given color.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-08-28 19:50:11 +02:00
Uri Blass e6482b7d97 Time management: move faster if PV is stable
Move faster but compensate by allocating more
time when the best move changes.

Passed short TC at 15+0.05
LLR: 2.93 (-2.94,2.94)
Total: 13895 W: 3030 L: 2882 D: 798

Long TC at 60+0.05
LLR: 2.96 (-2.94,2.94)
Total: 9266 W: 1777 L: 1624 D: 5865

At time increment 30+0.5
LLR: 2.96 (-2.94,2.94)
Total: 6703 W: 1238 L: 1134 D: 4331

And at fixed game number, longer TC 120+0.05
ELO: 5.17 +-2.8 (95%) LOS: 100.0%
Total: 19306 W: 3378 L: 3091 D: 12837

bench: 4728533
2013-08-26 10:29:58 -07:00
homoSapiensSapiens 6e6c5b6103 Simplify chain detection
No functional change.
2013-08-25 18:36:49 +02:00
homoSapiensSapiens 04fd7efdfa Simplify kf definition in shelter_storm()
No functional change.
2013-08-25 18:31:47 +02:00
Chris Caino 5e331f9618 Fix KBPK bug
With

position fen 7k/8/8/8/8/7P/6K1/7B w - - 0 1
go depth 25

The evaluation at depth 22 is not draw as it should be. The reason is that
when search reaches the position 8/6kP/8/8/8/3B4/6K1/8 w - - 0 1 if white plays
h8R or h8N then we get a position that is a "KNOWN_WIN" and is _not_ a check, so
futility pruning in qsearch kicks in and black may think that it is "futile"
to reply Kxh8 since, according to the logic of the code, it cannot raise the score
back towards a draw.

bench: 4728533
2013-08-24 10:08:03 +02:00
homoSapiensSapiens b9f5d1c6ff Simplify condition for backward pawn
No functional changes.
2013-08-22 14:39:08 +02:00
Marco Costalba c4533e0d94 Retire redundant endgames
The case of two lone kings on the board is already considered
by the "No pawns" scaling factor rules in material.cpp as is
KBK and KNK.

Moreover we had a small leak in endgames map because for
KK endgame it happens white and black material keys are the
same (both equal to zero), so when adding the black endgame in
Endgames::add() we were overwriting the already exsisting
white one, leading to a memory leak found by Valgrind.

So remove the endgames althogheter and rely on scaling
to correctly set the endgames value to a draw.

No functional change.
2013-08-22 13:13:06 +02:00
Joona Kiiski f39cf1b008 Use null move when depth >= 2 plies
Passed both short TC:
LLR: 2.96 (-2.94,2.94)
Total: 23725 W: 5031 L: 4855 D: 13839

And long TC:
LLR: 2.96 (-2.94,2.94)
Total: 15730 W: 2939 L: 2754 D: 10037

bench: 4729333
2013-08-22 09:06:48 +02:00
Marco Costalba c6baefb79d Restore development version
No functional change.
2013-08-21 08:41:47 +02:00
Marco Costalba 4d120ee02e Stockfish 4
Stockfish bench signature is: 4132374
2013-08-20 09:01:25 +02:00
Tom Vijlbrief f45eee318b Fix crash when reaching max ply
Bug introduced in 1b7223a53c that
updated the ss base stack without increasing
the dimension.

No functional change.
2013-08-19 16:53:46 +02:00
Tom Vijlbrief 8c2fd2170a Remove useless condition in KXK endgame
Because eval is never called when in check.

No functional change.
2013-08-19 08:55:17 +02:00
Leonid Pechenik 91c2c44fb1 Further tweak movecount pruning
Passed both short TC
LLR: 2.95 (-2.94,2.94)
Total: 15140 W: 3125 L: 2976 D: 9039

And long TC
LLR: 2.95 (-2.94,2.94)
Total: 17118 W: 3165 L: 2974 D: 10979

bench: 4132374
2013-08-18 09:13:57 +02:00
Marco Costalba 27e9fc1067 Normalize "pawn in front of minor" patch
No functional change.
2013-08-17 11:05:55 +02:00
homoSapiensSapiens e005270fb6 Use constants arguments where possible
No functional changes.
2013-08-16 09:57:21 +02:00
Marco Costalba 4f55ed14d3 Revert using exceptions
Due to crashes. It will be reapplied once
we understand what's happening.

No functional change.
2013-08-15 09:36:26 +02:00
homoSapiensSapiens a6c0ba2100 Simplify DistanceRingsBB init
Verified by same benchmark and picking some random values.

No functional change.
2013-08-14 10:53:43 +02:00
homoSapiensSapiens fc316cbca9 Some renaming in TT store()
No functional change.
2013-08-14 09:38:35 +02:00
Marco Costalba 11b1a76f35 Use exceptions to stop the search
Instead of classical flags, throw an
exception when we want to immediately halt
the search. Currently only one type
is used for both UCI stop and threads
cut off.

No functional change.
2013-08-14 08:29:57 +02:00
Tom Vijlbrief bd8f463b7e Bonus for a pawn in front of knight/bishop
Idea originated from a post of Don Dailey
on talkchess and reported by Eelco.

This is the last succesful attempt of a long
series of trials (as usually happens, the
'idea' alone is not enough).

Passed both short 15secs TC
LLR: 2.97 (-2.94,2.94)
Total: 7629 W: 1645 L: 1515 D: 4469

And long 60secs TC
LLR: 2.96 (-2.94,2.94)
Total: 10218 W: 1932 L: 1775 D: 6511

bench: 4944581
2013-08-13 14:20:02 +02:00
Ryan Takker 4d14f97482 Remove Now Unneeded Help Text
With the new automatic setting of split depth
instead of a default, the user no longer needs
guidance on setting the split point.

Also threads now defaults to one.

No functional change.
2013-08-13 07:36:33 +02:00
Marco Costalba 15616ad199 Don't set Search::RootColor in Eval::trace
Search::RootColor is a global parameter set
before to start a search, it is not something
trace() should change.

This patch allows to add trace() calls, for
debugging, inside search itself without altering
the bench, and also ensures that the values
returned by trace() and evaluate() are fully
equivalent.

No functional change.
2013-08-11 07:02:50 +02:00
Marco Costalba 94a3608ab9 Fix GrainSize rounding error
The rounding formula is different between
positive and negative scores due to the
GrainSize/2 term that is asymmetric.

So use truncation instead of rounding. This
guarantees that evaluation is rounded to zero
in the same way for both positive and negative
scores.

Found with position's flip

bench: 4634244
2013-08-10 17:11:13 +02:00
Marco Costalba 5769509d72 Fix 'improving' condition
Because VALUE_NONE is 30002, it happens that
after a check the next move is never an improving
one.

After this patch bench signature is independent from
VALUE_NONE actual value.

bench: 4303194
2013-08-09 08:21:55 +02:00
Marco Costalba fff6b9f061 Increase LMR when not improving
Apply to LMR the same Eelco's idea
applied to move count pruning.

This is the result of a series of
attempts started by Thomas Kolarik.

Passed both short TC
LLR: 2.95 (-2.94, 2.94)
Total: 5675 W: 1241 L: 1117 D: 3317


And long TC:
LLR: 2.95 (-2.94, 2.94)
Total: 8748 W: 1689 L: 1539 D: 5520

bench: 4356801
2013-08-08 10:28:48 +02:00
Marco Costalba 56d2c3844a Further tweak Position::flip
No functional change.
2013-08-05 14:44:06 +02:00
Marco Costalba 23b6809f3d Rewrite flip() to use FEN string manipulation
Instead of dealing directly with internal parameters
just "flip" the FEN string and set the position from
that.

No functional change.
2013-08-05 12:58:14 +02:00
Marco Costalba f31847302d Streamline time computation
No functional change.
2013-08-03 18:30:43 +02:00
Marco Costalba b1a4a18d63 Update polyglot.ini with new "Min Split Depth" default
No functional change.
2013-08-03 16:41:18 +02:00
Dan Schmidt f7096ea7ce Refactor do_castle()
Not a real functional change, but bench changed due to different piecelist
reordering. To verify it a temporary my canonicalize_rooks function was
written as follows. It just ensures that the rook on the "smaller" square
is listed first.

void Position::canonicalize_rooks(Color c)
{
   if (pieceCount[c][ROOK] == 2)
   {
      Square s0 = pieceList[c][ROOK][0];
      Square s1 = pieceList[c][ROOK][1];
      if (s0 > s1)
      {
         pieceList[c][ROOK][0] = s1;
         pieceList[c][ROOK][1] = s0;
         index[s0] = 1;
         index[s1] = 0;
      }
   }
}

With this both bench and the test on Chess960 positions

./stockfish bench 128 1 8 Chess960.epd file > /dev/null

Gives same result.

bench: 4424151
2013-08-03 16:18:28 +02:00
Joona Kiiski a16ba5bbd1 Retire cpu_count()
Set threads number always to 1 at startup and let the
user explicitly to chose the number of threads.

Also preserve the useful behavior of automatically set
"Min Split Depth" according to the requested threads,
indeed this parameter is too technical for a casual user,
so, when left to zero, we set it on a sensible value.

No functional change
2013-08-02 16:48:25 +02:00
Marco Costalba 408e6ee9b6 Further factor out position update code
Along the lines of previous patch.

No functional change
2013-08-01 16:32:46 +02:00
Dan Schmidt 7b4f5c8f72 Factor out pieceList updating code
The new Position methods add_piece, move_piece, and remove_piece
now manage the member variables pieceList, pieceCount, and index,
and 9 blocks of code in Position that used to manipulate those
data structures by hand now call the new methods.

There is a slightly slowdown (< 1%) on Clang and on perft,
but the cleanup compensates the little speed loss.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-08-01 15:50:19 +02:00
Marco Costalba 55948623e7 Rework Thread hierarchy
Introduce ThreadBase struct that is search
agnostic and just handles low level stuff,
and derive all the other specialized classes
form here.

In particular TimerThread does not hinerits
anymore all the search related stuff from Thread.

Also some renaming while there.

Suggested by Steven Edwards

No functional change.
2013-07-31 18:35:52 +02:00
Marco Costalba 4d46d29efe Fix a race at thread creation
At thread creation start_routine() is called
and from there the virtual function idle_loop()
because we do this inside Thread c'tor, where the
virtual mechanism is disabled, it could happen that
the base class idle_loop() is called instead.

The issue happens with TimerThread and MainThread
where, at launch, start_routine calls
Thread::idle_loop instead of the derived ones.

Normally this bug is hidden because c'tor finishes
before start_routine() is actually called in the
just created execution thread, but on some platforms
and in some cases this is not guaranteed and the
engine hangs.

Reported by Ted Wong on talkchess

No functional change.
2013-07-31 18:35:32 +02:00
Marco Costalba cc608a7aba Tidy up Position::pretty
No functional change.
2013-07-29 19:33:30 +02:00
Marco Costalba 1f40cd6d02 Small renaming
No functional change.
2013-07-29 19:32:59 +02:00
Marco Costalba 28bc8ed462 Speed up move generation
Pass the color as template parameter
to generate_all()

Speedup of 1,3% in perft and 2,5% in bench !

No functional change.
2013-07-29 19:01:50 +02:00
Eelco de Groot 5ee16a180a Increase pruning if evaluation is not improving
Add an additional set of margins to movecount pruning
to be used when static evaluation is getting worse
than previous move.

Here are the margins table with changing
depth (fm0 not improving, fm1 improving):

    d: 0, fm0: 3, fm1: 3
    d: 1, fm0: 4, fm1: 4
    d: 2, fm0: 6, fm1: 6
    d: 3, fm0: 7, fm1: 10
    d: 4, fm0: 11, fm1: 15
    d: 5, fm0: 15, fm1: 21
    d: 6, fm0: 21, fm1: 29
    d: 7, fm0: 27, fm1: 37
    d: 8, fm0: 35, fm1: 47
    d: 9, fm0: 42, fm1: 57
    d: 10, fm0: 51, fm1: 68
    d: 11, fm0: 60, fm1: 81
    d: 12, fm0: 70, fm1: 94
    d: 13, fm0: 81, fm1: 108
    d: 14, fm0: 92, fm1: 123
    d: 15, fm0: 104, fm1: 139

Good at both short TC

LLR: 2.97 (-2.94,2.94)
Total: 11502 W: 2503 L: 2361 D: 6638

And long TC

LLR: 2.98 (-2.94,2.94)
Total: 7189 W: 1421 L: 1277 D: 4491

bench: 4364793
2013-07-29 01:21:21 +02:00
Marco Costalba d30dfc084c Annotate an unlikely condition
No functional change.
2013-07-27 11:34:15 +02:00
Marco Costalba 6373e88b5b Fix an assert in KBK endgame
The endgame king + minor vs king is erroneusly
detected as king + minor vs king + minor

Here the fix is to detect king + minor earlier,
in particular to add these trivial cases to
endgame evaluation functions.

Spotted by Reuven Peleg

bench: 4727133
2013-07-27 08:25:45 +02:00
Tom Vijlbrief 7487eb0dca Rewrite pawn shield and storm code
Passes quickly both short TC:
LLR: 2.95 (-2.94,2.94)
Total: 5755 W: 1349 L: 1222 D: 3184

And long TC:
LLR: 2.95 (-2.94,2.94)
Total: 2744 W: 628 L: 505 D: 1611

bench: 4727133
2013-07-26 00:12:46 +02:00
Marco Costalba 2067a99c07 Fix a typo in bitboard.h
Introduced by previous patch.

Spotted by Joerg Oster

No functional change.
2013-07-25 07:44:27 +02:00
homoSapiensSapiens 002062ae93 Use #ifndef instead of #if !defined
And #ifdef instead of #if defined

This is more standard form (see for example iostream file).

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-07-24 19:49:17 +02:00
Marco Costalba 4064ee5406 Simplify captures ordering
A big simplification and removing of useless code.

Finished at 50% both at short TC (with SPRT) than
at long TC at fixed number of games:
ELO: -0.14 +-3.4 (95%) LOS: 46.8%
Total: 15206 W: 2836 L: 2842 D: 9528

bench: 5059948
2013-07-24 07:53:32 +02:00
Marco Costalba b0fd2b6b98 Revert "Halve king eval margin"
This reverts commit 4b3a0fdab0.

As Gary says: " It failed when I tried it at long TC previously, and only
barely passed this time.  Some anecdotal evidence is that it hurts vs other
engines as well (the Lightspeed rating list showed a 16 elo drop from previous
best version - still +- 5 error bars on both, but that's still significant)"

I also agree that if we have some doubts (like in this case) it is better to
be safe than sorry.

bench: 4615572
2013-07-24 07:46:25 +02:00
Ryan Schmitt d0bc951835 Tune pawn PSQT values
Reduces the influence of PSQT for entries such as
the extended center and the h-file.

Passed both short TC test:
LLR: 2.95 (-2.94,2.94)
Total: 23919 W: 5207 L: 5029 D: 13683

And long TC one:
LLR: 2.96 (-2.94,2.94)
Total: 5762 W: 1108 L: 974 D: 3680

Bench: 4617880

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-07-24 07:41:51 +02:00
Reuven Peleg 73131e7c78 Use arrow operator instead of * and .
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-07-22 20:22:40 +02:00
Reuven Peleg 1cc18d8a7a Better condition in is_pseudo_legal()
Simplify occupied destination condition.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-07-22 20:02:31 +02:00
Marco Costalba c45c6d308f Small touches in move generation
No functional change.
2013-07-21 11:01:24 +02:00
Marco Costalba f73bb438aa Some renaming in MovePicker
No functional change.
2013-07-21 09:55:08 +02:00
Marco Costalba 71dd8a333f Rewrite and simplify SEE
This very speed critical code was full of clever (!)
tricks and subtle details.

So I have rewritten it in a more straithforward way
and, as very often happens, result is even faster
than original.

No functional change.
2013-07-21 01:04:29 +02:00
Marco Costalba 9207baed65 Revert "Fix critical SEE bug (take 2)"
This reverts commit 3e95800814

For some reason it fails the short TC test:
LLR: -2.96 (-2.94,2.94)
Total: 20033 W: 4214 L: 4265 D: 11554

bench: 4769737
2013-07-20 18:45:38 +02:00
Marco Costalba b191df5ebe Revert "Yet another attempt at signature-build"
Still broken on OS X

No functional change.
2013-07-20 15:15:31 +02:00
Marco Costalba 3e95800814 Fix critical SEE bug (take 2)
It is somewhat unbilievable but our SEE is broken !

    If the first SEE move is a king capture and square is
    defended then SEE continues instead of breaking.

    The bug shows only on normal SEE, not see_sign() so
    probing with a:

    dbg_hit_on_c(slIndex==1, captured == KING);

    reports just a tiny:

    Total 3465656 Hits 6646 hit rate (%) 0

    Bug was there since Retire seeValues[] and move PieceValue[] out of Position of 26/6/2011 (!)
    although for some reason didn't show immediately, indeed the
    bougous patch was a "No functional change" (!!)

    bench: 4699504
2013-07-20 14:24:47 +02:00
Marco Costalba 2ed56f4d5f Revert all the SEE stuff
The speed up seems to introduce some
functionality change.

Revert to original master for now.

bench: 4769737
2013-07-20 14:20:33 +02:00
Marco Costalba 7b0b463720 Yet another attempt at signature-build
This one should work on all flavours of sed

Suggested by by Louis Zulli

No functional change.
2013-07-20 14:05:04 +02:00
Marco Costalba a6c5b4c6fb Fix critical SEE bug
It is somewhat unbilievable but our SEE is broken !

If the first SEE move is a king capture and square is
defended then SEE continues instead of breaking.

The bug shows only on normal SEE, not see_sign() so
probing with a:

dbg_hit_on_c(slIndex==1, captured == KING);

reports just a tiny:

Total 3465656 Hits 6646 hit rate (%) 0

Bug was there since 351ef5c85b of 26/6/2011 (!)
although for some reason didn't show immediately, indeed the
bougous patch was a "No functional change" (!!)

bench: 4793754
2013-07-20 13:37:12 +02:00
Marco Costalba 0504a6975d Speedup see()
And rename next_attacker() SEE helper

This very simple patch is able to speed up
bench run of almost 2% !

No functional change.
2013-07-20 12:34:35 +02:00
Marco Costalba a5b5a91512 Fix signature-build under OSX
On OS X when you use -i an extension for the in-place
substitution a backup files is required.

http://stackoverflow.com/questions/4247068/sed-command-failing-on-mac-but-works-on-linux

So rewrite to make sed flushing sign.txt in one go and avoid
using -i option.

Reported by Louis Zulli

No functional change.
2013-07-20 02:38:14 +02:00
Reuven Peleg 1a8f63a896 Microptimize gives_check() for castling case
Without patch we have 333198 nps, with patch 334249.

A very small +0.3%, not a lot manily becuase this is a
side path that is taken very few times.

Anyhow idea is correct becuase first 'quick' condition
has an hit rate of about 95%.

No functional change.
2013-07-19 17:07:54 +02:00
Marco Costalba ee5514b8fd Small simplification in space eval scoring
No functional change.
2013-07-19 11:13:18 +02:00
Marco Costalba 99e547f4cb Rename MoveStack to ExtMove
Stack has no meaning here, while ExtMove (extended move),
better clarifies that we have a move + a score.

No functional change.
2013-07-19 10:27:58 +02:00
Marco Costalba 110644d918 Better document what we skip when in check
No functional change.
2013-07-19 09:37:31 +02:00
homoSapiensSapiens 4b3a0fdab0 Halve king eval margin
But still keep the same original
margin for score.

Passed both short TC test
LR: 2.95 (-2.94,2.94)
Total: 3710 W: 845 L: 726 D: 2139

And long TC
LLR: 2.95 (-2.94,2.94)
Total: 57859 W: 10939 L: 10532 D: 36388

bench: 4769737

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-07-19 08:16:39 +02:00
Marco Costalba 05e31c5e5f Drop grep and tr dependency in Makefile
Use only sed to get the bench signature.

No functional change.
2013-07-15 21:39:06 +02:00
Marco Costalba 46fdb14b2f Don't use __builtin_expect
Partially revert previous patch and use
unlikey() just as code annotation.

Actually it is better to rely on a profiler for branch prediction:

http://blog.man7.org/2012/10/how-much-do-builtinexpect-likely-and.html

"In fact, even when only one in ten thousand values is nonzero,
we're still at only roughly the break-even point"

No functional change,
2013-07-15 21:09:06 +02:00
Marco Costalba cbb1a8ed31 Better annotate unlikely conditions
And in case of gcc we win also a small
speed optimization due to better branch
prediction.

No functional change.
2013-07-15 21:01:02 +02:00
Reuven Peleg a6c5f60caa Simplify a condition in refutes()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-07-15 20:40:49 +02:00
Marco Costalba 9518cc3254 Update 'make help'
No functional change.
2013-07-14 12:23:28 +02:00
Marco Costalba b0a177bc67 Add signature-profile-build make target
Extend patch 3f64a2af6a to profile builds.

here the make command is:

make signature-profile-build ARCH=xxx COMP=xxx

No functional change.
2013-07-14 11:57:06 +02:00
Marco Costalba 5b7b330616 Retire engine Tag
It is somewhat redundant and could make SF
name too long, so use just Version, in case
of a signature build Version will be set to
'sig-xxx' otherwise, if left empty, we fall
back on usual date stamp.

No functional change.
2013-07-14 11:13:13 +02:00
Marco Costalba 3f64a2af6a Add signature-build make target
When compiling with:

make signature-build ARCH=xxx COMP=xxx

After binary has been roduced, it will be run to
get the signature 'stockfish bench' and this
number will be used as Version, so that it
will be easy to track the original sources
from a binary.

No functinal change.
2013-07-14 10:59:25 +02:00
Marco Costalba 9749f1f14c Fix build on Intel compiler
Due to a strange issue (bug?) the ternary
operator does not return a BitCountType for
icc, so revert to the expression used
before bcbc9bfd1f

No functional change.
2013-07-13 23:13:30 +02:00
Marco Costalba 4ede49cd85 Fully qualify memset and memcpy
And other trivial touches.

Ispired by Lucas's DiscoCheck

No functional change.
2013-07-13 18:01:13 +02:00
Tom Vijlbrief 6960f41e03 Retire enoughMaterial + lower trapped rook threshold
Here speed up is the name of the game.

Speed up is gained:

- Removing the useless enoughMaterial code

- Limiting trapped rook evaluation to where it counts

Tested at long TC:
LLR: 2.97 (-2.94,2.94)
Total: 10061 W: 1948 L: 1790 D: 6323

bench: 4558173
2013-07-13 18:01:03 +02:00
Marco Costalba bf90499fc3 A useless assignment found by Clang’s static analyzer
Warning is: "Value stored to 'xxx' is never read" and
it is raised in SpNode case.

No functional change.
2013-07-13 16:57:03 +02:00
Marco Costalba 404c4122ce Fix build with MSVC 2013
Also add an assert hinted by MSVC code analysis tool.

No functional change.
2013-07-13 13:03:09 +02:00
Marco Costalba 2a0bbb9faa Fix printing of PV info: take 2
Now last PV line is printed twice, fix that.

No functional change.
2013-07-13 07:43:50 +02:00
Marco Costalba 128e097d03 Fix printing of PV info
It was erroneusly skipped after the
aspiration window rework.

Reported by Eelco.

No functional change.
2013-07-12 23:42:42 +02:00
Marco Costalba c1264e46d0 Rename some UCI options
Thanks to Don, Miguel, Louis and the other people
of talkchess forum for the suggestion:

http://www.talkchess.com/forum/viewtopic.php?t=48612

Also sync polyglot.ini with current UCI options

No functional change.
2013-07-11 16:07:10 +02:00
Marco Costalba 366f6b0dab Fix a crash with depth 1 perft
Bug recently introduced in e215a88cdd

No functional change.
2013-07-11 07:23:48 +02:00
Marco Costalba 58aee9a9ea Don't IID when in check also in PvNodes
This tiny functional change allows to
nicely simplify things.

Performed at 50% in short TC:
LLR: -0.43 (-2.94,2.94)
Total: 46406 W: 9681 L: 9565 D: 27160

And succesfully passed long TC reverse test:
LLR: -2.95 (-2.94,2.94)
Total: 4945 W: 858 L: 937 D: 3150

bench: 4507230
2013-07-09 19:03:11 +02:00
Marco Costalba 4b703b1429 Revert previous patch
Unfortunatly a reverse test at long TC failed:

master^ vs master
LLR: 1.37 (-2.94,2.94)
Total: 33682 W: 6294 L: 6071 D: 21317

So becuase short TC score is 50% there is a good
possibility patch is not scalable.

So revert it.

bench: 4507288
2013-07-09 08:04:12 +02:00
Marco Costalba 62d38f0196 Simplify "fail high upon reduction" in null search
Do not use threat move to detect the condition. This
let us to retire the big allows() function.

Test at short TC was within 50% score:
LLR: -2.95 (-2.94,2.94)
Total: 38272 W: 7941 L: 7940 D: 22391

To be verified with reverse long TC

bench: 4191565
2013-07-08 07:23:30 +02:00
Marco Costalba 7e575512ae Skip node-level cut-off tests when in check
No functional change.
2013-07-07 13:39:58 +02:00
Marco Costalba a55fb76dcc Simplify aspiration window code
Here the main difference is that now we center
aspiration window on last returned score. This allows
to simplify handling of mate scores.

We have done a reversed SPRT tests, where we wanted to
verify if master is stronger than this patch.

Long TC: master vs this patch (reverse test)
LLR: -2.95 (-2.94,2.94)
Total: 37992 W: 7012 L: 6920 D: 24060

bench: 4507288
2013-07-03 18:59:23 +02:00
Marco Costalba 838255ef91 Workaround github issue
Temporary revert aspiration window patch
so to be visible to everybody: it will be
re-applied with next patch

No functional change (together with next one)
2013-07-03 18:58:23 +02:00
Marco Costalba 6fbe027da0 Simplify aspiration window code
Here the main difference is that now we center
aspiration window on last returned score. This allows
to simplify handling of mate scores.

We have done a reversed SPRT tests, where we wanted to
verify if master is stronger than this patch.

Long TC: master vs this patch (reverse test)
LLR: -2.95 (-2.94,2.94)
Total: 37992 W: 7012 L: 6920 D: 24060

bench: 4507288
2013-07-03 09:06:03 +02:00
Marco Costalba ddcb572c41 Disable flto when debugging
Link-time optimization does not work well with
generation of debugging information:

http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html

Reported by Louis Zulli

No functional change.
2013-07-03 08:21:21 +02:00
Marco Costalba 12c0dfc113 Revert "Remove confusing optimization"
This reverts commit e05c80a088.

we gain a speed up of 1.5% under gcc !

No functional change.
2013-07-02 20:12:57 +02:00
Marco Costalba db53883f06 Merge branch 'master' into aspiration
bench: 4507288
2013-07-02 07:27:08 +02:00
Marco Costalba 13ffb08136 Revert "Increase earlier aspiration window size"
This reverts commit b88bc7b766.
2013-07-02 07:25:39 +02:00
Marco Costalba 2d82db1d14 Entering a pawn endgame is no more dangerous
A simplification of the 'dangerous' definition.

Seems neutral at reverse test at long TC

master vs patch
LLR: -2.96 (-2.94,2.94)
Total: 16974 W: 3122 L: 3139 D: 10713

bench: 4689029
2013-07-02 07:24:17 +02:00
Marco Costalba b88bc7b766 Increase earlier aspiration window size
bench: 4377851
2013-07-01 19:29:38 +02:00
Marco Costalba e074a19f5c Merge branch 'master' into aspiration 2013-07-01 19:25:23 +02:00
Marco Costalba b8930d0c26 Fix a stale comment
No functional change.
2013-06-30 13:12:04 +02:00
Marco Costalba 4fc7734547 Simplify search results update
Also some rename while there.

No functional change.
2013-06-30 12:49:39 +02:00
Marco Costalba 92dcbfa658 Reorder conditions according to their frequency
This should minimize useless tests.

No functional change.
2013-06-30 11:35:53 +02:00
Marco Costalba b50eb6bea8 Center aspiration window on last returned score
bench: 4428212
2013-06-30 11:14:02 +02:00
Marco Costalba 6f079ae720 Simplify aspiration window loop
Don't open the window in case we find a mate score: this
will be takes care with next patch.

No functional change.
2013-06-30 10:54:09 +02:00
Marco Costalba 203fdc9ac1 Use calloc() in TranspositionTable::set_size()
Function calloc() already initializes memory to
zero, so avoid calling clear() afterwards.

Also some renaming while there (inspired by DiscoCheck).

No functional change.
2013-06-29 11:23:07 +02:00
Marco Costalba 17d41b3861 Fix some stale comments
No functional change.
2013-06-23 13:19:03 +02:00
Marco Costalba 8cff4862a6 Move SquareDistance[] to bitboard.cpp
No functional change.
2013-06-23 13:13:13 +02:00
Marco Costalba 908d98820b Don't explicitize enum values when not needed
Compiler will chose the correct values in sequential
order for you.

Also move file and rank bitboards definitions to
bitboard.h

No functional change.
2013-06-23 11:30:40 +02:00
Marco Costalba a4c11b71ac Retire in_front_bb(Color c, Square s) overload
Explciitly call rank_of() in the few places where
it is used.

No functional change.
2013-06-23 10:16:43 +02:00
Marco Costalba b2fadf32aa Retire ThisAndAdjacentFilesBB[]
It is unused. Also renamed attack_span_mask to
pawn_attack_span

No functional change.
2013-06-23 10:09:24 +02:00
Marco Costalba 378bcfe760 Simplify hidden_checkers()
De-templetize and pass color as function argument.
No speed change.

No functional change.
2013-06-23 09:03:40 +02:00
Marco Costalba fe2ed42661 Name functions along corresponding UCI commands
No functional change.
2013-06-22 12:46:03 +02:00
Marco Costalba e215a88cdd Micro-optimize perft
Avoid to call perft function when we just need to count
moves, at leaf nodes.

Speed up of almost 2%

No functional change.
2013-06-21 09:10:03 +02:00
Ryan Schmitt e95e69515a Include file attacks in 'major on pawn'
Passed both short TC:
LLR: 2.97 (-2.94,2.94)
Total: 57846 W: 12248 L: 11974 D: 33624

And long one:
LLR: 2.95 (-2.94,2.94)
Total: 9181 W: 1732 L: 1581 D: 5868

bench: 4609948
2013-06-19 07:22:10 +02:00
Reuven Peleg e05c80a088 Remove confusing optimization
Here we skip the call to pos.attacks_from<ROOK>(s) in the 98%
of cases, testing the first 2 members first. Unfortunatly
code is a bit triky and not clear. So we give up to the
speed optimization in exchange of more code clarity.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-06-17 18:12:15 +02:00
Reuven Peleg 7b31e81d77 Merge some if statements in pos_is_ok()
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-06-17 18:02:59 +02:00
Marco Costalba c0964fc70f Remove redundant condition in probcut
When !ss->skipNullMove it is assured that excludedMove == MOVE_NONE

No functional change.
2013-06-17 09:30:59 +02:00
Marco Costalba cd782c11ec Rename piece_count and piece_list
No functional change.
2013-06-16 13:21:10 +02:00
Marco Costalba 5ea984ac35 Don't calculate pawnsOnSquares twice
And reformat some code while there.

No functional change.
2013-06-16 11:32:10 +02:00
Marco Costalba 1fd020a8ba Use move_pawns() in Pawns::probe
And rename some stuff.

No functional change.
2013-06-16 10:40:36 +02:00
Marco Costalba 02420d4670 Revert "Reduce more CUT nodes only if parent node is reduced"
This reverts commit d54e8a5955.

It was not proved with SPRT this tweak is stronger. So revert it
for now to follow fishtest guidelines.

bench: 5108393
2013-06-14 08:27:06 +02:00
Marco Costalba 2c7ab488a8 Fix description of TT entry
It was way outdated and wrong !

No functional change.
2013-06-14 08:21:02 +02:00
Marco Costalba d54e8a5955 Reduce more CUT nodes only if parent node is reduced
So when we are doing a LMR search at the parent ALL node.

This patch didn't prove stronger at 60" TC
LLR: -2.97 (-2.94,2.94)
Total: 22398 W: 4070 L: 4060 D: 14268

But, first, it scores at 50%, second (and most important for me) the opposite,
i.e. normal reduction when parent node is not reduced, seems very bad:
LLR: -2.95 (-2.94,2.94)
Total: 7036 W: 1446 L: 1534 D: 4056

According to Don, this idea of increased reduction of CUT nodes
works because if parent node is reduced, missing a cut-off due to
reduced depth search (meaning position is somehow tricky) forces
a full depth research at parent node, giving due insight in this
set of sensible positions.

IOW if we expect a node to fail-high at depth n, then we assume it
should fail-high also at depth n-1, if this doesn't happen it means
position is tricky enough to deserve a research at depth n+1.

bench: 4687419
2013-06-13 20:05:02 +02:00
Marco Costalba 4bebb15e94 Reduce more CUT nodes
We got a good result from this tweak, in line with
what was already found by Don Dailey.

At short TC:
LLR: 2.95 (-2.94,2.94)
Total: 13097 W: 2742 L: 2598 D: 7757

At long TC:
LLR: 2.97 (-2.94,2.94)
Total: 7281 W: 1408 L: 1265 D: 4608

bench: 5108393
2013-06-13 19:50:32 +02:00
Marco Costalba 3b8f66f8ac Introduce Cut/All node definitions
Follow Don Dailey definition of cut/all node:

"If the previous node was a cut node, we consider this an ALL node.
The only exception is for PV nodes which are a special case of ALL nodes.
In the PVS framework, the first zero width window searched from a PV
node is by our definition a CUT node and if you have to do a re-search
then it is suddenly promoted to a PV nodes (as per PVS search) and only
then can the cut and all nodes swap positions. In other words, these
internal search failures can force the status of every node in the subtree
to swap if it propagates back to the last PV nodes."

http://talkchess.com/forum/viewtopic.php?topic_view=threads&p=519741&t=47577

With this definition we have an hit rate higher than 90% on:

    if (!PvNode && depth > 4 * ONE_PLY)
        dbg_hit_on_c(cutNode, (bestValue >= beta));

And an hit rate of just 28% on:

    if (!PvNode && depth > 4 * ONE_PLY)
        dbg_hit_on_c(!cutNode, (bestValue >= beta));

No functional change.
2013-06-13 19:46:49 +02:00
Marco Costalba b6e9d901b0 Don't use std::vector::data()
It is a C++11 only function.

Reported by Eelco.

No functional change.
2013-06-13 07:42:43 +02:00
Marco Costalba 1b7223a53c Fix again early stop ss pointer
Fix was wrong becuase search starts from ss+1,
code is a bit tricky here, so rewrite in a way
to be more easy to read and understand.

Spotted by Eelco.

No functional change.
2013-06-09 23:36:46 +02:00
Marco Costalba de1dc4f2de Don't need to expose namespace Zobrist
It can be local to position.cpp

No functional change.
2013-06-09 23:27:07 +02:00
Marco Costalba a6e0f62a4f Zobrist::init() should be Position::init()
No functional change.
2013-06-09 13:54:38 +02:00
Marco Costalba 81e242a96d Convert pieceSquareTable to 3 dimensions
No functional change.
2013-06-09 13:10:21 +02:00
Marco Costalba dd1855eb2f More consistent 'piece' variable naming
No functional change.
2013-06-09 12:56:05 +02:00
Marco Costalba db4cd89cb8 Introduce operator~(Piece c)
Small syntactic sugar to reverse piece color.

No functional change.
2013-06-09 12:44:04 +02:00
Marco Costalba 7e95495b35 Retire psq_delta()
No functional change.
2013-06-09 12:32:16 +02:00
Marco Costalba 55eb7dd1e9 Use alpha instead of beta-1
It is more directly related to a fail-low.

No functional change.
2013-06-09 11:52:39 +02:00
Marco Costalba 902c0566a6 Fix incorrect 'ss' pointer in early stop check
The exclusion search used to verify one move is much
better than other shall be called with 'ss' and not
'ss+1'

No functional change.
2013-06-09 11:01:11 +02:00
Dariusz Orzechowski bc02cc0c8a Fix a typo
No functional change.
2013-06-08 11:01:28 +02:00
Marco Costalba 05c6f7a40b Fix search log when using skills
In case of we pick a sub-optimal move be
sure to print this, and not the best one
on seach log file.

Bug spotted by Guenther Demetz.

No functional change.
2013-06-08 10:56:20 +02:00
Marco Costalba 2a98042c21 Fix a crash when 'go' multiple times
Search is started after setting a position and
issuing UCI 'go' command. Then if we stop the search
and call 'go' again without setting a new position it
is assumed that the previous setup is preserved, but
this is not the case because what happens is that
SetupStates is reset to NULL, leading to a crash as
soon as RootPos.is_draw() is called because st->previous
is now stale.

UCI protocol is not very clear about requiring that a
position is setup always before launching a search,
so here we easy the life of GUI developers assuming
that the current state is preserved after returning
from a 'stop' command.

Bug reported by Gregor Cramer.

No functional change.
2013-06-01 16:19:42 +02:00
Marco Costalba 46409a7852 Assorted renaming in evaluation
And some reshuffle too.

No functional change.
2013-06-01 13:17:39 +02:00
jundery d8b266af8b Passed pawn tuning
A small number of tests with simulated
annealing at 15s indicated these values
may be better

And this is verified at long 60+0.05 TC
LLR: 2.95 (-2.94,2.94)
Total: 40658 W: 7821 L: 7501 D: 25336

bench: 4931544
2013-05-31 09:17:48 +02:00
Marco Costalba abb40777bf Shrink engine UCI name
Some GUI have problems with long names.

Reported by George Speight.

No functional change.
2013-05-27 17:43:38 +02:00
Marco Costalba 90abcac1c7 Add Pawn Structure also to polyglot.ini
No functional change.
2013-05-25 13:14:41 +02:00
Marco Costalba 7222f47350 Re-add "Pawn Structure" UCI option
And reshuffle the code to not special case
this parameter.

No functional change.
2013-05-25 12:38:14 +02:00
Marco Costalba eafb66e1aa More uniform tracing code
No functional change.
2013-05-25 12:01:50 +02:00
Uri Blass d4a02b135d Bunch of 3 small patches
This patch is the sum of:

- Grainsize of 4 instead of 8

- Removing "depth < DEPTH_ZERO"

- Change DEPTH_QS_RECAPTURES = -5 to -7

All the patches individually failed to pass SPRT but scored
around 50%.

Together they pass easily short TC:
LLR: 2.96 (-2.94,2.94)
Total: 4429 W: 964 L: 844 D: 2621

And with some difficult long TC of 60+0.05:
LLR: 2.95 (-2.94,2.94)
Total: 64133 W: 11968 L: 11532 D: 40633

bench: 4821467
2013-05-23 17:59:39 +02:00
Marco Costalba d3608c4e79 Microptimize MoveList loop
Add MOVE_NONE at the tail, this allows to loop
across MoveList checking for *it != MOVE_NONE,
and because *it is used imediately after compiler
is able to reuse it.

With this small patch perft speed increased of 3%

And it is also a semplification !

No functional change.
2013-05-19 22:00:49 +02:00
Marco Costalba 38cfbeeb50 Delay killers[] initialization
Most of the time we cut-off earlier, at captures, so this
results in useless work.

There is a small functionality change becuase 'ss' can change
from MovePicker c'tor to when killers are tried due, for
instance, to singular search.

bench: 4603795
2013-05-19 21:41:56 +02:00
Marco Costalba 77547a4ef1 Reduce countermoves less in LMR
Passed SPRT for both short TC 15+0.05:
LLR: 2.95 (-2.94,2.94)
Total: 17724 W: 3756 L: 3598 D: 10370

And long TC 60+0.05:
LLR: 2.95 (-2.94,2.94)
Total: 22672 W: 4232 L: 4011 D: 14429

bench: 4418832
2013-05-19 21:36:23 +02:00
Marco Costalba 8ceef92266 Mimic an iterator for looping across MoveList
Seems more conventional.

No functional change.
2013-05-19 13:28:25 +02:00
Joona Kiiski f7c013edd0 Use two counter moves instead of one
Very good at long 60"+0.05 TC
LLR: 2.95 (-2.94,2.94)
Total: 5954 W: 1151 L: 1016 D: 3787

[edit: slightly changed form original patch to avoid useless loop
 across killers when killer is MOVE_NONE]

bench: 4327405
2013-05-16 16:20:50 +02:00
Marco Costalba 148490f04c Rename Refutation to Countermove
Use proper naming according to:

http://chessprogramming.wikispaces.com/Countermove+Heuristic

The name of this idea is "Countermove Heuristic" and was
first introduced by Jos Uiterwijk in 1992

No functional change.
2013-05-15 20:59:56 +02:00
Uri Blass 7c6f346c90 Increased mobility array
Performed more or less well at short TC
LLR: 2.95 (-2.94,2.94)
Total: 50517 W: 9815 L: 9574 D: 31128

And a bit better at long TC
LLR: 2.96 (-2.94,2.94)
Total: 15564 W: 2805 L: 2624 D: 10135

bench: 4375253
2013-05-15 00:34:05 +02:00
Marco Costalba 1a34496761 Revert trapped rook bug fix
It seems that do  not limiting checking the
trapped rook only on rank 1 improves the
score.

At long TC
LLR: 2.97 (-2.94,2.94)
Total: 6581 W: 1346 L: 1204 D: 4031

bench: 4985012
2013-05-15 00:06:11 +02:00
Gary Linscott 049e5ca191 Minor bugfixes to refutation table
Don't update refutation table in case of
previous move is MOVE_NULL or MOVE_NONE
and don't try refutation if is already
a killer move.

Pass both short TC
LLR: 2.96 (-2.94,2.94)
Total: 4310 W: 953 L: 869 D: 2488

And long one
LLR: 2.95 (-2.94,2.94)
Total: 6707 W: 1254 L: 1184 D: 4269

bench: 4785954
2013-05-14 23:52:44 +02:00
Marco Costalba 19dd0de4ff Reformat previous patch
No functional change.
2013-05-13 20:42:44 +02:00
Joona Kiiski 8a61b030a6 Enable refuation table
Very good result both at short TC 15+0.05
LLR: 2.95 (-2.94,2.94)
Total: 2803 W: 596 L: 483 D: 1724

And at long TC 60+0.05
LLR: 2.95 (-2.94,2.94)
Total: 2862 W: 548 L: 431 D: 1883

bench: 4329221
2013-05-13 19:48:41 +02:00
Joona Kiiski c7e31d5aa8 Simple always overwrite Refutation table 2013-05-12 21:21:46 +01:00
Marco Costalba 818a3537a7 Use Them instead of ~Us
Unortunatly we have no guarantee that the call to
operator~(Color c) is resolved at compile time.

Perhaps the solution would be to use C++11 const_expr,
but for now simply use the good old-style ternary operator
that works as expected.

No functional change.
2013-05-11 11:49:44 +02:00
Marco Costalba bcbc9bfd1f Some code reformat in evaluate_pieces
No functional change.
2013-05-11 11:13:06 +02:00
Marco Costalba 7eda7335fd Simplify previous patch
No functional change.
2013-05-09 19:09:51 +02:00
Marco Costalba 02606a8c83 Merge 'passed_pawns' tweaks
Good at both short and long TC

15+0.05
LLR: 2.96 (-2.94,2.94)
Total: 28220 W: 5531 L: 5349 D: 17340

TC 60+0.05
LLR: 2.95 (-2.94,2.94)
Total: 12612 W: 2221 L: 2057 D: 8334

bench: 4857939
2013-05-08 23:04:11 +02:00
Reuven Peleg e7505324f6 Avoid explicit bitwise operators
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-05-05 22:51:18 +02:00
jundery 653c0527a7 Passed pawn eval
Use a stepped function to evaluate bonuses and add the bonus to the
middle game

bench: 4857939
2013-05-05 11:12:04 -06:00
Marco Costalba 7f4c7cd785 Merge increased 'movecount' pruning
Good at both short and long TC

15+0.05
LLR: 2.95 (-2.94,2.94)
Total: 13814 W: 2731 L: 2588 D: 8495

TC 60+0.05
LLR: 2.95 (-2.94,2.94)
Total: 18013 W: 3136 L: 2946 D: 11931

bench: 4306557
2013-05-05 13:46:26 +02:00
Marco Costalba 0958e5c6d3 Simplify previous condition
No functional change.
2013-05-05 12:31:32 +02:00
Marco Costalba 9fc77bc414 Fix trapped rook condition
A rook is trapped if on rank 1 as is the king.
Currently the condition aloows for the rook
to be also in front of the pawns as long
as king is on first rank.

Verified with short TC test:
LLR: -1.71 (-2.94,2.94)
Total: 23234 W: 4317 L: 4317 D: 14600

Here what it counts is that after 23K games
result is equal.

bench: 4696542
2013-05-05 12:27:11 +02:00
Marco Costalba 3b92159908 Further simplify previous patch
No functional change.
2013-05-04 12:15:31 +02:00
homoSapiensSapiens e00bb13e85 Merge some conditions
No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>
2013-05-04 11:13:22 +02:00
Marco Costalba 3b41e62666 Drop some redundant defined(_WIN64)
When it is already defined(_WIN32).

According to Microsoft documentation:
http://msdn.microsoft.com/en-us/library/b0084kay.aspx

_WIN32 Defined for applications for Win32 and Win64. Always defined.

_WIN64 Defined for applications for Win64.

Patch suggested by Joona.

No functional change.
2013-05-03 15:24:54 +02:00
Marco Costalba 37c91aa94c Print time and node count before search ends
This info is normally printed together with
PV info in uci_pv() but when search is stopped,
for instance when max search time is reached,
uci_pv is not called and we miss this bits.

Suggested by gravy_train

No functional change.
2013-05-03 10:26:03 +02:00
Marco Costalba 43f67eab5f Merge mobility area tweak
A nice improvment.

Was good at 15+0.05
LLR: 2.96 (-2.94,2.94)
Total: 10731 W: 2176 L: 2040 D: 6515

And at 60"+0.05
LLR: 2.96 (-2.94,2.94)
Total: 10601 W: 1968 L: 1810 D: 6823

bench: 4676606
2013-05-03 10:12:31 +02:00
Gary Linscott 11d30b6298 Fix rounding issue 2013-05-02 14:37:55 -04:00
Gary Linscott 3edb15d183 More aggressive move count pruning 2013-05-02 09:47:34 -04:00
Marco Costalba d44ac0a485 Another take at TT alignment
This time revert to original version but using
uintptr_t instead of size_t

Suggested by Lucas.

No functional change.
2013-05-02 09:38:23 +02:00
Marco Costalba 481eda4ca0 Re-add "Cache line aligned TT"
But this time do not play with pointers, in
particular do not assume that size_t is an
unsigned type of the same width as pointers.

This code should be fully portable.

No functional change.
2013-05-01 23:42:16 +02:00
jhellis3 7323231786 Tweak Mobility Area
Only consider pawns and the king as restricting.
2013-05-01 02:37:50 -05:00
Marco Costalba e381951a24 Restore development version
No functional change.
2013-04-30 20:01:07 +02:00
44 changed files with 2274 additions and 2734 deletions
+5 -9
View File
@@ -7,14 +7,10 @@ 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 it.
This version of Stockfish supports up to 64 CPUs, but has not been
tested thoroughly with more than 4. The program tries to detect the
number of CPUs on your computer and sets 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 eight threads, it is
recommended to raise the value of the *Min Split Depth* UCI parameter to 7.
This version of Stockfish supports up to 64 CPUs. The engine defaults
to one search thread 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.
### Files
@@ -25,7 +21,7 @@ This distribution of Stockfish consists of the following files:
* Copying.txt, a text file containing the GNU General Public License.
* src/, a subdirectory containing the full source code, including a Makefile
* 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 below.
+9 -5
View File
@@ -14,22 +14,25 @@ ResignScore = 600
[Engine]
Use Search Log = false
Write Debug Log = false
Write Search Log = false
Search Log Filename = SearchLog.txt
Book File = book.bin
Best Book Move = false
Contempt Factor = 0
Mobility (Middle Game) = 100
Mobility (Midgame) = 100
Mobility (Endgame) = 100
Passed Pawns (Middle Game) = 100
Pawn Structure (Midgame) = 100
Pawn Structure (Endgame) = 100
Passed Pawns (Midgame) = 100
Passed Pawns (Endgame) = 100
Space = 100
Aggressiveness = 100
Cowardice = 100
Min Split Depth = 4
Min Split Depth = 0
Max Threads per Split Point = 5
Threads = 1
Use Sleeping Threads = false
Idle Threads Sleep = false
Hash = 128
Ponder = true
OwnBook = false
@@ -39,5 +42,6 @@ Emergency Move Horizon = 40
Emergency Base Time = 200
Emergency Move Time = 70
Minimum Thinking Time = 20
Slow Mover = 100
UCI_Chess960 = false
UCI_AnalyseMode = false
+66 -36
View File
@@ -34,8 +34,9 @@ ifeq ($(UNAME),Haiku)
endif
BINDIR = $(PREFIX)/bin
### Built-in benchmark for pgo-builds
### Built-in benchmark for pgo-builds and signature
PGOBENCH = ./$(EXE) bench 32 1 10 default depth
SIGNBENCH = ./$(EXE) bench
### Object files
OBJS = benchmark.o bitbase.o bitboard.o book.o endgame.o evaluate.o main.o \
@@ -164,6 +165,16 @@ ifeq ($(ARCH),osx-ppc-32)
sse = no
endif
ifeq ($(ARCH),linux-ppc-64)
arch = ppc64
os = any
bits = 64
prefetch = no
bsfq = no
popcnt = no
sse = no
endif
ifeq ($(ARCH),osx-x86-64)
arch = x86_64
os = osx
@@ -231,7 +242,7 @@ ifeq ($(COMP),clang)
endif
### 3.2 General compiler settings
CXXFLAGS = -g -Wall -Wcast-qual -fno-exceptions -fno-rtti $(EXTRACXXFLAGS)
CXXFLAGS = -Wall -Wcast-qual -fno-exceptions -fno-rtti $(EXTRACXXFLAGS)
ifeq ($(comp),gcc)
CXXFLAGS += -ansi -pedantic -Wno-long-long -Wextra -Wshadow
@@ -242,7 +253,7 @@ ifeq ($(comp),mingw)
endif
ifeq ($(comp),icc)
CXXFLAGS += -wd383,981,1418,1419,1476,10187,10188,11505,11503 -Wcheck -Wabi -Wdeprecated -strict-ansi
CXXFLAGS += -diag-disable 1476,10120 -Wcheck -Wabi -Wdeprecated -strict-ansi
endif
ifeq ($(comp),clang)
@@ -262,9 +273,12 @@ endif
### On mingw use Windows threads, otherwise POSIX
ifneq ($(comp),mingw)
# Haiku has pthreads in its libroot, so only link it in on other platforms
ifneq ($(UNAME),Haiku)
LDFLAGS += -lpthread
# On Android Bionic's C library comes with its own pthread implementation bundled in
ifneq ($(arch),armv7)
# Haiku has pthreads in its libroot, so only link it in on other platforms
ifneq ($(UNAME),Haiku)
LDFLAGS += -lpthread
endif
endif
endif
@@ -275,6 +289,8 @@ endif
### 3.4 Debugging
ifeq ($(debug),no)
CXXFLAGS += -DNDEBUG
else
CXXFLAGS += -g
endif
### 3.5 Optimization
@@ -293,7 +309,7 @@ ifeq ($(optimize),yes)
endif
ifeq ($(arch),armv7)
CXXFLAGS += -fno-gcse
CXXFLAGS += -fno-gcse -mthumb -march=armv7-a -mfloat-abi=softfp
endif
endif
@@ -305,7 +321,7 @@ ifeq ($(optimize),yes)
ifeq ($(os),osx)
CXXFLAGS += -fast -mdynamic-no-pic
else
CXXFLAGS += -O3
CXXFLAGS += -fast
endif
endif
@@ -354,6 +370,7 @@ endif
### needs access to the optimization flags.
ifeq ($(comp),gcc)
ifeq ($(optimize),yes)
ifeq ($(debug),no)
GCC_MAJOR := `$(CXX) -dumpversion | cut -f1 -d.`
GCC_MINOR := `$(CXX) -dumpversion | cut -f2 -d.`
ifeq (1,$(shell expr \( $(GCC_MAJOR) \> 4 \) \| \( $(GCC_MAJOR) \= 4 \& $(GCC_MINOR) \>= 5 \)))
@@ -361,6 +378,7 @@ ifeq ($(comp),gcc)
LDFLAGS += $(CXXFLAGS)
endif
endif
endif
endif
### ==========================================================================
@@ -375,44 +393,47 @@ help:
@echo ""
@echo "Supported targets:"
@echo ""
@echo "build > Build unoptimized version"
@echo "profile-build > Build PGO-optimized version"
@echo "strip > Strip executable"
@echo "install > Install executable"
@echo "clean > Clean up"
@echo "testrun > Make sample run"
@echo "build > Standard build"
@echo "signature-build > Standard build with embedded signature"
@echo "profile-build > PGO build"
@echo "signature-profile-build > PGO build with embedded signature"
@echo "strip > Strip executable"
@echo "install > Install executable"
@echo "clean > Clean up"
@echo ""
@echo "Supported archs:"
@echo ""
@echo "x86-64 > x86 64-bit"
@echo "x86-64-modern > x86 64-bit with runtime support for popcnt instruction"
@echo "x86-32 > x86 32-bit excluding old hardware without SSE-support"
@echo "x86-32-old > x86 32-bit including also very old hardware"
@echo "osx-ppc-64 > PPC-Mac OS X 64 bit"
@echo "osx-ppc-32 > PPC-Mac OS X 32 bit"
@echo "osx-x86-64 > x86-Mac OS X 64 bit"
@echo "osx-x86-32 > x86-Mac OS X 32 bit"
@echo "armv7 > ARMv7 32 bit"
@echo "general-64 > unspecified 64-bit"
@echo "general-32 > unspecified 32-bit"
@echo "x86-64 > x86 64-bit"
@echo "x86-64-modern > x86 64-bit with popcnt support"
@echo "x86-32 > x86 32-bit with SSE support"
@echo "x86-32-old > x86 32-bit fall back for old hardware"
@echo "linux-ppc-64 > PPC-Linux 64 bit"
@echo "osx-ppc-64 > PPC-Mac OS X 64 bit"
@echo "osx-ppc-32 > PPC-Mac OS X 32 bit"
@echo "osx-x86-64 > x86-Mac OS X 64 bit"
@echo "osx-x86-32 > x86-Mac OS X 32 bit"
@echo "armv7 > ARMv7 32 bit"
@echo "general-64 > unspecified 64-bit"
@echo "general-32 > unspecified 32-bit"
@echo ""
@echo "Supported comps:"
@echo "Supported compilers:"
@echo ""
@echo "gcc > Gnu compiler (default)"
@echo "icc > Intel compiler"
@echo "mingw > Gnu compiler with MinGW under Windows"
@echo "clang > LLVM Clang compiler"
@echo "gcc > Gnu compiler (default)"
@echo "mingw > Gnu compiler with MinGW under Windows"
@echo "clang > LLVM Clang compiler"
@echo "icc > Intel compiler"
@echo ""
@echo "Non-standard targets:"
@echo ""
@echo "make hpux > Compile for HP-UX. Compiler = aCC"
@echo "make hpux > Compile for HP-UX. Compiler = aCC"
@echo ""
@echo "Examples. If you don't know what to do, you likely want to run: "
@echo ""
@echo "make profile-build ARCH=x86-64 (This is for 64-bit systems)"
@echo "make profile-build ARCH=x86-32 (This is for 32-bit systems)"
@echo "make build ARCH=x86-64 (This is for 64-bit systems)"
@echo "make build ARCH=x86-32 (This is for 32-bit systems)"
@echo ""
.PHONY: build profile-build embed-signature
build:
$(MAKE) ARCH=$(ARCH) COMP=$(COMP) config-sanity
$(MAKE) ARCH=$(ARCH) COMP=$(COMP) all
@@ -437,6 +458,18 @@ profile-build:
@echo "Step 4/4. Deleting profile data ..."
$(MAKE) ARCH=$(ARCH) COMP=$(COMP) $(profile_clean)
embed-signature:
@echo "Running benchmark for getting the signature ..."
@$(SIGNBENCH) 2>&1 | sed -n 's/Nodes searched : \(.*\)/\/string Version\/s\/"\\(.*\\)"\/"sig-\1"\//p' > sign.txt
@sed -f sign.txt misc.cpp > misc2.cpp
@mv misc2.cpp misc.cpp
@rm sign.txt
signature-build: build embed-signature
$(MAKE) ARCH=$(ARCH) COMP=$(COMP) all
signature-profile-build: build embed-signature profile-build
strip:
strip $(EXE)
@@ -448,9 +481,6 @@ install:
clean:
$(RM) $(EXE) $(EXE).exe *.o .depend *~ core bench.txt *.gcda
testrun:
@$(PGOBENCH)
default:
help
+19 -5
View File
@@ -47,7 +47,21 @@ static const char* Defaults[] = {
"3r1rk1/p5pp/bpp1pp2/8/q1PP1P2/b3P3/P2NQRPP/1R2B1K1 b - - 6 22",
"r1q2rk1/2p1bppp/2Pp4/p6b/Q1PNp3/4B3/PP1R1PPP/2K4R w - - 2 18",
"4k2r/1pb2ppp/1p2p3/1R1p4/3P4/2r1PN2/P4PPP/1R4K1 b - - 3 22",
"3q2k1/pb3p1p/4pbp1/2r5/PpN2N2/1P2P2P/5PP1/Q2R2K1 b - - 4 26"
"3q2k1/pb3p1p/4pbp1/2r5/PpN2N2/1P2P2P/5PP1/Q2R2K1 b - - 4 26",
"6k1/6p1/6Pp/ppp5/3pn2P/1P3K2/1PP2P2/3N4 b - - 0 1",
"3b4/5kp1/1p1p1p1p/pP1PpP1P/P1P1P3/3KN3/8/8 w - - 0 1",
"2K5/p7/7P/5pR1/8/5k2/r7/8 w - - 0 1",
"8/6pk/1p6/8/PP3p1p/5P2/4KP1q/3Q4 w - - 0 1",
"7k/3p2pp/4q3/8/4Q3/5Kp1/P6b/8 w - - 0 1",
"8/2p5/8/2kPKp1p/2p4P/2P5/3P4/8 w - - 0 1",
"8/1p3pp1/7p/5P1P/2k3P1/8/2K2P2/8 w - - 0 1",
"8/pp2r1k1/2p1p3/3pP2p/1P1P1P1P/P5KR/8/8 w - - 0 1",
"8/3p4/p1bk3p/Pp6/1Kp1PpPp/2P2P1P/2P5/5B2 b - - 0 1",
"5k2/7R/4P2p/5K2/p1r2P1p/8/8/8 b - - 0 1",
"6k1/6p1/P6p/r1N5/5p2/7P/1b3PP1/4R1K1 w - - 0 1",
"1r3k2/4q3/2Pp3b/3Bp3/2Q2p2/1p1P2P1/1P2KP2/3N4 w - - 0 1",
"6k1/4pp1p/3p2p1/P1pPb3/R7/1r2P1PP/3B1P2/6K1 w - - 0 1",
"8/3p3B/5p2/5P2/p7/PP5b/k7/6K1 w - - 0 1"
};
@@ -68,7 +82,7 @@ void benchmark(const Position& current, istream& is) {
// Assign default values to missing arguments
string ttSize = (is >> token) ? token : "32";
string threads = (is >> token) ? token : "1";
string limit = (is >> token) ? token : "12";
string limit = (is >> token) ? token : "13";
string fenFile = (is >> token) ? token : "default";
string limitType = (is >> token) ? token : "depth";
@@ -89,7 +103,7 @@ void benchmark(const Position& current, istream& is) {
limits.depth = atoi(limit.c_str());
if (fenFile == "default")
fens.assign(Defaults, Defaults + 16);
fens.assign(Defaults, Defaults + 30);
else if (fenFile == "current")
fens.push_back(current.fen());
@@ -116,9 +130,9 @@ void benchmark(const Position& current, istream& is) {
Search::StateStackPtr st;
Time::point elapsed = Time::now();
for (size_t i = 0; i < fens.size(); i++)
for (size_t i = 0; i < fens.size(); ++i)
{
Position pos(fens[i], Options["UCI_Chess960"], Threads.main_thread());
Position pos(fens[i], Options["UCI_Chess960"], Threads.main());
cerr << "\nPosition: " << i + 1 << '/' << fens.size() << endl;
+30 -30
View File
@@ -39,9 +39,9 @@ namespace {
// bit 6-11: black king square (from SQ_A1 to SQ_H8)
// bit 12: side to move (WHITE or BLACK)
// bit 13-14: white pawn file (from FILE_A to FILE_D)
// bit 15-17: white pawn 6 - rank (from 6 - RANK_7 to 6 - RANK_2)
// bit 15-17: white pawn RANK_7 - rank (from RANK_7 - RANK_7 to RANK_7 - RANK_2)
unsigned index(Color us, Square bksq, Square wksq, Square psq) {
return wksq + (bksq << 6) + (us << 12) + (file_of(psq) << 13) + ((6 - rank_of(psq)) << 15);
return wksq + (bksq << 6) + (us << 12) + (file_of(psq) << 13) + ((RANK_7 - rank_of(psq)) << 15);
}
enum Result {
@@ -55,8 +55,8 @@ namespace {
struct KPKPosition {
operator Result() const { return res; }
Result classify_leaf(unsigned idx);
KPKPosition(unsigned idx);
operator Result() const { return result; }
Result classify(const std::vector<KPKPosition>& db)
{ return us == WHITE ? classify<WHITE>(db) : classify<BLACK>(db); }
@@ -65,7 +65,7 @@ namespace {
Color us;
Square bksq, wksq, psq;
Result res;
Result result;
};
} // namespace
@@ -83,21 +83,21 @@ bool Bitbases::probe_kpk(Square wksq, Square wpsq, Square bksq, Color us) {
void Bitbases::init_kpk() {
unsigned idx, repeat = 1;
std::vector<KPKPosition> db(IndexMax);
std::vector<KPKPosition> db;
db.reserve(IndexMax);
// Initialize db with known win / draw positions
for (idx = 0; idx < IndexMax; idx++)
db[idx].classify_leaf(idx);
for (idx = 0; idx < IndexMax; ++idx)
db.push_back(KPKPosition(idx));
// Iterate through the positions until no more of the unknown positions can be
// changed to either wins or draws (15 cycles needed).
while (repeat)
for (repeat = idx = 0; idx < IndexMax; idx++)
if (db[idx] == UNKNOWN && db[idx].classify(db) != UNKNOWN)
repeat = 1;
for (repeat = idx = 0; idx < IndexMax; ++idx)
repeat |= (db[idx] == UNKNOWN && db[idx].classify(db) != UNKNOWN);
// Map 32 results into one KPKBitbase[] entry
for (idx = 0; idx < IndexMax; idx++)
for (idx = 0; idx < IndexMax; ++idx)
if (db[idx] == WIN)
KPKBitbase[idx / 32] |= 1 << (idx & 0x1F);
}
@@ -105,34 +105,32 @@ void Bitbases::init_kpk() {
namespace {
Result KPKPosition::classify_leaf(unsigned idx) {
KPKPosition::KPKPosition(unsigned idx) {
wksq = Square((idx >> 0) & 0x3F);
bksq = Square((idx >> 6) & 0x3F);
us = Color((idx >> 12) & 0x01);
psq = File((idx >> 13) & 3) | Rank(6 - (idx >> 15));
wksq = Square((idx >> 0) & 0x3F);
bksq = Square((idx >> 6) & 0x3F);
us = Color ((idx >> 12) & 0x01);
psq = File ((idx >> 13) & 0x03) | Rank(RANK_7 - (idx >> 15));
result = UNKNOWN;
// Check if two pieces are on the same square or if a king can be captured
if ( wksq == psq || wksq == bksq || bksq == psq
|| (StepAttacksBB[KING][wksq] & bksq)
if ( square_distance(wksq, bksq) <= 1 || wksq == psq || bksq == psq
|| (us == WHITE && (StepAttacksBB[PAWN][psq] & bksq)))
return res = INVALID;
result = INVALID;
if (us == WHITE)
else if (us == WHITE)
{
// Immediate win if pawn can be promoted without getting captured
if ( rank_of(psq) == RANK_7
&& wksq != psq + DELTA_N
&& ( square_distance(bksq, psq + DELTA_N) > 1
||(StepAttacksBB[KING][wksq] & (psq + DELTA_N))))
return res = WIN;
result = WIN;
}
// Immediate draw if is stalemate or king captures undefended pawn
else if ( !(StepAttacksBB[KING][bksq] & ~(StepAttacksBB[KING][wksq] | StepAttacksBB[PAWN][psq]))
|| (StepAttacksBB[KING][bksq] & psq & ~StepAttacksBB[KING][wksq]))
return res = DRAW;
return res = UNKNOWN;
result = DRAW;
}
template<Color Us>
@@ -148,26 +146,28 @@ namespace {
// as WIN, the position is classified WIN otherwise the current position is
// classified UNKNOWN.
const Color Them = (Us == WHITE ? BLACK : WHITE);
Result r = INVALID;
Bitboard b = StepAttacksBB[KING][Us == WHITE ? wksq : bksq];
while (b)
r |= Us == WHITE ? db[index(~Us, bksq, pop_lsb(&b), psq)]
: db[index(~Us, pop_lsb(&b), wksq, psq)];
r |= Us == WHITE ? db[index(Them, bksq, pop_lsb(&b), psq)]
: db[index(Them, pop_lsb(&b), wksq, psq)];
if (Us == WHITE && rank_of(psq) < RANK_7)
{
Square s = psq + DELTA_N;
r |= db[index(BLACK, bksq, wksq, s)]; // Single push
if (rank_of(s) == RANK_3 && s != wksq && s != bksq)
if (rank_of(psq) == RANK_2 && s != wksq && s != bksq)
r |= db[index(BLACK, bksq, wksq, s + DELTA_N)]; // Double push
}
if (Us == WHITE)
return res = r & WIN ? WIN : r & UNKNOWN ? UNKNOWN : DRAW;
return result = r & WIN ? WIN : r & UNKNOWN ? UNKNOWN : DRAW;
else
return res = r & DRAW ? DRAW : r & UNKNOWN ? UNKNOWN : WIN;
return result = r & DRAW ? DRAW : r & UNKNOWN ? UNKNOWN : WIN;
}
} // namespace
+36 -38
View File
@@ -42,14 +42,14 @@ Bitboard SquareBB[SQUARE_NB];
Bitboard FileBB[FILE_NB];
Bitboard RankBB[RANK_NB];
Bitboard AdjacentFilesBB[FILE_NB];
Bitboard ThisAndAdjacentFilesBB[FILE_NB];
Bitboard InFrontBB[COLOR_NB][RANK_NB];
Bitboard StepAttacksBB[PIECE_NB][SQUARE_NB];
Bitboard BetweenBB[SQUARE_NB][SQUARE_NB];
Bitboard LineBB[SQUARE_NB][SQUARE_NB];
Bitboard DistanceRingsBB[SQUARE_NB][8];
Bitboard ForwardBB[COLOR_NB][SQUARE_NB];
Bitboard PassedPawnMask[COLOR_NB][SQUARE_NB];
Bitboard AttackSpanMask[COLOR_NB][SQUARE_NB];
Bitboard PawnAttackSpan[COLOR_NB][SQUARE_NB];
Bitboard PseudoAttacks[PIECE_TYPE_NB][SQUARE_NB];
int SquareDistance[SQUARE_NB][SQUARE_NB];
@@ -84,7 +84,7 @@ namespace {
/// lsb()/msb() finds the least/most significant bit in a nonzero bitboard.
/// pop_lsb() finds and clears the least significant bit in a nonzero bitboard.
#if !defined(USE_BSFQ)
#ifndef USE_BSFQ
Square lsb(Bitboard b) { return BSFTable[bsf_index(b)]; }
@@ -123,7 +123,7 @@ Square msb(Bitboard b) {
return (Square)(result + MS1BTable[b32]);
}
#endif // !defined(USE_BSFQ)
#endif // ifndef USE_BSFQ
/// Bitboards::print() prints a bitboard in an easily readable format to the
@@ -133,11 +133,11 @@ void Bitboards::print(Bitboard b) {
sync_cout;
for (Rank rank = RANK_8; rank >= RANK_1; rank--)
for (Rank rank = RANK_8; rank >= RANK_1; --rank)
{
std::cout << "+---+---+---+---+---+---+---+---+" << '\n';
for (File file = FILE_A; file <= FILE_H; file++)
for (File file = FILE_A; file <= FILE_H; ++file)
std::cout << "| " << (b & (file | rank) ? "X " : " ");
std::cout << "|\n";
@@ -151,59 +151,54 @@ void Bitboards::print(Bitboard b) {
void Bitboards::init() {
for (int k = 0, i = 0; i < 8; i++)
for (int k = 0, i = 0; i < 8; ++i)
while (k < (2 << i))
MS1BTable[k++] = i;
for (int i = 0; i < 64; i++)
for (int i = 0; i < 64; ++i)
BSFTable[bsf_index(1ULL << i)] = Square(i);
for (Square s = SQ_A1; s <= SQ_H8; s++)
for (Square s = SQ_A1; s <= SQ_H8; ++s)
SquareBB[s] = 1ULL << s;
FileBB[FILE_A] = FileABB;
RankBB[RANK_1] = Rank1BB;
for (int i = 1; i < 8; i++)
for (int i = 1; i < 8; ++i)
{
FileBB[i] = FileBB[i - 1] << 1;
RankBB[i] = RankBB[i - 1] << 8;
}
for (File f = FILE_A; f <= FILE_H; f++)
{
for (File f = FILE_A; f <= FILE_H; ++f)
AdjacentFilesBB[f] = (f > FILE_A ? FileBB[f - 1] : 0) | (f < FILE_H ? FileBB[f + 1] : 0);
ThisAndAdjacentFilesBB[f] = FileBB[f] | AdjacentFilesBB[f];
}
for (Rank r = RANK_1; r < RANK_8; r++)
for (Rank r = RANK_1; r < RANK_8; ++r)
InFrontBB[WHITE][r] = ~(InFrontBB[BLACK][r + 1] = InFrontBB[BLACK][r] | RankBB[r]);
for (Color c = WHITE; c <= BLACK; c++)
for (Square s = SQ_A1; s <= SQ_H8; s++)
for (Color c = WHITE; c <= BLACK; ++c)
for (Square s = SQ_A1; s <= SQ_H8; ++s)
{
ForwardBB[c][s] = InFrontBB[c][rank_of(s)] & FileBB[file_of(s)];
PassedPawnMask[c][s] = InFrontBB[c][rank_of(s)] & ThisAndAdjacentFilesBB[file_of(s)];
AttackSpanMask[c][s] = InFrontBB[c][rank_of(s)] & AdjacentFilesBB[file_of(s)];
PawnAttackSpan[c][s] = InFrontBB[c][rank_of(s)] & AdjacentFilesBB[file_of(s)];
PassedPawnMask[c][s] = ForwardBB[c][s] | PawnAttackSpan[c][s];
}
for (Square s1 = SQ_A1; s1 <= SQ_H8; s1++)
for (Square s2 = SQ_A1; s2 <= SQ_H8; s2++)
for (Square s1 = SQ_A1; s1 <= SQ_H8; ++s1)
for (Square s2 = SQ_A1; s2 <= SQ_H8; ++s2)
{
SquareDistance[s1][s2] = std::max(file_distance(s1, s2), rank_distance(s1, s2));
for (Square s1 = SQ_A1; s1 <= SQ_H8; s1++)
for (int d = 1; d < 8; d++)
for (Square s2 = SQ_A1; s2 <= SQ_H8; s2++)
if (SquareDistance[s1][s2] == d)
DistanceRingsBB[s1][d - 1] |= s2;
if (s1 != s2)
DistanceRingsBB[s1][SquareDistance[s1][s2] - 1] |= s2;
}
int steps[][9] = { {}, { 7, 9 }, { 17, 15, 10, 6, -6, -10, -15, -17 },
{}, {}, {}, { 9, 7, -7, -9, 8, 1, -1, -8 } };
for (Color c = WHITE; c <= BLACK; c++)
for (PieceType pt = PAWN; pt <= KING; pt++)
for (Square s = SQ_A1; s <= SQ_H8; s++)
for (int k = 0; steps[pt][k]; k++)
for (Color c = WHITE; c <= BLACK; ++c)
for (PieceType pt = PAWN; pt <= KING; ++pt)
for (Square s = SQ_A1; s <= SQ_H8; ++s)
for (int k = 0; steps[pt][k]; ++k)
{
Square to = s + Square(c == WHITE ? steps[pt][k] : -steps[pt][k]);
@@ -217,20 +212,23 @@ void Bitboards::init() {
init_magics(RTable, RAttacks, RMagics, RMasks, RShifts, RDeltas, magic_index<ROOK>);
init_magics(BTable, BAttacks, BMagics, BMasks, BShifts, BDeltas, magic_index<BISHOP>);
for (Square s = SQ_A1; s <= SQ_H8; s++)
for (Square s = SQ_A1; s <= SQ_H8; ++s)
{
PseudoAttacks[QUEEN][s] = PseudoAttacks[BISHOP][s] = attacks_bb<BISHOP>(s, 0);
PseudoAttacks[QUEEN][s] |= PseudoAttacks[ ROOK][s] = attacks_bb< ROOK>(s, 0);
}
for (Square s1 = SQ_A1; s1 <= SQ_H8; s1++)
for (Square s2 = SQ_A1; s2 <= SQ_H8; s2++)
for (Square s1 = SQ_A1; s1 <= SQ_H8; ++s1)
for (Square s2 = SQ_A1; s2 <= SQ_H8; ++s2)
if (PseudoAttacks[QUEEN][s1] & s2)
{
Square delta = (s2 - s1) / square_distance(s1, s2);
for (Square s = s1 + delta; s != s2; s += delta)
BetweenBB[s1][s2] |= s;
PieceType pt = (PseudoAttacks[BISHOP][s1] & s2) ? BISHOP : ROOK;
LineBB[s1][s2] = (PseudoAttacks[pt][s1] & PseudoAttacks[pt][s2]) | s1 | s2;
}
}
@@ -241,7 +239,7 @@ namespace {
Bitboard attack = 0;
for (int i = 0; i < 4; i++)
for (int i = 0; i < 4; ++i)
for (Square s = sq + deltas[i];
is_ok(s) && square_distance(s, s - deltas[i]) == 1;
s += deltas[i])
@@ -287,7 +285,7 @@ namespace {
// attacks[s] is a pointer to the beginning of the attacks table for square 's'
attacks[SQ_A1] = table;
for (Square s = SQ_A1; s <= SQ_H8; s++)
for (Square s = SQ_A1; s <= SQ_H8; ++s)
{
// Board edges are not considered in the relevant occupancies
edges = ((Rank1BB | Rank8BB) & ~rank_bb(s)) | ((FileABB | FileHBB) & ~file_bb(s));
@@ -322,13 +320,13 @@ namespace {
do magics[s] = pick_random(rk, booster);
while (popcount<Max15>((magics[s] * masks[s]) >> 56) < 6);
memset(attacks[s], 0, size * sizeof(Bitboard));
std::memset(attacks[s], 0, size * sizeof(Bitboard));
// A good magic must map every possible occupancy to an index that
// looks up the correct sliding attack in the attacks[s] database.
// Note that we build up the database for square 's' as a side
// effect of verifying the magic.
for (i = 0; i < size; i++)
for (i = 0; i < size; ++i)
{
Bitboard& attack = attacks[s][index(s, occupancy[i])];
+83 -50
View File
@@ -18,7 +18,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(BITBOARD_H_INCLUDED)
#ifndef BITBOARD_H_INCLUDED
#define BITBOARD_H_INCLUDED
#include "types.h"
@@ -37,6 +37,24 @@ bool probe_kpk(Square wksq, Square wpsq, Square bksq, Color us);
}
const Bitboard FileABB = 0x0101010101010101ULL;
const Bitboard FileBBB = FileABB << 1;
const Bitboard FileCBB = FileABB << 2;
const Bitboard FileDBB = FileABB << 3;
const Bitboard FileEBB = FileABB << 4;
const Bitboard FileFBB = FileABB << 5;
const Bitboard FileGBB = FileABB << 6;
const Bitboard FileHBB = FileABB << 7;
const Bitboard Rank1BB = 0xFF;
const Bitboard Rank2BB = Rank1BB << (8 * 1);
const Bitboard Rank3BB = Rank1BB << (8 * 2);
const Bitboard Rank4BB = Rank1BB << (8 * 3);
const Bitboard Rank5BB = Rank1BB << (8 * 4);
const Bitboard Rank6BB = Rank1BB << (8 * 5);
const Bitboard Rank7BB = Rank1BB << (8 * 6);
const Bitboard Rank8BB = Rank1BB << (8 * 7);
CACHE_LINE_ALIGNMENT
extern Bitboard RMasks[SQUARE_NB];
@@ -53,17 +71,19 @@ extern Bitboard SquareBB[SQUARE_NB];
extern Bitboard FileBB[FILE_NB];
extern Bitboard RankBB[RANK_NB];
extern Bitboard AdjacentFilesBB[FILE_NB];
extern Bitboard ThisAndAdjacentFilesBB[FILE_NB];
extern Bitboard InFrontBB[COLOR_NB][RANK_NB];
extern Bitboard StepAttacksBB[PIECE_NB][SQUARE_NB];
extern Bitboard BetweenBB[SQUARE_NB][SQUARE_NB];
extern Bitboard LineBB[SQUARE_NB][SQUARE_NB];
extern Bitboard DistanceRingsBB[SQUARE_NB][8];
extern Bitboard ForwardBB[COLOR_NB][SQUARE_NB];
extern Bitboard PassedPawnMask[COLOR_NB][SQUARE_NB];
extern Bitboard AttackSpanMask[COLOR_NB][SQUARE_NB];
extern Bitboard PawnAttackSpan[COLOR_NB][SQUARE_NB];
extern Bitboard PseudoAttacks[PIECE_TYPE_NB][SQUARE_NB];
const Bitboard BlackSquares = 0xAA55AA55AA55AA55ULL;
extern int SquareDistance[SQUARE_NB][SQUARE_NB];
const Bitboard DarkSquares = 0xAA55AA55AA55AA55ULL;
/// Overloads of bitwise operators between a Bitboard and a Square for testing
/// whether a given bit is set in a bitboard, and for setting and clearing bits.
@@ -88,13 +108,34 @@ inline Bitboard operator^(Bitboard b, Square s) {
return b ^ SquareBB[s];
}
/// more_than_one() returns true if in 'b' there is more than one bit set
inline bool more_than_one(Bitboard b) {
return b & (b - 1);
}
inline int square_distance(Square s1, Square s2) {
return SquareDistance[s1][s2];
}
inline int file_distance(Square s1, Square s2) {
return abs(file_of(s1) - file_of(s2));
}
inline int rank_distance(Square s1, Square s2) {
return abs(rank_of(s1) - rank_of(s2));
}
/// shift_bb() moves bitboard one step along direction Delta. Mainly for pawns.
template<Square Delta>
inline Bitboard shift_bb(Bitboard b) {
return Delta == DELTA_N ? b << 8 : Delta == DELTA_S ? b >> 8
: Delta == DELTA_NE ? (b & ~FileHBB) << 9 : Delta == DELTA_SE ? (b & ~FileHBB) >> 7
: Delta == DELTA_NW ? (b & ~FileABB) << 7 : Delta == DELTA_SW ? (b & ~FileABB) >> 9
: 0;
}
/// rank_bb() and file_bb() take a file or a square as input and return
/// a bitboard representing all squares on the given file or rank.
@@ -116,7 +157,7 @@ inline Bitboard file_bb(Square s) {
}
/// adjacent_files_bb takes a file as input and returns a bitboard representing
/// adjacent_files_bb() takes a file as input and returns a bitboard representing
/// all squares on the adjacent files.
inline Bitboard adjacent_files_bb(File f) {
@@ -124,30 +165,17 @@ inline Bitboard adjacent_files_bb(File f) {
}
/// this_and_adjacent_files_bb takes a file as input and returns a bitboard
/// representing all squares on the given and adjacent files.
inline Bitboard this_and_adjacent_files_bb(File f) {
return ThisAndAdjacentFilesBB[f];
}
/// in_front_bb() takes a color and a rank or square as input, and returns a
/// bitboard representing all the squares on all ranks in front of the rank
/// (or square), from the given color's point of view. For instance,
/// in_front_bb(WHITE, RANK_5) will give all squares on ranks 6, 7 and 8, while
/// in_front_bb(BLACK, SQ_D3) will give all squares on ranks 1 and 2.
/// in_front_bb() takes a color and a rank as input, and returns a bitboard
/// representing all the squares on all ranks in front of the rank, from the
/// given color's point of view. For instance, in_front_bb(BLACK, RANK_3) will
/// give all squares on ranks 1 and 2.
inline Bitboard in_front_bb(Color c, Rank r) {
return InFrontBB[c][r];
}
inline Bitboard in_front_bb(Color c, Square s) {
return InFrontBB[c][rank_of(s)];
}
/// between_bb returns a bitboard representing all squares between two squares.
/// between_bb() returns a bitboard representing all squares between two squares.
/// For instance, between_bb(SQ_C4, SQ_F7) returns a bitboard with the bits for
/// square d5 and e6 set. If s1 and s2 are not on the same line, file or diagonal,
/// 0 is returned.
@@ -157,7 +185,7 @@ inline Bitboard between_bb(Square s1, Square s2) {
}
/// forward_bb takes a color and a square as input, and returns a bitboard
/// forward_bb() takes a color and a square as input, and returns a bitboard
/// representing all squares along the line in front of the square, from the
/// point of view of the given color. Definition of the table is:
/// ForwardBB[c][s] = in_front_bb(c, s) & file_bb(s)
@@ -167,40 +195,39 @@ inline Bitboard forward_bb(Color c, Square s) {
}
/// passed_pawn_mask takes a color and a square as input, and returns a
/// pawn_attack_span() takes a color and a square as input, and returns a bitboard
/// representing all squares that can be attacked by a pawn of the given color
/// when it moves along its file starting from the given square. Definition is:
/// PawnAttackSpan[c][s] = in_front_bb(c, s) & adjacent_files_bb(s);
inline Bitboard pawn_attack_span(Color c, Square s) {
return PawnAttackSpan[c][s];
}
/// passed_pawn_mask() takes a color and a square as input, and returns a
/// bitboard mask which can be used to test if a pawn of the given color on
/// the given square is a passed pawn. Definition of the table is:
/// PassedPawnMask[c][s] = in_front_bb(c, s) & this_and_adjacent_files_bb(s)
/// PassedPawnMask[c][s] = pawn_attack_span(c, s) | forward_bb(c, s)
inline Bitboard passed_pawn_mask(Color c, Square s) {
return PassedPawnMask[c][s];
}
/// attack_span_mask takes a color and a square as input, and returns a bitboard
/// representing all squares that can be attacked by a pawn of the given color
/// when it moves along its file starting from the given square. Definition is:
/// AttackSpanMask[c][s] = in_front_bb(c, s) & adjacent_files_bb(s);
/// squares_of_color() returns a bitboard representing all squares with the same
/// color of the given square.
inline Bitboard attack_span_mask(Color c, Square s) {
return AttackSpanMask[c][s];
inline Bitboard squares_of_color(Square s) {
return DarkSquares & s ? DarkSquares : ~DarkSquares;
}
/// squares_aligned returns true if the squares s1, s2 and s3 are aligned
/// aligned() returns true if the squares s1, s2 and s3 are aligned
/// either on a straight or on a diagonal line.
inline bool squares_aligned(Square s1, Square s2, Square s3) {
return (BetweenBB[s1][s2] | BetweenBB[s1][s3] | BetweenBB[s2][s3])
& ( SquareBB[s1] | SquareBB[s2] | SquareBB[s3]);
}
/// same_color_squares() returns a bitboard representing all squares with
/// the same color of the given square.
inline Bitboard same_color_squares(Square s) {
return BlackSquares & s ? BlackSquares : ~BlackSquares;
inline bool aligned(Square s1, Square s2, Square s3) {
return LineBB[s1][s2] & s3;
}
@@ -231,7 +258,7 @@ inline Bitboard attacks_bb(Square s, Bitboard occ) {
/// lsb()/msb() finds the least/most significant bit in a nonzero bitboard.
/// pop_lsb() finds and clears the least significant bit in a nonzero bitboard.
#if defined(USE_BSFQ)
#ifdef USE_BSFQ
# if defined(_MSC_VER) && !defined(__INTEL_COMPILER)
@@ -284,7 +311,7 @@ FORCE_INLINE Square pop_lsb(Bitboard* b) {
return s;
}
#else // if !defined(USE_BSFQ)
#else // if defined(USE_BSFQ)
extern Square msb(Bitboard b);
extern Square lsb(Bitboard b);
@@ -292,4 +319,10 @@ extern Square pop_lsb(Bitboard* b);
#endif
#endif // !defined(BITBOARD_H_INCLUDED)
/// frontmost_sq() and backmost_sq() find the square corresponding to the
/// most/least advanced bit relative to the given color.
inline Square frontmost_sq(Color c, Bitboard b) { return c == WHITE ? msb(b) : lsb(b); }
inline Square backmost_sq(Color c, Bitboard b) { return c == WHITE ? lsb(b) : msb(b); }
#endif // #ifndef BITBOARD_H_INCLUDED
+3 -3
View File
@@ -18,7 +18,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(BITCOUNT_H_INCLUDED)
#ifndef BITCOUNT_H_INCLUDED
#define BITCOUNT_H_INCLUDED
#include <cassert>
@@ -81,7 +81,7 @@ inline int popcount<CNT_32_MAX15>(Bitboard b) {
template<>
inline int popcount<CNT_HW_POPCNT>(Bitboard b) {
#if !defined(USE_POPCNT)
#ifndef USE_POPCNT
assert(false);
return b != 0; // Avoid 'b not used' warning
@@ -102,4 +102,4 @@ inline int popcount<CNT_HW_POPCNT>(Bitboard b) {
#endif
}
#endif // !defined(BITCOUNT_H_INCLUDED)
#endif // #ifndef BITCOUNT_H_INCLUDED
+5 -5
View File
@@ -361,7 +361,7 @@ PolyglotBook::~PolyglotBook() { if (is_open()) close(); }
template<typename T> PolyglotBook& PolyglotBook::operator>>(T& n) {
n = 0;
for (size_t i = 0; i < sizeof(T); i++)
for (size_t i = 0; i < sizeof(T); ++i)
n = T((n << 8) + ifstream::get());
return *this;
@@ -413,7 +413,7 @@ Move PolyglotBook::probe(const Position& pos, const string& fName, bool pickBest
// Choose book move according to its score. If a move has a very
// high score it has higher probability to be choosen than a move
// with lower score. Note that first entry is always chosen.
if ( (sum && rkiss.rand<unsigned>() % sum < e.count)
if ( (!pickBest && sum && rkiss.rand<unsigned>() % sum < e.count)
|| (pickBest && e.count == best))
move = Move(e.move);
}
@@ -436,9 +436,9 @@ Move PolyglotBook::probe(const Position& pos, const string& fName, bool pickBest
move = make<PROMOTION>(from_sq(move), to_sq(move), PieceType(pt + 1));
// Add 'special move' flags and verify it is legal
for (MoveList<LEGAL> ml(pos); !ml.end(); ++ml)
if (move == (ml.move() ^ type_of(ml.move())))
return ml.move();
for (MoveList<LEGAL> it(pos); *it; ++it)
if (move == (*it ^ type_of(*it)))
return *it;
return MOVE_NONE;
}
+2 -2
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(BOOK_H_INCLUDED)
#ifndef BOOK_H_INCLUDED
#define BOOK_H_INCLUDED
#include <fstream>
@@ -42,4 +42,4 @@ private:
std::string fileName;
};
#endif // !defined(BOOK_H_INCLUDED)
#endif // #ifndef BOOK_H_INCLUDED
+314 -397
View File
File diff suppressed because it is too large Load Diff
+7 -6
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(ENDGAME_H_INCLUDED)
#ifndef ENDGAME_H_INCLUDED
#define ENDGAME_H_INCLUDED
#include <map>
@@ -33,6 +33,7 @@ enum EndgameType {
// Evaluation functions
KNNK, // KNN vs K
KXK, // Generic "mate lone king" eval
KBNK, // KBN vs K
KPK, // KP vs K
@@ -42,7 +43,6 @@ enum EndgameType {
KQKP, // KQ vs KP
KQKR, // KQ vs KR
KBBKN, // KBB vs KN
KNNK, // KNN vs K
KmmKm, // K and two minors vs K and one or two minors
@@ -52,6 +52,7 @@ enum EndgameType {
KBPsK, // KB+pawns vs K
KQKRPs, // KQ vs KR+pawns
KRPKR, // KRP vs KR
KRPKB, // KRP vs KB
KRPPKRP, // KRPP vs KRP
KPsK, // King and pawns vs king
KBPKB, // KBP vs KB
@@ -85,12 +86,12 @@ struct EndgameBase {
template<EndgameType E, typename T = typename eg_fun<(E > SCALE_FUNS)>::type>
struct Endgame : public EndgameBase<T> {
explicit Endgame(Color c) : strongerSide(c), weakerSide(~c) {}
Color color() const { return strongerSide; }
explicit Endgame(Color c) : strongSide(c), weakSide(~c) {}
Color color() const { return strongSide; }
T operator()(const Position&) const;
private:
Color strongerSide, weakerSide;
const Color strongSide, weakSide;
};
@@ -119,4 +120,4 @@ public:
{ return eg = map(eg).count(key) ? map(eg)[key] : NULL; }
};
#endif // !defined(ENDGAME_H_INCLUDED)
#endif // #ifndef ENDGAME_H_INCLUDED
+398 -575
View File
File diff suppressed because it is too large Load Diff
+3 -3
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(EVALUATE_H_INCLUDED)
#ifndef EVALUATE_H_INCLUDED
#define EVALUATE_H_INCLUDED
#include "types.h"
@@ -27,9 +27,9 @@ class Position;
namespace Eval {
extern void init();
extern Value evaluate(const Position& pos, Value& margin);
extern Value evaluate(const Position& pos);
extern std::string trace(const Position& pos);
}
#endif // !defined(EVALUATE_H_INCLUDED)
#endif // #ifndef EVALUATE_H_INCLUDED
+3 -2
View File
@@ -34,16 +34,17 @@ int main(int argc, char* argv[]) {
UCI::init(Options);
Bitboards::init();
Zobrist::init();
Position::init();
Bitbases::init_kpk();
Search::init();
Pawns::init();
Eval::init();
Threads.init();
TT.set_size(Options["Hash"]);
std::string args;
for (int i = 1; i < argc; i++)
for (int i = 1; i < argc; ++i)
args += std::string(argv[i]) + " ";
UCI::loop(args);
+45 -52
View File
@@ -35,31 +35,29 @@ namespace {
const int NoPawnsSF[4] = { 6, 12, 32 };
// Polynomial material balance parameters
const Value RedundantQueenPenalty = Value(320);
const Value RedundantRookPenalty = Value(554);
// pair pawn knight bishop rook queen
const int LinearCoefficients[6] = { 1617, -162, -1172, -190, 105, 26 };
const int LinearCoefficients[6] = { 1852, -162, -1122, -183, 249, -52 };
const int QuadraticCoefficientsSameColor[][PIECE_TYPE_NB] = {
// pair pawn knight bishop rook queen
{ 7 }, // Bishop pair
{ 0 }, // Bishop pair
{ 39, 2 }, // Pawn
{ 35, 271, -4 }, // Knight
{ 7, 105, 4, 7 }, // Bishop
{ -27, -2, 46, 100, 56 }, // Rook
{ 58, 29, 83, 148, -3, -25 } // Queen
{ 0, 105, 4, 0 }, // Bishop
{ -27, -2, 46, 100, -141 }, // Rook
{ 58, 29, 83, 148, -163, 0 } // Queen
};
const int QuadraticCoefficientsOppositeColor[][PIECE_TYPE_NB] = {
// THEIR PIECES
// pair pawn knight bishop rook queen
{ 41 }, // Bishop pair
{ 37, 41 }, // Pawn
{ 10, 62, 41 }, // Knight OUR PIECES
{ 57, 64, 39, 41 }, // Bishop
{ 50, 40, 23, -22, 41 }, // Rook
{ 106, 101, 3, 151, 171, 41 } // Queen
{ 0 }, // Bishop pair
{ 37, 0 }, // Pawn
{ 10, 62, 0 }, // Knight OUR PIECES
{ 57, 64, 39, 0 }, // Bishop
{ 50, 40, 23, -22, 0 }, // Rook
{ 106, 101, 3, 151, 171, 0 } // Queen
};
// Endgame evaluation and scaling functions accessed direcly and not through
@@ -75,24 +73,24 @@ namespace {
// Helper templates used to detect a given material distribution
template<Color Us> bool is_KXK(const Position& pos) {
const Color Them = (Us == WHITE ? BLACK : WHITE);
return pos.non_pawn_material(Them) == VALUE_ZERO
&& pos.piece_count(Them, PAWN) == 0
&& pos.non_pawn_material(Us) >= RookValueMg;
return !pos.count<PAWN>(Them)
&& pos.non_pawn_material(Them) == VALUE_ZERO
&& pos.non_pawn_material(Us) >= RookValueMg;
}
template<Color Us> bool is_KBPsKs(const Position& pos) {
return pos.non_pawn_material(Us) == BishopValueMg
&& pos.piece_count(Us, BISHOP) == 1
&& pos.piece_count(Us, PAWN) >= 1;
return pos.non_pawn_material(Us) == BishopValueMg
&& pos.count<BISHOP>(Us) == 1
&& pos.count<PAWN >(Us) >= 1;
}
template<Color Us> bool is_KQKRPs(const Position& pos) {
const Color Them = (Us == WHITE ? BLACK : WHITE);
return pos.piece_count(Us, PAWN) == 0
&& pos.non_pawn_material(Us) == QueenValueMg
&& pos.piece_count(Us, QUEEN) == 1
&& pos.piece_count(Them, ROOK) == 1
&& pos.piece_count(Them, PAWN) >= 1;
return !pos.count<PAWN>(Us)
&& pos.non_pawn_material(Us) == QueenValueMg
&& pos.count<QUEEN>(Us) == 1
&& pos.count<ROOK>(Them) == 1
&& pos.count<PAWN>(Them) >= 1;
}
/// imbalance() calculates imbalance comparing piece count of each
@@ -106,14 +104,8 @@ namespace {
int pt1, pt2, pc, v;
int value = 0;
// Redundancy of major pieces, formula based on Kaufman's paper
// "The Evaluation of Material Imbalances in Chess"
if (pieceCount[Us][ROOK] > 0)
value -= RedundantRookPenalty * (pieceCount[Us][ROOK] - 1)
+ RedundantQueenPenalty * pieceCount[Us][QUEEN];
// Second-degree polynomial material imbalance by Tord Romstad
for (pt1 = NO_PIECE_TYPE; pt1 <= QUEEN; pt1++)
for (pt1 = NO_PIECE_TYPE; pt1 <= QUEEN; ++pt1)
{
pc = pieceCount[Us][pt1];
if (!pc)
@@ -121,7 +113,7 @@ namespace {
v = LinearCoefficients[pt1];
for (pt2 = NO_PIECE_TYPE; pt2 <= pt1; pt2++)
for (pt2 = NO_PIECE_TYPE; pt2 <= pt1; ++pt2)
v += QuadraticCoefficientsSameColor[pt1][pt2] * pieceCount[Us][pt2]
+ QuadraticCoefficientsOppositeColor[pt1][pt2] * pieceCount[Them][pt2];
@@ -150,7 +142,7 @@ Entry* probe(const Position& pos, Table& entries, Endgames& endgames) {
if (e->key == key)
return e;
memset(e, 0, sizeof(Entry));
std::memset(e, 0, sizeof(Entry));
e->key = key;
e->factor[WHITE] = e->factor[BLACK] = (uint8_t)SCALE_FACTOR_NORMAL;
e->gamePhase = game_phase(pos);
@@ -180,8 +172,8 @@ Entry* probe(const Position& pos, Table& entries, Endgames& endgames) {
assert((pos.pieces(WHITE, KNIGHT) | pos.pieces(WHITE, BISHOP)));
assert((pos.pieces(BLACK, KNIGHT) | pos.pieces(BLACK, BISHOP)));
if ( pos.piece_count(WHITE, BISHOP) + pos.piece_count(WHITE, KNIGHT) <= 2
&& pos.piece_count(BLACK, BISHOP) + pos.piece_count(BLACK, KNIGHT) <= 2)
if ( pos.count<BISHOP>(WHITE) + pos.count<KNIGHT>(WHITE) <= 2
&& pos.count<BISHOP>(BLACK) + pos.count<KNIGHT>(BLACK) <= 2)
{
e->evaluationFunction = &EvaluateKmmKm[pos.side_to_move()];
return e;
@@ -221,17 +213,17 @@ Entry* probe(const Position& pos, Table& entries, Endgames& endgames) {
if (npm_w + npm_b == VALUE_ZERO)
{
if (pos.piece_count(BLACK, PAWN) == 0)
if (!pos.count<PAWN>(BLACK))
{
assert(pos.piece_count(WHITE, PAWN) >= 2);
assert(pos.count<PAWN>(WHITE) >= 2);
e->scalingFunction[WHITE] = &ScaleKPsK[WHITE];
}
else if (pos.piece_count(WHITE, PAWN) == 0)
else if (!pos.count<PAWN>(WHITE))
{
assert(pos.piece_count(BLACK, PAWN) >= 2);
assert(pos.count<PAWN>(BLACK) >= 2);
e->scalingFunction[BLACK] = &ScaleKPsK[BLACK];
}
else if (pos.piece_count(WHITE, PAWN) == 1 && pos.piece_count(BLACK, PAWN) == 1)
else if (pos.count<PAWN>(WHITE) == 1 && pos.count<PAWN>(BLACK) == 1)
{
// This is a special case because we set scaling functions
// for both colors instead of only one.
@@ -240,36 +232,37 @@ Entry* probe(const Position& pos, Table& entries, Endgames& endgames) {
}
}
// No pawns makes it difficult to win, even with a material advantage
if (pos.piece_count(WHITE, PAWN) == 0 && npm_w - npm_b <= BishopValueMg)
// No pawns makes it difficult to win, even with a material advantage. This
// catches some trivial draws like KK, KBK and KNK
if (!pos.count<PAWN>(WHITE) && npm_w - npm_b <= BishopValueMg)
{
e->factor[WHITE] = (uint8_t)
(npm_w == npm_b || npm_w < RookValueMg ? 0 : NoPawnsSF[std::min(pos.piece_count(WHITE, BISHOP), 2)]);
(npm_w == npm_b || npm_w < RookValueMg ? 0 : NoPawnsSF[std::min(pos.count<BISHOP>(WHITE), 2)]);
}
if (pos.piece_count(BLACK, PAWN) == 0 && npm_b - npm_w <= BishopValueMg)
if (!pos.count<PAWN>(BLACK) && npm_b - npm_w <= BishopValueMg)
{
e->factor[BLACK] = (uint8_t)
(npm_w == npm_b || npm_b < RookValueMg ? 0 : NoPawnsSF[std::min(pos.piece_count(BLACK, BISHOP), 2)]);
(npm_w == npm_b || npm_b < RookValueMg ? 0 : NoPawnsSF[std::min(pos.count<BISHOP>(BLACK), 2)]);
}
// Compute the space weight
if (npm_w + npm_b >= 2 * QueenValueMg + 4 * RookValueMg + 2 * KnightValueMg)
{
int minorPieceCount = pos.piece_count(WHITE, KNIGHT) + pos.piece_count(WHITE, BISHOP)
+ pos.piece_count(BLACK, KNIGHT) + pos.piece_count(BLACK, BISHOP);
int minorPieceCount = pos.count<KNIGHT>(WHITE) + pos.count<BISHOP>(WHITE)
+ pos.count<KNIGHT>(BLACK) + pos.count<BISHOP>(BLACK);
e->spaceWeight = minorPieceCount * minorPieceCount;
e->spaceWeight = make_score(minorPieceCount * minorPieceCount, 0);
}
// Evaluate the material imbalance. We use PIECE_TYPE_NONE as a place holder
// for the bishop pair "extended piece", this allow us to be more flexible
// in defining bishop pair bonuses.
const int pieceCount[COLOR_NB][PIECE_TYPE_NB] = {
{ 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) } };
{ pos.count<BISHOP>(WHITE) > 1, pos.count<PAWN>(WHITE), pos.count<KNIGHT>(WHITE),
pos.count<BISHOP>(WHITE) , pos.count<ROOK>(WHITE), pos.count<QUEEN >(WHITE) },
{ pos.count<BISHOP>(BLACK) > 1, pos.count<PAWN>(BLACK), pos.count<KNIGHT>(BLACK),
pos.count<BISHOP>(BLACK) , pos.count<ROOK>(BLACK), pos.count<QUEEN >(BLACK) } };
e->value = (int16_t)((imbalance<WHITE>(pieceCount) - imbalance<BLACK>(pieceCount)) / 16);
return e;
+4 -4
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(MATERIAL_H_INCLUDED)
#ifndef MATERIAL_H_INCLUDED
#define MATERIAL_H_INCLUDED
#include "endgame.h"
@@ -39,7 +39,7 @@ namespace Material {
struct Entry {
Score material_value() const { return make_score(value, value); }
int space_weight() const { return spaceWeight; }
Score space_weight() const { return spaceWeight; }
Phase game_phase() const { return gamePhase; }
bool specialized_eval_exists() const { return evaluationFunction != NULL; }
Value evaluate(const Position& p) const { return (*evaluationFunction)(p); }
@@ -50,7 +50,7 @@ struct Entry {
uint8_t factor[COLOR_NB];
EndgameBase<Value>* evaluationFunction;
EndgameBase<ScaleFactor>* scalingFunction[COLOR_NB];
int spaceWeight;
Score spaceWeight;
Phase gamePhase;
};
@@ -74,4 +74,4 @@ inline ScaleFactor Entry::scale_factor(const Position& pos, Color c) const {
}
#endif // !defined(MATERIAL_H_INCLUDED)
#endif // #ifndef MATERIAL_H_INCLUDED
+13 -56
View File
@@ -24,17 +24,11 @@
#include "misc.h"
#include "thread.h"
#if defined(__hpux)
# include <sys/pstat.h>
#endif
using namespace std;
/// Version number. If Version is left empty, then Tag plus current
/// date, in the format DD-MM-YY, are used as a version number.
static const string Version = "3";
static const string Tag = "";
/// Version number. If Version is left empty, then compile date, in the
/// format DD-MM-YY, is shown in engine_info.
static const string Version = "DD";
/// engine_info() returns the full name of the current Stockfish version. This
@@ -45,43 +39,33 @@ static const string Tag = "";
const string engine_info(bool to_uci) {
const string months("Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec");
const string cpu64(Is64Bit ? " 64bit" : "");
const string popcnt(HasPopCnt ? " SSE4.2" : "");
string month, day, year;
stringstream s, date(__DATE__); // From compiler, format is "Sep 21 2008"
s << "Stockfish " << Version;
s << "Stockfish " << Version << setfill('0');
if (Version.empty())
{
date >> month >> day >> year;
s << Tag << string(Tag.empty() ? "" : " ") << setfill('0') << setw(2) << day
<< "-" << setw(2) << (1 + months.find(month) / 4) << "-" << year.substr(2);
s << setw(2) << day << setw(2) << (1 + months.find(month) / 4) << year.substr(2);
}
s << cpu64 << popcnt << (to_uci ? "\nid author ": " by ")
s << (Is64Bit ? " 64" : "")
<< (HasPopCnt ? " SSE4.2" : "")
<< (to_uci ? "\nid author ": " by ")
<< "Tord Romstad, Marco Costalba and Joona Kiiski";
return s.str();
}
/// Convert system time to milliseconds. That's all we need.
Time::point Time::now() {
sys_time_t t; system_time(&t); return time_to_msec(t);
}
/// Debug functions used mainly to collect run-time statistics
static uint64_t hits[2], means[2];
void dbg_hit_on(bool b) { hits[0]++; if (b) hits[1]++; }
void dbg_hit_on(bool b) { ++hits[0]; if (b) ++hits[1]; }
void dbg_hit_on_c(bool c, bool b) { if (c) dbg_hit_on(b); }
void dbg_mean_of(int v) { means[0]++; means[1] += v; }
void dbg_mean_of(int v) { ++means[0]; means[1] += v; }
void dbg_print() {
@@ -91,7 +75,7 @@ void dbg_print() {
if (means[0])
cerr << "Total " << means[0] << " Mean "
<< (float)means[1] / means[0] << endl;
<< (double)means[1] / means[0] << endl;
}
@@ -174,37 +158,12 @@ std::ostream& operator<<(std::ostream& os, SyncCout sc) {
void start_logger(bool b) { Logger::start(b); }
/// cpu_count() tries to detect the number of CPU cores
int cpu_count() {
#if defined(_WIN32) || defined(_WIN64)
SYSTEM_INFO s;
GetSystemInfo(&s);
return s.dwNumberOfProcessors;
#else
# if defined(_SC_NPROCESSORS_ONLN)
return sysconf(_SC_NPROCESSORS_ONLN);
# elif defined(__hpux)
struct pst_dynamic psd;
if (pstat_getdynamic(&psd, sizeof(psd), (size_t)1, 0) == -1)
return 1;
return psd.psd_proc_cnt;
# else
return 1;
# endif
#endif
}
/// timed_wait() waits for msec milliseconds. It is mainly an helper to wrap
/// conversion from milliseconds to struct timespec, as used by pthreads.
void timed_wait(WaitCondition& sleepCond, Lock& sleepLock, int msec) {
#if defined(_WIN32) || defined(_WIN64)
#ifdef _WIN32
int tm = msec;
#else
timespec ts, *tm = &ts;
@@ -221,7 +180,7 @@ void timed_wait(WaitCondition& sleepCond, Lock& sleepLock, int msec) {
/// prefetch() preloads the given address in L1/L2 cache. This is a non
/// blocking function and do not stalls the CPU waiting for data to be
/// loaded from memory, that can be quite slow.
#if defined(NO_PREFETCH)
#ifdef NO_PREFETCH
void prefetch(char*) {}
@@ -237,10 +196,8 @@ void prefetch(char* addr) {
# if defined(__INTEL_COMPILER) || defined(_MSC_VER)
_mm_prefetch(addr, _MM_HINT_T0);
_mm_prefetch(addr+64, _MM_HINT_T0); // 64 bytes ahead
# else
__builtin_prefetch(addr);
__builtin_prefetch(addr+64);
# endif
}
+3 -4
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(MISC_H_INCLUDED)
#ifndef MISC_H_INCLUDED
#define MISC_H_INCLUDED
#include <fstream>
@@ -27,7 +27,6 @@
#include "types.h"
extern const std::string engine_info(bool to_uci = false);
extern int cpu_count();
extern void timed_wait(WaitCondition&, Lock&, int);
extern void prefetch(char* addr);
extern void start_logger(bool b);
@@ -46,7 +45,7 @@ struct Log : public std::ofstream {
namespace Time {
typedef int64_t point;
point now();
inline point now() { return system_time_to_msec(); }
}
@@ -66,4 +65,4 @@ std::ostream& operator<<(std::ostream&, SyncCout);
#define sync_cout std::cout << io_lock
#define sync_endl std::endl << io_unlock
#endif // !defined(MISC_H_INCLUDED)
#endif // #ifndef MISC_H_INCLUDED
+76 -105
View File
@@ -24,15 +24,15 @@
/// Simple macro to wrap a very common while loop, no facny, no flexibility,
/// hardcoded names 'mlist' and 'from'.
#define SERIALIZE(b) while (b) (*mlist++).move = make_move(from, pop_lsb(&b))
#define SERIALIZE(b) while (b) (mlist++)->move = make_move(from, pop_lsb(&b))
/// Version used for pawns, where the 'from' square is given as a delta from the 'to' square
#define SERIALIZE_PAWNS(b, d) while (b) { Square to = pop_lsb(&b); \
(*mlist++).move = make_move(to - (d), to); }
(mlist++)->move = make_move(to - (d), to); }
namespace {
template<CastlingSide Side, bool Checks, bool Chess960>
MoveStack* generate_castle(const Position& pos, MoveStack* mlist, Color us) {
ExtMove* generate_castle(const Position& pos, ExtMove* mlist, Color us) {
if (pos.castle_impeded(us, Side) || !pos.can_castle(make_castle_right(us, Side)))
return mlist;
@@ -56,54 +56,42 @@ namespace {
// Because we generate only legal castling moves we need to verify that
// when moving the castling rook we do not discover some hidden checker.
// For instance an enemy queen in SQ_A1 when castling rook is in SQ_B1.
if (Chess960 && (pos.attackers_to(kto, pos.pieces() ^ rfrom) & enemies))
if (Chess960 && (attacks_bb<ROOK>(kto, pos.pieces() ^ rfrom) & pos.pieces(~us, ROOK, QUEEN)))
return mlist;
(*mlist++).move = make<CASTLE>(kfrom, rfrom);
(mlist++)->move = make<CASTLE>(kfrom, rfrom);
if (Checks && !pos.move_gives_check((mlist - 1)->move, CheckInfo(pos)))
mlist--;
if (Checks && !pos.gives_check((mlist - 1)->move, CheckInfo(pos)))
--mlist;
return mlist;
}
template<Square Delta>
inline Bitboard move_pawns(Bitboard p) {
return Delta == DELTA_N ? p << 8
: Delta == DELTA_S ? p >> 8
: Delta == DELTA_NE ? (p & ~FileHBB) << 9
: Delta == DELTA_SE ? (p & ~FileHBB) >> 7
: Delta == DELTA_NW ? (p & ~FileABB) << 7
: Delta == DELTA_SW ? (p & ~FileABB) >> 9 : 0;
}
template<GenType Type, Square Delta>
inline MoveStack* generate_promotions(MoveStack* mlist, Bitboard pawnsOn7,
Bitboard target, const CheckInfo* ci) {
inline ExtMove* generate_promotions(ExtMove* mlist, Bitboard pawnsOn7,
Bitboard target, const CheckInfo* ci) {
Bitboard b = move_pawns<Delta>(pawnsOn7) & target;
Bitboard b = shift_bb<Delta>(pawnsOn7) & target;
while (b)
{
Square to = pop_lsb(&b);
if (Type == CAPTURES || Type == EVASIONS || Type == NON_EVASIONS)
(*mlist++).move = make<PROMOTION>(to - Delta, to, QUEEN);
(mlist++)->move = make<PROMOTION>(to - Delta, to, QUEEN);
if (Type == QUIETS || Type == EVASIONS || Type == NON_EVASIONS)
{
(*mlist++).move = make<PROMOTION>(to - Delta, to, ROOK);
(*mlist++).move = make<PROMOTION>(to - Delta, to, BISHOP);
(*mlist++).move = make<PROMOTION>(to - Delta, to, KNIGHT);
(mlist++)->move = make<PROMOTION>(to - Delta, to, ROOK);
(mlist++)->move = make<PROMOTION>(to - Delta, to, BISHOP);
(mlist++)->move = make<PROMOTION>(to - Delta, to, KNIGHT);
}
// Knight-promotion is the only one that can give a direct check not
// already included in the queen-promotion.
if (Type == QUIET_CHECKS && (StepAttacksBB[W_KNIGHT][to] & ci->ksq))
(*mlist++).move = make<PROMOTION>(to - Delta, to, KNIGHT);
(mlist++)->move = make<PROMOTION>(to - Delta, to, KNIGHT);
else
(void)ci; // Silence a warning under MSVC
}
@@ -113,8 +101,8 @@ namespace {
template<Color Us, GenType Type>
MoveStack* generate_pawn_moves(const Position& pos, MoveStack* mlist,
Bitboard target, const CheckInfo* ci) {
ExtMove* generate_pawn_moves(const Position& pos, ExtMove* mlist,
Bitboard target, const CheckInfo* ci) {
// Compute our parametrized parameters at compile time, named according to
// the point of view of white side.
@@ -122,9 +110,9 @@ namespace {
const Bitboard TRank8BB = (Us == WHITE ? Rank8BB : Rank1BB);
const Bitboard TRank7BB = (Us == WHITE ? Rank7BB : Rank2BB);
const Bitboard TRank3BB = (Us == WHITE ? Rank3BB : Rank6BB);
const Square UP = (Us == WHITE ? DELTA_N : DELTA_S);
const Square RIGHT = (Us == WHITE ? DELTA_NE : DELTA_SW);
const Square LEFT = (Us == WHITE ? DELTA_NW : DELTA_SE);
const Square Up = (Us == WHITE ? DELTA_N : DELTA_S);
const Square Right = (Us == WHITE ? DELTA_NE : DELTA_SW);
const Square Left = (Us == WHITE ? DELTA_NW : DELTA_SE);
Bitboard b1, b2, dc1, dc2, emptySquares;
@@ -139,8 +127,8 @@ namespace {
{
emptySquares = (Type == QUIETS || Type == QUIET_CHECKS ? target : ~pos.pieces());
b1 = move_pawns<UP>(pawnsNotOn7) & emptySquares;
b2 = move_pawns<UP>(b1 & TRank3BB) & emptySquares;
b1 = shift_bb<Up>(pawnsNotOn7) & emptySquares;
b2 = shift_bb<Up>(b1 & TRank3BB) & emptySquares;
if (Type == EVASIONS) // Consider only blocking squares
{
@@ -159,16 +147,16 @@ namespace {
// promotion has been already generated among captures.
if (pawnsNotOn7 & ci->dcCandidates)
{
dc1 = move_pawns<UP>(pawnsNotOn7 & ci->dcCandidates) & emptySquares & ~file_bb(ci->ksq);
dc2 = move_pawns<UP>(dc1 & TRank3BB) & emptySquares;
dc1 = shift_bb<Up>(pawnsNotOn7 & ci->dcCandidates) & emptySquares & ~file_bb(ci->ksq);
dc2 = shift_bb<Up>(dc1 & TRank3BB) & emptySquares;
b1 |= dc1;
b2 |= dc2;
}
}
SERIALIZE_PAWNS(b1, UP);
SERIALIZE_PAWNS(b2, UP + UP);
SERIALIZE_PAWNS(b1, Up);
SERIALIZE_PAWNS(b2, Up + Up);
}
// Promotions and underpromotions
@@ -180,19 +168,19 @@ namespace {
if (Type == EVASIONS)
emptySquares &= target;
mlist = generate_promotions<Type, RIGHT>(mlist, pawnsOn7, enemies, ci);
mlist = generate_promotions<Type, LEFT>(mlist, pawnsOn7, enemies, ci);
mlist = generate_promotions<Type, UP>(mlist, pawnsOn7, emptySquares, ci);
mlist = generate_promotions<Type, Right>(mlist, pawnsOn7, enemies, ci);
mlist = generate_promotions<Type, Left >(mlist, pawnsOn7, enemies, ci);
mlist = generate_promotions<Type, Up>(mlist, pawnsOn7, emptySquares, ci);
}
// Standard and en-passant captures
if (Type == CAPTURES || Type == EVASIONS || Type == NON_EVASIONS)
{
b1 = move_pawns<RIGHT>(pawnsNotOn7) & enemies;
b2 = move_pawns<LEFT >(pawnsNotOn7) & enemies;
b1 = shift_bb<Right>(pawnsNotOn7) & enemies;
b2 = shift_bb<Left >(pawnsNotOn7) & enemies;
SERIALIZE_PAWNS(b1, RIGHT);
SERIALIZE_PAWNS(b2, LEFT);
SERIALIZE_PAWNS(b1, Right);
SERIALIZE_PAWNS(b2, Left);
if (pos.ep_square() != SQ_NONE)
{
@@ -201,7 +189,7 @@ namespace {
// An en passant capture can be an evasion only if the checking piece
// is the double pushed pawn and so is in the target. Otherwise this
// is a discovery check and we are forced to do otherwise.
if (Type == EVASIONS && !(target & (pos.ep_square() - UP)))
if (Type == EVASIONS && !(target & (pos.ep_square() - Up)))
return mlist;
b1 = pawnsNotOn7 & pos.attacks_from<PAWN>(pos.ep_square(), Them);
@@ -209,7 +197,7 @@ namespace {
assert(b1);
while (b1)
(*mlist++).move = make<ENPASSANT>(pop_lsb(&b1), pos.ep_square());
(mlist++)->move = make<ENPASSANT>(pop_lsb(&b1), pos.ep_square());
}
}
@@ -218,12 +206,12 @@ namespace {
template<PieceType Pt, bool Checks> FORCE_INLINE
MoveStack* generate_moves(const Position& pos, MoveStack* mlist, Color us,
Bitboard target, const CheckInfo* ci) {
ExtMove* generate_moves(const Position& pos, ExtMove* mlist, Color us,
Bitboard target, const CheckInfo* ci) {
assert(Pt != KING && Pt != PAWN);
const Square* pl = pos.piece_list(us, Pt);
const Square* pl = pos.list<Pt>(us);
for (Square from = *pl; from != SQ_NONE; from = *++pl)
{
@@ -233,7 +221,7 @@ namespace {
&& !(PseudoAttacks[Pt][from] & target & ci->checkSq[Pt]))
continue;
if (ci->dcCandidates && (ci->dcCandidates & from))
if (unlikely(ci->dcCandidates) && (ci->dcCandidates & from))
continue;
}
@@ -249,38 +237,36 @@ namespace {
}
template<GenType Type> FORCE_INLINE
MoveStack* generate_all(const Position& pos, MoveStack* mlist, Color us,
Bitboard target, const CheckInfo* ci = NULL) {
template<Color Us, GenType Type> FORCE_INLINE
ExtMove* generate_all(const Position& pos, ExtMove* mlist, Bitboard target,
const CheckInfo* ci = NULL) {
const bool Checks = Type == QUIET_CHECKS;
mlist = (us == WHITE ? generate_pawn_moves<WHITE, Type>(pos, mlist, target, ci)
: generate_pawn_moves<BLACK, Type>(pos, mlist, target, ci));
mlist = generate_moves<KNIGHT, Checks>(pos, mlist, us, target, ci);
mlist = generate_moves<BISHOP, Checks>(pos, mlist, us, target, ci);
mlist = generate_moves<ROOK, Checks>(pos, mlist, us, target, ci);
mlist = generate_moves<QUEEN, Checks>(pos, mlist, us, target, ci);
mlist = generate_pawn_moves<Us, Type>(pos, mlist, target, ci);
mlist = generate_moves<KNIGHT, Checks>(pos, mlist, Us, target, ci);
mlist = generate_moves<BISHOP, Checks>(pos, mlist, Us, target, ci);
mlist = generate_moves< ROOK, Checks>(pos, mlist, Us, target, ci);
mlist = generate_moves< QUEEN, Checks>(pos, mlist, Us, target, ci);
if (Type != QUIET_CHECKS && Type != EVASIONS)
{
Square from = pos.king_square(us);
Square from = pos.king_square(Us);
Bitboard b = pos.attacks_from<KING>(from) & target;
SERIALIZE(b);
}
if (Type != CAPTURES && Type != EVASIONS && pos.can_castle(us))
if (Type != CAPTURES && Type != EVASIONS && pos.can_castle(Us))
{
if (pos.is_chess960())
{
mlist = generate_castle<KING_SIDE, Checks, true>(pos, mlist, us);
mlist = generate_castle<QUEEN_SIDE, Checks, true>(pos, mlist, us);
mlist = generate_castle< KING_SIDE, Checks, true>(pos, mlist, Us);
mlist = generate_castle<QUEEN_SIDE, Checks, true>(pos, mlist, Us);
}
else
{
mlist = generate_castle<KING_SIDE, Checks, false>(pos, mlist, us);
mlist = generate_castle<QUEEN_SIDE, Checks, false>(pos, mlist, us);
mlist = generate_castle< KING_SIDE, Checks, false>(pos, mlist, Us);
mlist = generate_castle<QUEEN_SIDE, Checks, false>(pos, mlist, Us);
}
}
@@ -301,7 +287,7 @@ namespace {
/// non-captures. Returns a pointer to the end of the move list.
template<GenType Type>
MoveStack* generate(const Position& pos, MoveStack* mlist) {
ExtMove* generate(const Position& pos, ExtMove* mlist) {
assert(Type == CAPTURES || Type == QUIETS || Type == NON_EVASIONS);
assert(!pos.checkers());
@@ -312,22 +298,24 @@ MoveStack* generate(const Position& pos, MoveStack* mlist) {
: Type == QUIETS ? ~pos.pieces()
: Type == NON_EVASIONS ? ~pos.pieces(us) : 0;
return generate_all<Type>(pos, mlist, us, target);
return us == WHITE ? generate_all<WHITE, Type>(pos, mlist, target)
: generate_all<BLACK, Type>(pos, mlist, target);
}
// Explicit template instantiations
template MoveStack* generate<CAPTURES>(const Position&, MoveStack*);
template MoveStack* generate<QUIETS>(const Position&, MoveStack*);
template MoveStack* generate<NON_EVASIONS>(const Position&, MoveStack*);
template ExtMove* generate<CAPTURES>(const Position&, ExtMove*);
template ExtMove* generate<QUIETS>(const Position&, ExtMove*);
template ExtMove* generate<NON_EVASIONS>(const Position&, ExtMove*);
/// generate<QUIET_CHECKS> generates all pseudo-legal non-captures and knight
/// underpromotions that give check. Returns a pointer to the end of the move list.
template<>
MoveStack* generate<QUIET_CHECKS>(const Position& pos, MoveStack* mlist) {
ExtMove* generate<QUIET_CHECKS>(const Position& pos, ExtMove* mlist) {
assert(!pos.checkers());
Color us = pos.side_to_move();
CheckInfo ci(pos);
Bitboard dc = ci.dcCandidates;
@@ -347,21 +335,21 @@ MoveStack* generate<QUIET_CHECKS>(const Position& pos, MoveStack* mlist) {
SERIALIZE(b);
}
return generate_all<QUIET_CHECKS>(pos, mlist, pos.side_to_move(), ~pos.pieces(), &ci);
return us == WHITE ? generate_all<WHITE, QUIET_CHECKS>(pos, mlist, ~pos.pieces(), &ci)
: generate_all<BLACK, QUIET_CHECKS>(pos, mlist, ~pos.pieces(), &ci);
}
/// generate<EVASIONS> generates all pseudo-legal check evasions when the side
/// to move is in check. Returns a pointer to the end of the move list.
template<>
MoveStack* generate<EVASIONS>(const Position& pos, MoveStack* mlist) {
ExtMove* generate<EVASIONS>(const Position& pos, ExtMove* mlist) {
assert(pos.checkers());
Square from, checksq;
int checkersCnt = 0;
Color us = pos.side_to_move();
Square ksq = pos.king_square(us);
Square ksq = pos.king_square(us), from = ksq /* For SERIALIZE */, checksq;
Bitboard sliderAttacks = 0;
Bitboard b = pos.checkers();
@@ -371,65 +359,48 @@ MoveStack* generate<EVASIONS>(const Position& pos, MoveStack* mlist) {
// evasions so to skip known illegal moves avoiding useless legality check later.
do
{
checkersCnt++;
++checkersCnt;
checksq = pop_lsb(&b);
assert(color_of(pos.piece_on(checksq)) == ~us);
switch (type_of(pos.piece_on(checksq)))
{
case BISHOP: sliderAttacks |= PseudoAttacks[BISHOP][checksq]; break;
case ROOK: sliderAttacks |= PseudoAttacks[ROOK][checksq]; break;
case QUEEN:
// If queen and king are far or not on a diagonal line we can safely
// remove all the squares attacked in the other direction becuase are
// not reachable by the king anyway.
if (between_bb(ksq, checksq) || !(PseudoAttacks[BISHOP][checksq] & ksq))
sliderAttacks |= PseudoAttacks[QUEEN][checksq];
if (type_of(pos.piece_on(checksq)) > KNIGHT) // A slider
sliderAttacks |= LineBB[checksq][ksq] ^ checksq;
// Otherwise we need to use real rook attacks to check if king is safe
// to move in the other direction. For example: king in B2, queen in A1
// a knight in B1, and we can safely move to C1.
else
sliderAttacks |= PseudoAttacks[BISHOP][checksq] | pos.attacks_from<ROOK>(checksq);
default:
break;
}
} while (b);
// Generate evasions for king, capture and non capture moves
b = pos.attacks_from<KING>(ksq) & ~pos.pieces(us) & ~sliderAttacks;
from = ksq;
SERIALIZE(b);
if (checkersCnt > 1)
return mlist; // Double check, only a king move can save the day
// Generate blocking evasions or captures of the checking piece
Bitboard target = between_bb(checksq, ksq) | pos.checkers();
Bitboard target = between_bb(checksq, ksq) | checksq;
return generate_all<EVASIONS>(pos, mlist, us, target);
return us == WHITE ? generate_all<WHITE, EVASIONS>(pos, mlist, target)
: generate_all<BLACK, EVASIONS>(pos, mlist, target);
}
/// generate<LEGAL> generates all the legal moves in the given position
template<>
MoveStack* generate<LEGAL>(const Position& pos, MoveStack* mlist) {
ExtMove* generate<LEGAL>(const Position& pos, ExtMove* mlist) {
MoveStack *end, *cur = mlist;
Bitboard pinned = pos.pinned_pieces();
ExtMove *end, *cur = mlist;
Bitboard pinned = pos.pinned_pieces(pos.side_to_move());
Square ksq = pos.king_square(pos.side_to_move());
end = pos.checkers() ? generate<EVASIONS>(pos, mlist)
: generate<NON_EVASIONS>(pos, mlist);
while (cur != end)
if ( (pinned || from_sq(cur->move) == ksq || type_of(cur->move) == ENPASSANT)
&& !pos.pl_move_is_legal(cur->move, pinned))
&& !pos.legal(cur->move, pinned))
cur->move = (--end)->move;
else
cur++;
++cur;
return end;
}
+9 -10
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(MOVEGEN_H_INCLUDED)
#ifndef MOVEGEN_H_INCLUDED
#define MOVEGEN_H_INCLUDED
#include "types.h"
@@ -34,26 +34,25 @@ enum GenType {
class Position;
template<GenType>
MoveStack* generate(const Position& pos, MoveStack* mlist);
ExtMove* generate(const Position& pos, ExtMove* mlist);
/// The MoveList struct is a simple wrapper around generate(), sometimes comes
/// handy to use this class instead of the low level generate() function.
template<GenType T>
struct MoveList {
explicit MoveList(const Position& pos) : cur(mlist), last(generate<T>(pos, mlist)) {}
void operator++() { cur++; }
bool end() const { return cur == last; }
Move move() const { return cur->move; }
explicit MoveList(const Position& pos) : cur(mlist), last(generate<T>(pos, mlist)) { last->move = MOVE_NONE; }
void operator++() { ++cur; }
Move operator*() const { return cur->move; }
size_t size() const { return last - mlist; }
bool contains(Move m) const {
for (const MoveStack* it(mlist); it != last; ++it) if (it->move == m) return true;
for (const ExtMove* it(mlist); it != last; ++it) if (it->move == m) return true;
return false;
}
private:
MoveStack mlist[MAX_MOVES];
MoveStack *cur, *last;
ExtMove mlist[MAX_MOVES];
ExtMove *cur, *last;
};
#endif // !defined(MOVEGEN_H_INCLUDED)
#endif // #ifndef MOVEGEN_H_INCLUDED
+56 -56
View File
@@ -25,7 +25,7 @@
namespace {
enum Sequencer {
enum Stages {
MAIN_SEARCH, CAPTURES_S1, KILLERS_S1, QUIETS_1_S1, QUIETS_2_S1, BAD_CAPTURES_S1,
EVASION, EVASIONS_S2,
QSEARCH_0, CAPTURES_S3, QUIET_CHECKS_S3,
@@ -36,9 +36,9 @@ namespace {
};
// Our insertion sort, guaranteed to be stable, as is needed
void insertion_sort(MoveStack* begin, MoveStack* end)
void insertion_sort(ExtMove* begin, ExtMove* end)
{
MoveStack tmp, *p, *q;
ExtMove tmp, *p, *q;
for (p = begin + 1; p < end; ++p)
{
@@ -51,12 +51,12 @@ namespace {
// Unary predicate used by std::partition to split positive scores from remaining
// ones so to sort separately the two sets, and with the second sort delayed.
inline bool has_positive_score(const MoveStack& ms) { return ms.score > 0; }
inline bool has_positive_score(const ExtMove& ms) { return ms.score > 0; }
// Picks and moves to the front the best move in the range [begin, end),
// it is faster than sorting all the moves in advance when moves are few, as
// normally are the possible captures.
inline MoveStack* pick_best(MoveStack* begin, MoveStack* end)
inline ExtMove* pick_best(ExtMove* begin, ExtMove* end)
{
std::swap(*begin, *std::max_element(begin, end));
return begin;
@@ -70,83 +70,70 @@ namespace {
/// search captures, promotions and some checks) and about how important good
/// move ordering is at the current node.
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const History& h,
Search::Stack* s, Value beta) : pos(p), Hist(h), depth(d) {
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const HistoryStats& h,
Move* cm, Search::Stack* s) : pos(p), history(h), depth(d) {
assert(d > DEPTH_ZERO);
captureThreshold = 0;
cur = end = moves;
endBadCaptures = moves + MAX_MOVES - 1;
countermoves = cm;
ss = s;
if (p.checkers())
phase = EVASION;
stage = EVASION;
else
{
phase = MAIN_SEARCH;
stage = MAIN_SEARCH;
killers[0].move = ss->killers[0];
killers[1].move = ss->killers[1];
// Consider sligtly negative captures as good if at low depth and far from beta
if (ss && ss->staticEval < beta - PawnValueMg && d < 3 * ONE_PLY)
captureThreshold = -PawnValueMg;
// Consider negative captures as good if still enough to reach beta
else if (ss && ss->staticEval > beta)
captureThreshold = beta - ss->staticEval;
}
ttMove = (ttm && pos.is_pseudo_legal(ttm) ? ttm : MOVE_NONE);
ttMove = (ttm && pos.pseudo_legal(ttm) ? ttm : MOVE_NONE);
end += (ttMove != MOVE_NONE);
}
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const History& h,
Square sq) : pos(p), Hist(h), cur(moves), end(moves) {
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const HistoryStats& h,
Square sq) : pos(p), history(h), cur(moves), end(moves) {
assert(d <= DEPTH_ZERO);
if (p.checkers())
phase = EVASION;
stage = EVASION;
else if (d > DEPTH_QS_NO_CHECKS)
phase = QSEARCH_0;
stage = QSEARCH_0;
else if (d > DEPTH_QS_RECAPTURES)
{
phase = QSEARCH_1;
stage = QSEARCH_1;
// 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 && !pos.is_capture_or_promotion(ttm))
if (ttm && !pos.capture_or_promotion(ttm))
ttm = MOVE_NONE;
}
else
{
phase = RECAPTURE;
stage = RECAPTURE;
recaptureSquare = sq;
ttm = MOVE_NONE;
}
ttMove = (ttm && pos.is_pseudo_legal(ttm) ? ttm : MOVE_NONE);
ttMove = (ttm && pos.pseudo_legal(ttm) ? ttm : MOVE_NONE);
end += (ttMove != MOVE_NONE);
}
MovePicker::MovePicker(const Position& p, Move ttm, const History& h, PieceType pt)
: pos(p), Hist(h), cur(moves), end(moves) {
MovePicker::MovePicker(const Position& p, Move ttm, const HistoryStats& h, PieceType pt)
: pos(p), history(h), cur(moves), end(moves) {
assert(!pos.checkers());
phase = PROBCUT;
stage = PROBCUT;
// In ProbCut we generate only captures better than parent's captured piece
captureThreshold = PieceValue[MG][pt];
ttMove = (ttm && pos.is_pseudo_legal(ttm) ? ttm : MOVE_NONE);
ttMove = (ttm && pos.pseudo_legal(ttm) ? ttm : MOVE_NONE);
if (ttMove && (!pos.is_capture(ttMove) || pos.see(ttMove) <= captureThreshold))
if (ttMove && (!pos.capture(ttMove) || pos.see(ttMove) <= captureThreshold))
ttMove = MOVE_NONE;
end += (ttMove != MOVE_NONE);
@@ -172,11 +159,11 @@ void MovePicker::score<CAPTURES>() {
// some SEE calls in case we get a cutoff (idea from Pablo Vazquez).
Move m;
for (MoveStack* it = moves; it != end; ++it)
for (ExtMove* it = moves; it != end; ++it)
{
m = it->move;
it->score = PieceValue[MG][pos.piece_on(to_sq(m))]
- type_of(pos.piece_moved(m));
- type_of(pos.moved_piece(m));
if (type_of(m) == PROMOTION)
it->score += PieceValue[MG][promotion_type(m)] - PieceValue[MG][PAWN];
@@ -191,10 +178,10 @@ void MovePicker::score<QUIETS>() {
Move m;
for (MoveStack* it = moves; it != end; ++it)
for (ExtMove* it = moves; it != end; ++it)
{
m = it->move;
it->score = Hist[pos.piece_moved(m)][to_sq(m)];
it->score = history[pos.moved_piece(m)][to_sq(m)];
}
}
@@ -206,17 +193,17 @@ void MovePicker::score<EVASIONS>() {
Move m;
int seeScore;
for (MoveStack* it = moves; it != end; ++it)
for (ExtMove* it = moves; it != end; ++it)
{
m = it->move;
if ((seeScore = pos.see_sign(m)) < 0)
it->score = seeScore - History::Max; // At the bottom
it->score = seeScore - HistoryStats::Max; // At the bottom
else if (pos.is_capture(m))
else if (pos.capture(m))
it->score = PieceValue[MG][pos.piece_on(to_sq(m))]
- type_of(pos.piece_moved(m)) + History::Max;
- type_of(pos.moved_piece(m)) + HistoryStats::Max;
else
it->score = Hist[pos.piece_moved(m)][to_sq(m)];
it->score = history[pos.moved_piece(m)][to_sq(m)];
}
}
@@ -228,7 +215,7 @@ void MovePicker::generate_next() {
cur = moves;
switch (++phase) {
switch (++stage) {
case CAPTURES_S1: case CAPTURES_S3: case CAPTURES_S4: case CAPTURES_S5: case CAPTURES_S6:
end = generate<CAPTURES>(pos, moves);
@@ -238,6 +225,19 @@ void MovePicker::generate_next() {
case KILLERS_S1:
cur = killers;
end = cur + 2;
killers[0].move = ss->killers[0];
killers[1].move = ss->killers[1];
killers[2].move = killers[3].move = MOVE_NONE;
// Be sure countermoves are different from killers
for (int i = 0; i < 2; ++i)
if (countermoves[i] != cur->move && countermoves[i] != (cur+1)->move)
(end++)->move = countermoves[i];
if (countermoves[1] && countermoves[1] == countermoves[0]) // Due to SMP races
killers[3].move = MOVE_NONE;
return;
case QUIETS_1_S1:
@@ -271,7 +271,7 @@ void MovePicker::generate_next() {
return;
case EVASION: case QSEARCH_0: case QSEARCH_1: case PROBCUT: case RECAPTURE:
phase = STOP;
stage = STOP;
case STOP:
end = cur + 1; // Avoid another next_phase() call
return;
@@ -296,19 +296,17 @@ Move MovePicker::next_move<false>() {
while (cur == end)
generate_next();
switch (phase) {
switch (stage) {
case MAIN_SEARCH: case EVASION: case QSEARCH_0: case QSEARCH_1: case PROBCUT:
cur++;
++cur;
return ttMove;
case CAPTURES_S1:
move = pick_best(cur++, end)->move;
if (move != ttMove)
{
assert(captureThreshold <= 0); // Otherwise we cannot use see_sign()
if (pos.see_sign(move) >= captureThreshold)
if (pos.see_sign(move) >= 0)
return move;
// Losing capture, move it to the tail of the array
@@ -319,9 +317,9 @@ Move MovePicker::next_move<false>() {
case KILLERS_S1:
move = (cur++)->move;
if ( move != MOVE_NONE
&& pos.is_pseudo_legal(move)
&& pos.pseudo_legal(move)
&& move != ttMove
&& !pos.is_capture(move))
&& !pos.capture(move))
return move;
break;
@@ -329,7 +327,9 @@ Move MovePicker::next_move<false>() {
move = (cur++)->move;
if ( move != ttMove
&& move != killers[0].move
&& move != killers[1].move)
&& move != killers[1].move
&& move != killers[2].move
&& move != killers[3].move)
return move;
break;
+36 -24
View File
@@ -17,11 +17,11 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined MOVEPICK_H_INCLUDED
#ifndef MOVEPICK_H_INCLUDED
#define MOVEPICK_H_INCLUDED
#include <algorithm> // For std::max
#include <cstring> // For memset
#include <cstring> // For std::memset
#include "movegen.h"
#include "position.h"
@@ -30,20 +30,29 @@
/// The Stats struct stores moves statistics. According to the template parameter
/// the class can store both History and Gains type statistics. History records
/// how often different moves have been successful or unsuccessful during the
/// current search and is used for reduction and move ordering decisions. Gains
/// records the move's best evaluation gain from one ply to the next and is used
/// for pruning decisions. Entries are stored according only to moving piece and
/// destination square, in particular two moves with different origin but same
/// destination and same piece will be considered identical.
template<bool Gain>
/// the class can store History, Gains and Countermoves. History records how often
/// different moves have been successful or unsuccessful during the current search
/// and is used for reduction and move ordering decisions. Gains records the move's
/// best evaluation gain from one ply to the next and is used for pruning decisions.
/// Countermoves store the move that refute a previous one. Entries are stored
/// according only to moving piece and destination square, hence two moves with
/// different origin but same destination and piece will be considered identical.
template<bool Gain, typename T>
struct Stats {
static const Value Max = Value(2000);
const Value* operator[](Piece p) const { return &table[p][0]; }
void clear() { memset(table, 0, sizeof(table)); }
const T* operator[](Piece p) const { return table[p]; }
void clear() { std::memset(table, 0, sizeof(table)); }
void update(Piece p, Square to, Move m) {
if (m == table[p][to].first)
return;
table[p][to].second = table[p][to].first;
table[p][to].first = m;
}
void update(Piece p, Square to, Value v) {
@@ -55,11 +64,12 @@ struct Stats {
}
private:
Value table[PIECE_NB][SQUARE_NB];
T table[PIECE_NB][SQUARE_NB];
};
typedef Stats<false> History;
typedef Stats<true> Gains;
typedef Stats< true, Value> GainsStats;
typedef Stats<false, Value> HistoryStats;
typedef Stats<false, std::pair<Move, Move> > CountermovesStats;
/// MovePicker class is used to pick one pseudo legal move at a time from the
@@ -74,9 +84,10 @@ class MovePicker {
MovePicker& operator=(const MovePicker&); // Silence a warning under MSVC
public:
MovePicker(const Position&, Move, Depth, const History&, Search::Stack*, Value);
MovePicker(const Position&, Move, Depth, const History&, Square);
MovePicker(const Position&, Move, const History&, PieceType);
MovePicker(const Position&, Move, Depth, const HistoryStats&, Square);
MovePicker(const Position&, Move, const HistoryStats&, PieceType);
MovePicker(const Position&, Move, Depth, const HistoryStats&, Move*, Search::Stack*);
template<bool SpNode> Move next_move();
private:
@@ -84,15 +95,16 @@ private:
void generate_next();
const Position& pos;
const History& Hist;
const HistoryStats& history;
Search::Stack* ss;
Move* countermoves;
Depth depth;
Move ttMove;
MoveStack killers[2];
ExtMove killers[4];
Square recaptureSquare;
int captureThreshold, phase;
MoveStack *cur, *end, *endQuiets, *endBadCaptures;
MoveStack moves[MAX_MOVES];
int captureThreshold, stage;
ExtMove *cur, *end, *endQuiets, *endBadCaptures;
ExtMove moves[MAX_MOVES];
};
#endif // !defined(MOVEPICK_H_INCLUDED)
#endif // #ifndef MOVEPICK_H_INCLUDED
+8 -8
View File
@@ -89,9 +89,9 @@ Move move_from_uci(const Position& pos, string& str) {
if (str.length() == 5) // Junior could send promotion piece in uppercase
str[4] = char(tolower(str[4]));
for (MoveList<LEGAL> ml(pos); !ml.end(); ++ml)
if (str == move_to_uci(ml.move(), pos.is_chess960()))
return ml.move();
for (MoveList<LEGAL> it(pos); *it; ++it)
if (str == move_to_uci(*it, pos.is_chess960()))
return *it;
return MOVE_NONE;
}
@@ -133,7 +133,7 @@ const string move_to_san(Position& pos, Move m) {
while (b)
{
Move move = make_move(pop_lsb(&b), to);
if (!pos.pl_move_is_legal(move, pos.pinned_pieces()))
if (!pos.legal(move, pos.pinned_pieces(pos.side_to_move())))
others ^= from_sq(move);
}
@@ -149,10 +149,10 @@ const string move_to_san(Position& pos, Move m) {
san += square_to_string(from);
}
}
else if (pos.is_capture(m))
else if (pos.capture(m))
san = file_to_char(file_of(from));
if (pos.is_capture(m))
if (pos.capture(m))
san += 'x';
san += square_to_string(to);
@@ -161,7 +161,7 @@ const string move_to_san(Position& pos, Move m) {
san += string("=") + PieceToChar[WHITE][promotion_type(m)];
}
if (pos.move_gives_check(m, CheckInfo(pos)))
if (pos.gives_check(m, CheckInfo(pos)))
{
StateInfo st;
pos.do_move(m, st);
@@ -207,7 +207,7 @@ static string score_to_string(Value v) {
s << "-#" << (VALUE_MATE + v) / 2;
else
s << setprecision(2) << fixed << showpos << float(v) / PawnValueMg;
s << setprecision(2) << fixed << showpos << double(v) / PawnValueMg;
return s.str();
}
+2 -2
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(NOTATION_H_INCLUDED)
#ifndef NOTATION_H_INCLUDED
#define NOTATION_H_INCLUDED
#include <string>
@@ -32,4 +32,4 @@ const std::string move_to_uci(Move m, bool chess960);
const std::string move_to_san(Position& pos, Move m);
std::string pretty_pv(Position& pos, int depth, Value score, int64_t msecs, Move pv[]);
#endif // !defined(NOTATION_H_INCLUDED)
#endif // #ifndef NOTATION_H_INCLUDED
+98 -102
View File
@@ -17,6 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
#include <cassert>
#include "bitboard.h"
@@ -29,71 +30,75 @@ namespace {
#define V Value
#define S(mg, eg) make_score(mg, eg)
// Doubled pawn penalty by opposed flag and file
const Score DoubledPawnPenalty[2][FILE_NB] = {
{ S(13, 43), S(20, 48), S(23, 48), S(23, 48),
S(23, 48), S(23, 48), S(20, 48), S(13, 43) },
{ 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
const Score Doubled[FILE_NB] = {
S(13, 43), S(20, 48), S(23, 48), S(23, 48),
S(23, 48), S(23, 48), S(20, 48), S(13, 43) };
// Isolated pawn penalty by opposed flag and file
const Score IsolatedPawnPenalty[2][FILE_NB] = {
const Score Isolated[2][FILE_NB] = {
{ S(37, 45), S(54, 52), S(60, 52), S(60, 52),
S(60, 52), S(60, 52), S(54, 52), S(37, 45) },
{ S(25, 30), S(36, 35), S(40, 35), S(40, 35),
S(40, 35), S(40, 35), S(36, 35), S(25, 30) }};
S(40, 35), S(40, 35), S(36, 35), S(25, 30) } };
// Backward pawn penalty by opposed flag and file
const Score BackwardPawnPenalty[2][FILE_NB] = {
const Score Backward[2][FILE_NB] = {
{ S(30, 42), S(43, 46), S(49, 46), S(49, 46),
S(49, 46), S(49, 46), S(43, 46), S(30, 42) },
{ S(20, 28), S(29, 31), S(33, 31), S(33, 31),
S(33, 31), S(33, 31), S(29, 31), S(20, 28) }};
S(33, 31), S(33, 31), S(29, 31), S(20, 28) } };
// Pawn chain membership bonus by file
const Score ChainBonus[FILE_NB] = {
S(11,-1), S(13,-1), S(13,-1), S(14,-1),
S(14,-1), S(13,-1), S(13,-1), S(11,-1)
};
// Pawn chain membership bonus by file and rank (initialized by formula)
Score ChainMember[FILE_NB][RANK_NB];
// Candidate passed pawn bonus by rank
const Score CandidateBonus[RANK_NB] = {
const Score CandidatePassed[RANK_NB] = {
S( 0, 0), S( 6, 13), S(6,13), S(14,29),
S(34,68), S(83,166), S(0, 0), S( 0, 0)
};
S(34,68), S(83,166), S(0, 0), S( 0, 0) };
const Score PawnStructureWeight = S(233, 201);
// Weakness of our pawn shelter in front of the king indexed by [rank]
const Value ShelterWeakness[RANK_NB] =
{ V(100), V(0), V(27), V(73), V(92), V(101), V(101) };
// Weakness of our pawn shelter in front of the king indexed by [king pawn][rank]
const Value ShelterWeakness[2][RANK_NB] =
{ { V(141), V(0), V(38), V(102), V(128), V(141), V(141) },
{ V( 61), V(0), V(16), V( 44), V( 56), V( 61), V( 61) } };
// Danger of enemy pawns moving toward our king indexed by [pawn blocked][rank]
const Value StormDanger[2][RANK_NB] =
{ { V(26), V(0), V(128), V(51), V(26) },
{ V(13), V(0), V( 64), V(25), V(13) } };
// Danger of enemy pawns moving toward our king indexed by
// [no friendly pawn | pawn unblocked | pawn blocked][rank of enemy pawn]
const Value StormDanger[3][RANK_NB] = {
{ V( 0), V(64), V(128), V(51), V(26) },
{ V(26), V(32), V( 96), V(38), V(20) },
{ V( 0), V( 0), V( 64), V(25), V(13) } };
// Max bonus for king safety. Corresponds to start position with all the pawns
// in front of the king and no enemy pawn on the horizont.
// in front of the king and no enemy pawn on the horizon.
const Value MaxSafetyBonus = V(263);
#undef S
#undef V
template<Color Us>
Score evaluate_pawns(const Position& pos, Bitboard ourPawns,
Bitboard theirPawns, Pawns::Entry* e) {
Score evaluate(const Position& pos, Pawns::Entry* e) {
const Color Them = (Us == WHITE ? BLACK : WHITE);
const Color Them = (Us == WHITE ? BLACK : WHITE);
const Square Up = (Us == WHITE ? DELTA_N : DELTA_S);
const Square Right = (Us == WHITE ? DELTA_NE : DELTA_SW);
const Square Left = (Us == WHITE ? DELTA_NW : DELTA_SE);
Bitboard b;
Square s;
File f;
Rank r;
bool passed, isolated, doubled, opposed, chain, backward, candidate;
Score value = SCORE_ZERO;
const Square* pl = pos.piece_list(Us, PAWN);
const Square* pl = pos.list<PAWN>(Us);
Bitboard ourPawns = pos.pieces(Us, PAWN);
Bitboard theirPawns = pos.pieces(Them, PAWN);
e->passedPawns[Us] = e->candidatePawns[Us] = 0;
e->kingSquares[Us] = SQ_NONE;
e->semiopenFiles[Us] = 0xFF;
e->pawnAttacks[Us] = shift_bb<Right>(ourPawns) | shift_bb<Left>(ourPawns);
e->pawnsOnSquares[Us][BLACK] = popcount<Max15>(ourPawns & DarkSquares);
e->pawnsOnSquares[Us][WHITE] = pos.count<PAWN>(Us) - e->pawnsOnSquares[Us][BLACK];
// Loop through all pawns of the current color and score each pawn
while ((s = *pl++) != SQ_NONE)
@@ -101,13 +106,12 @@ namespace {
assert(pos.piece_on(s) == make_piece(Us, PAWN));
f = file_of(s);
r = rank_of(s);
// This file cannot be half open
e->halfOpenFiles[Us] &= ~(1 << f);
// This file cannot be semi-open
e->semiopenFiles[Us] &= ~(1 << f);
// Our rank plus previous one. Used for chain detection
b = rank_bb(r) | rank_bb(Us == WHITE ? r - Rank(1) : r + Rank(1));
b = rank_bb(s) | rank_bb(s - pawn_push(Us));
// Flag the pawn as passed, isolated, doubled or member of a pawn
// chain (but not the backward one).
@@ -117,41 +121,37 @@ namespace {
opposed = theirPawns & forward_bb(Us, s);
passed = !(theirPawns & passed_pawn_mask(Us, s));
// Test for backward pawn
backward = false;
// Test for backward pawn.
// If the pawn is passed, isolated, or member of a pawn chain it cannot
// be backward. If there are friendly pawns behind on adjacent files
// or if can capture an enemy pawn it cannot be backward either.
if ( !(passed | isolated | chain)
&& !(ourPawns & attack_span_mask(Them, s))
&& !(pos.attacks_from<PAWN>(s, Us) & theirPawns))
if ( (passed | isolated | chain)
|| (ourPawns & pawn_attack_span(Them, s))
|| (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 adjacent files. We now check whether the pawn is
// backward by looking in the forward direction on the adjacent
// files, and seeing whether we meet a friendly or an enemy pawn first.
b = pos.attacks_from<PAWN>(s, Us);
// files, and picking the closest pawn there.
b = pawn_attack_span(Us, s) & (ourPawns | theirPawns);
b = pawn_attack_span(Us, s) & rank_bb(backmost_sq(Us, b));
// 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;
// If we have an enemy pawn in the same or next rank, the pawn is
// backward because it cannot advance without being captured.
backward = (b | shift_bb<Up>(b)) & theirPawns;
}
assert(opposed | passed | (attack_span_mask(Us, s) & theirPawns));
assert(opposed | passed | (pawn_attack_span(Us, s) & theirPawns));
// A not passed pawn is a candidate to become passed if it is free to
// A not passed pawn is a candidate to become passed, if it is free to
// advance and if the number of friendly pawns beside or behind this
// pawn on adjacent files is higher or equal than the number of
// enemy pawns in the forward direction on the adjacent files.
candidate = !(opposed | passed | backward | isolated)
&& (b = attack_span_mask(Them, s + pawn_push(Us)) & ourPawns) != 0
&& popcount<Max15>(b) >= popcount<Max15>(attack_span_mask(Us, s) & theirPawns);
&& (b = pawn_attack_span(Them, s + pawn_push(Us)) & ourPawns) != 0
&& popcount<Max15>(b) >= popcount<Max15>(pawn_attack_span(Us, s) & theirPawns);
// Passed pawns will be properly scored in evaluation because we need
// full attack info to evaluate passed pawns. Only the frontmost passed
@@ -161,33 +161,49 @@ namespace {
// Score this pawn
if (isolated)
value -= IsolatedPawnPenalty[opposed][f];
value -= Isolated[opposed][f];
if (doubled)
value -= DoubledPawnPenalty[opposed][f];
value -= Doubled[f];
if (backward)
value -= BackwardPawnPenalty[opposed][f];
value -= Backward[opposed][f];
if (chain)
value += ChainBonus[f];
value += ChainMember[f][relative_rank(Us, s)];
if (candidate)
value += CandidateBonus[relative_rank(Us, s)];
{
value += CandidatePassed[relative_rank(Us, s)];
if (!doubled)
e->candidatePawns[Us] |= s;
}
}
e->pawnsOnSquares[Us][BLACK] = popcount<Max15>(ourPawns & BlackSquares);
e->pawnsOnSquares[Us][WHITE] = pos.piece_count(Us, PAWN) - e->pawnsOnSquares[Us][BLACK];
e->pawnsOnSquares[Them][BLACK] = popcount<Max15>(theirPawns & BlackSquares);
e->pawnsOnSquares[Them][WHITE] = pos.piece_count(Them, PAWN) - e->pawnsOnSquares[Them][BLACK];
return value;
}
}
} // namespace
namespace Pawns {
/// init() initializes some tables by formula instead of hard-code their values
void init() {
const int chainByFile[8] = { 1, 3, 3, 4, 4, 3, 3, 1 };
int bonus;
for (Rank r = RANK_1; r < RANK_8; ++r)
for (File f = FILE_A; f <= FILE_H; ++f)
{
bonus = r * (r-1) * (r-2) + chainByFile[f] * (r/2 + 1);
ChainMember[f][r] = make_score(bonus, bonus);
}
}
/// probe() takes a position object as input, computes a Entry 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.
@@ -197,27 +213,11 @@ Entry* probe(const Position& pos, Table& entries) {
Key key = pos.pawn_key();
Entry* e = entries[key];
// If e->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.
if (e->key == key)
return e;
e->key = key;
e->passedPawns[WHITE] = e->passedPawns[BLACK] = 0;
e->kingSquares[WHITE] = e->kingSquares[BLACK] = SQ_NONE;
e->halfOpenFiles[WHITE] = e->halfOpenFiles[BLACK] = 0xFF;
Bitboard wPawns = pos.pieces(WHITE, PAWN);
Bitboard bPawns = pos.pieces(BLACK, PAWN);
e->pawnAttacks[WHITE] = ((wPawns & ~FileHBB) << 9) | ((wPawns & ~FileABB) << 7);
e->pawnAttacks[BLACK] = ((bPawns & ~FileHBB) >> 7) | ((bPawns & ~FileABB) >> 9);
e->value = evaluate_pawns<WHITE>(pos, wPawns, bPawns, e)
- evaluate_pawns<BLACK>(pos, bPawns, wPawns, e);
e->value = apply_weight(e->value, PawnStructureWeight);
e->value = evaluate<WHITE>(pos, e) - evaluate<BLACK>(pos, e);
return e;
}
@@ -231,25 +231,21 @@ Value Entry::shelter_storm(const Position& pos, Square ksq) {
const Color Them = (Us == WHITE ? BLACK : WHITE);
Value safety = MaxSafetyBonus;
Bitboard b = pos.pieces(PAWN) & (in_front_bb(Us, ksq) | rank_bb(ksq));
Bitboard ourPawns = b & pos.pieces(Us) & ~rank_bb(ksq);
Bitboard b = pos.pieces(PAWN) & (in_front_bb(Us, rank_of(ksq)) | rank_bb(ksq));
Bitboard ourPawns = b & pos.pieces(Us);
Bitboard theirPawns = b & pos.pieces(Them);
Rank rkUs, rkThem;
File kf = file_of(ksq);
File kf = std::max(FILE_B, std::min(FILE_G, file_of(ksq)));
kf = (kf == FILE_A) ? FILE_B : (kf == FILE_H) ? FILE_G : kf;
for (int f = kf - 1; f <= kf + 1; f++)
for (File f = kf - File(1); f <= kf + File(1); ++f)
{
// Shelter penalty is higher for the pawn in front of the king
b = ourPawns & FileBB[f];
rkUs = b ? rank_of(Us == WHITE ? lsb(b) : ~msb(b)) : RANK_1;
safety -= ShelterWeakness[f != kf][rkUs];
b = ourPawns & file_bb(f);
rkUs = b ? relative_rank(Us, backmost_sq(Us, b)) : RANK_1;
safety -= ShelterWeakness[rkUs];
// Storm danger is smaller if enemy pawn is blocked
b = theirPawns & FileBB[f];
rkThem = b ? rank_of(Us == WHITE ? lsb(b) : ~msb(b)) : RANK_1;
safety -= StormDanger[rkThem == rkUs + 1][rkThem];
b = theirPawns & file_bb(f);
rkThem = b ? relative_rank(Us, frontmost_sq(Them, b)) : RANK_1;
safety -= StormDanger[rkUs == RANK_1 ? 0 : rkThem == rkUs + 1 ? 2 : 1][rkThem];
}
return safety;
@@ -275,7 +271,7 @@ Score Entry::update_safety(const Position& pos, Square ksq) {
Value bonus = shelter_storm<Us>(pos, ksq);
// If we can castle use the bonus after the castle if is bigger
// If we can castle use the bonus after the castle if it is bigger
if (pos.can_castle(make_castle_right(Us, KING_SIDE)))
bonus = std::max(bonus, shelter_storm<Us>(pos, relative_square(Us, SQ_G1)));
+12 -7
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(PAWNS_H_INCLUDED)
#ifndef PAWNS_H_INCLUDED
#define PAWNS_H_INCLUDED
#include "misc.h"
@@ -37,10 +37,13 @@ struct Entry {
Score pawns_value() const { return value; }
Bitboard pawn_attacks(Color c) const { return pawnAttacks[c]; }
Bitboard passed_pawns(Color c) const { return passedPawns[c]; }
int file_is_half_open(Color c, File f) const { return halfOpenFiles[c] & (1 << int(f)); }
int has_open_file_to_left(Color c, File f) const { return halfOpenFiles[c] & ((1 << int(f)) - 1); }
int has_open_file_to_right(Color c, File f) const { return halfOpenFiles[c] & ~((1 << int(f+1)) - 1); }
int pawns_on_same_color_squares(Color c, Square s) const { return pawnsOnSquares[c][!!(BlackSquares & s)]; }
Bitboard candidate_pawns(Color c) const { return candidatePawns[c]; }
int pawns_on_same_color_squares(Color c, Square s) const { return pawnsOnSquares[c][!!(DarkSquares & s)]; }
int semiopen(Color c, File f) const { return semiopenFiles[c] & (1 << int(f)); }
int semiopen_on_side(Color c, File f, bool left) const {
return semiopenFiles[c] & (left ? ((1 << int(f)) - 1) : ~((1 << int(f+1)) - 1));
}
template<Color Us>
Score king_safety(const Position& pos, Square ksq) {
@@ -57,20 +60,22 @@ struct Entry {
Key key;
Bitboard passedPawns[COLOR_NB];
Bitboard candidatePawns[COLOR_NB];
Bitboard pawnAttacks[COLOR_NB];
Square kingSquares[COLOR_NB];
int minKPdistance[COLOR_NB];
int castleRights[COLOR_NB];
Score value;
int halfOpenFiles[COLOR_NB];
int semiopenFiles[COLOR_NB];
Score kingSafety[COLOR_NB];
int pawnsOnSquares[COLOR_NB][COLOR_NB];
};
typedef HashTable<Entry, 16384> Table;
void init();
Entry* probe(const Position& pos, Table& entries);
}
#endif // !defined(PAWNS_H_INCLUDED)
#endif // #ifndef PAWNS_H_INCLUDED
+17 -14
View File
@@ -17,10 +17,10 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(PLATFORM_H_INCLUDED)
#ifndef PLATFORM_H_INCLUDED
#define PLATFORM_H_INCLUDED
#if defined(_MSC_VER)
#ifdef _MSC_VER
// Disable some silly and noisy warning from MSVC compiler
#pragma warning(disable: 4127) // Conditional expression is constant
@@ -40,16 +40,17 @@ typedef unsigned __int64 uint64_t;
#else
# include <inttypes.h>
# include <unistd.h> // Used by sysconf(_SC_NPROCESSORS_ONLN)
#endif
#if !defined(_WIN32) && !defined(_WIN64) // Linux - Unix
#ifndef _WIN32 // Linux - Unix
# include <sys/time.h>
typedef timeval sys_time_t;
inline void system_time(sys_time_t* t) { gettimeofday(t, NULL); }
inline int64_t time_to_msec(const sys_time_t& t) { return t.tv_sec * 1000LL + t.tv_usec / 1000; }
inline int64_t system_time_to_msec() {
timeval t;
gettimeofday(&t, NULL);
return t.tv_sec * 1000LL + t.tv_usec / 1000;
}
# include <pthread.h>
typedef pthread_mutex_t Lock;
@@ -66,18 +67,20 @@ typedef void*(*pt_start_fn)(void*);
# define cond_signal(x) pthread_cond_signal(&(x))
# define cond_wait(x,y) pthread_cond_wait(&(x),&(y))
# define cond_timedwait(x,y,z) pthread_cond_timedwait(&(x),&(y),z)
# define thread_create(x,f,t) !pthread_create(&(x),NULL,(pt_start_fn)f,t)
# define thread_create(x,f,t) pthread_create(&(x),NULL,(pt_start_fn)f,t)
# define thread_join(x) pthread_join(x, NULL)
#else // Windows and MinGW
# include <sys/timeb.h>
typedef _timeb sys_time_t;
inline void system_time(sys_time_t* t) { _ftime(t); }
inline int64_t time_to_msec(const sys_time_t& t) { return t.time * 1000LL + t.millitm; }
inline int64_t system_time_to_msec() {
_timeb t;
_ftime(&t);
return t.time * 1000LL + t.millitm;
}
#if !defined(NOMINMAX)
#ifndef NOMINMAX
# define NOMINMAX // disable macros min() and max()
#endif
@@ -105,9 +108,9 @@ inline DWORD* dwWin9xKludge() { static DWORD dw; return &dw; }
# define cond_signal(x) SetEvent(x)
# define cond_wait(x,y) { lock_release(y); WaitForSingleObject(x, INFINITE); lock_grab(y); }
# define cond_timedwait(x,y,z) { lock_release(y); WaitForSingleObject(x,z); lock_grab(y); }
# define thread_create(x,f,t) (x = CreateThread(NULL,0,(LPTHREAD_START_ROUTINE)f,t,0,dwWin9xKludge()), x != NULL)
# define thread_create(x,f,t) (x = CreateThread(NULL,0,(LPTHREAD_START_ROUTINE)f,t,0,dwWin9xKludge()))
# define thread_join(x) { WaitForSingleObject(x, INFINITE); CloseHandle(x); }
#endif
#endif // !defined(PLATFORM_H_INCLUDED)
#endif // #ifndef PLATFORM_H_INCLUDED
+244 -377
View File
File diff suppressed because it is too large Load Diff
+91 -72
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(POSITION_H_INCLUDED)
#ifndef POSITION_H_INCLUDED
#define POSITION_H_INCLUDED
#include <cassert>
@@ -43,16 +43,16 @@ struct CheckInfo {
};
/// The StateInfo struct stores information we need to restore a Position
/// The StateInfo struct stores information needed to restore a Position
/// object to its previous state when we retract a move. Whenever a move
/// is made on the board (by calling Position::do_move), a StateInfo object
/// must be passed as a parameter.
/// is made on the board (by calling Position::do_move), a StateInfo
/// object must be passed as a parameter.
struct StateInfo {
Key pawnKey, materialKey;
Value npMaterial[COLOR_NB];
int castleRights, rule50, pliesFromNull;
Score psqScore;
Score psq;
Square epSquare;
Key key;
@@ -67,27 +67,10 @@ struct StateInfo {
const size_t StateCopySize64 = offsetof(StateInfo, key) / sizeof(uint64_t) + 1;
/// The position data structure. A position consists of the following data:
///
/// * For each piece type, a bitboard representing the squares occupied
/// by pieces of that type.
/// * For each color, a bitboard representing the squares occupied by
/// pieces of that color.
/// * A bitboard of all occupied squares.
/// * A bitboard of all checking pieces.
/// * A 64-entry array of pieces, indexed by the squares of the board.
/// * The current side to move.
/// * Information about the castling rights for both sides.
/// * The initial files of the kings and both pairs of rooks. This is
/// used to implement the Chess960 castling rules.
/// * The en passant square (which is SQ_NONE if no en passant capture is
/// possible).
/// * The squares of the kings for both sides.
/// * Hash keys for the position itself, the current pawn structure, and
/// the current material situation.
/// * Hash keys for all previous positions in the game for detecting
/// repetition draws.
/// * A counter for detecting 50 move rule draws.
/// The Position class stores the information regarding the board representation
/// like pieces, side to move, hash keys, castling info, etc. The most important
/// methods are do_move() and undo_move(), used by the search to update node info
/// when traversing the search tree.
class Position {
public:
@@ -95,6 +78,7 @@ public:
Position(const Position& p, Thread* t) { *this = p; thisThread = t; }
Position(const std::string& f, bool c960, Thread* t) { set(f, c960, t); }
Position& operator=(const Position&);
static void init();
// Text input/output
void set(const std::string& fen, bool isChess960, Thread* th);
@@ -111,9 +95,9 @@ public:
Piece piece_on(Square s) const;
Square king_square(Color c) const;
Square ep_square() const;
bool is_empty(Square s) const;
const Square* piece_list(Color c, PieceType pt) const;
int piece_count(Color c, PieceType pt) const;
bool empty(Square s) const;
template<PieceType Pt> int count(Color c) const;
template<PieceType Pt> const Square* list(Color c) const;
// Castling
int can_castle(CastleRight f) const;
@@ -124,7 +108,7 @@ public:
// Checking
Bitboard checkers() const;
Bitboard discovered_check_candidates() const;
Bitboard pinned_pieces() const;
Bitboard pinned_pieces(Color toMove) const;
// Attacks to/from a given square
Bitboard attackers_to(Square s) const;
@@ -135,20 +119,20 @@ public:
template<PieceType> Bitboard attacks_from(Square s, Color c) const;
// Properties of moves
bool move_gives_check(Move m, const CheckInfo& ci) const;
bool pl_move_is_legal(Move m, Bitboard pinned) const;
bool is_pseudo_legal(const Move m) const;
bool is_capture(Move m) const;
bool is_capture_or_promotion(Move m) const;
bool is_passed_pawn_push(Move m) const;
Piece piece_moved(Move m) const;
bool legal(Move m, Bitboard pinned) const;
bool pseudo_legal(const Move m) const;
bool capture(Move m) const;
bool capture_or_promotion(Move m) const;
bool gives_check(Move m, const CheckInfo& ci) const;
bool passed_pawn_push(Move m) const;
Piece moved_piece(Move m) const;
PieceType captured_piece_type() const;
// Piece specific
bool pawn_is_passed(Color c, Square s) const;
bool pawn_passed(Color c, Square s) const;
bool pawn_on_7th(Color c) const;
bool opposite_bishops() const;
bool bishop_pair(Color c) const;
bool opposite_bishops() const;
// Doing and undoing moves
void do_move(Move m, StateInfo& st);
@@ -169,7 +153,6 @@ public:
// Incremental piece-square evaluation
Score psq_score() const;
Score psq_delta(Piece p, Square from, Square to) const;
Value non_pawn_material(Color c) const;
// Other properties of the position
@@ -188,12 +171,14 @@ public:
private:
// Initialization helpers (used while setting up a position)
void clear();
void put_piece(Piece p, Square s);
void set_castle_right(Color c, Square rfrom);
// Helper functions
void do_castle(Square kfrom, Square kto, Square rfrom, Square rto);
template<bool FindPinned> Bitboard hidden_checkers() const;
Bitboard hidden_checkers(Square ksq, Color c, Color toMove) const;
void put_piece(Square s, Color c, PieceType pt);
void remove_piece(Square s, Color c, PieceType pt);
void move_piece(Square from, Square to, Color c, PieceType pt);
// Computing hash keys from scratch (for initialization and debugging)
Key compute_key() const;
@@ -237,11 +222,11 @@ inline Piece Position::piece_on(Square s) const {
return board[s];
}
inline Piece Position::piece_moved(Move m) const {
inline Piece Position::moved_piece(Move m) const {
return board[from_sq(m)];
}
inline bool Position::is_empty(Square s) const {
inline bool Position::empty(Square s) const {
return board[s] == NO_PIECE;
}
@@ -273,12 +258,12 @@ inline Bitboard Position::pieces(Color c, PieceType pt1, PieceType pt2) const {
return byColorBB[c] & (byTypeBB[pt1] | byTypeBB[pt2]);
}
inline int Position::piece_count(Color c, PieceType pt) const {
return pieceCount[c][pt];
template<PieceType Pt> inline int Position::count(Color c) const {
return pieceCount[c][Pt];
}
inline const Square* Position::piece_list(Color c, PieceType pt) const {
return pieceList[c][pt];
template<PieceType Pt> inline const Square* Position::list(Color c) const {
return pieceList[c][Pt];
}
inline Square Position::ep_square() const {
@@ -331,25 +316,27 @@ inline Bitboard Position::checkers() const {
}
inline Bitboard Position::discovered_check_candidates() const {
return hidden_checkers<false>();
return hidden_checkers(king_square(~sideToMove), sideToMove, sideToMove);
}
inline Bitboard Position::pinned_pieces() const {
return hidden_checkers<true>();
inline Bitboard Position::pinned_pieces(Color toMove) const {
return hidden_checkers(king_square(toMove), ~toMove, toMove);
}
inline bool Position::pawn_is_passed(Color c, Square s) const {
inline bool Position::pawn_passed(Color c, Square s) const {
return !(pieces(~c, PAWN) & passed_pawn_mask(c, s));
}
inline bool Position::passed_pawn_push(Move m) const {
return type_of(moved_piece(m)) == PAWN
&& pawn_passed(sideToMove, to_sq(m));
}
inline Key Position::key() const {
return st->key;
}
inline Key Position::exclusion_key() const {
return st->key ^ Zobrist::exclusion;
}
inline Key Position::pawn_key() const {
return st->pawnKey;
}
@@ -358,24 +345,14 @@ inline Key Position::material_key() const {
return st->materialKey;
}
inline Score Position::psq_delta(Piece p, Square from, Square to) const {
return pieceSquareTable[p][to] - pieceSquareTable[p][from];
}
inline Score Position::psq_score() const {
return st->psqScore;
return st->psq;
}
inline Value Position::non_pawn_material(Color c) const {
return st->npMaterial[c];
}
inline bool Position::is_passed_pawn_push(Move m) const {
return type_of(piece_moved(m)) == PAWN
&& pawn_is_passed(sideToMove, to_sq(m));
}
inline int Position::game_ply() const {
return gamePly;
}
@@ -401,17 +378,17 @@ inline bool Position::is_chess960() const {
return chess960;
}
inline bool Position::is_capture_or_promotion(Move m) const {
inline bool Position::capture_or_promotion(Move m) const {
assert(is_ok(m));
return type_of(m) ? type_of(m) != CASTLE : !is_empty(to_sq(m));
return type_of(m) ? type_of(m) != CASTLE : !empty(to_sq(m));
}
inline bool Position::is_capture(Move m) const {
inline bool Position::capture(Move m) const {
// Note that castle is coded as "king captures the rook"
assert(is_ok(m));
return (!is_empty(to_sq(m)) && type_of(m) != CASTLE) || type_of(m) == ENPASSANT;
return (!empty(to_sq(m)) && type_of(m) != CASTLE) || type_of(m) == ENPASSANT;
}
inline PieceType Position::captured_piece_type() const {
@@ -422,4 +399,46 @@ inline Thread* Position::this_thread() const {
return thisThread;
}
#endif // !defined(POSITION_H_INCLUDED)
inline void Position::put_piece(Square s, Color c, PieceType pt) {
board[s] = make_piece(c, pt);
byTypeBB[ALL_PIECES] |= s;
byTypeBB[pt] |= s;
byColorBB[c] |= s;
pieceCount[c][ALL_PIECES]++;
index[s] = pieceCount[c][pt]++;
pieceList[c][pt][index[s]] = s;
}
inline void Position::move_piece(Square from, Square to, Color c, PieceType pt) {
// index[from] is not updated and becomes stale. This works as long
// as index[] is accessed just by known occupied squares.
Bitboard from_to_bb = SquareBB[from] ^ SquareBB[to];
byTypeBB[ALL_PIECES] ^= from_to_bb;
byTypeBB[pt] ^= from_to_bb;
byColorBB[c] ^= from_to_bb;
board[from] = NO_PIECE;
board[to] = make_piece(c, pt);
index[to] = index[from];
pieceList[c][pt][index[to]] = to;
}
inline void Position::remove_piece(Square s, Color c, PieceType pt) {
// WARNING: This is not a reversible operation. If we remove a piece in
// do_move() and then replace it in undo_move() we will put it at the end of
// the list and not in its original place, it means index[] and pieceList[]
// are not guaranteed to be invariant to a do_move() + undo_move() sequence.
byTypeBB[ALL_PIECES] ^= s;
byTypeBB[pt] ^= s;
byColorBB[c] ^= s;
/* board[s] = NO_PIECE; */ // Not needed, will be overwritten by capturing
pieceCount[c][ALL_PIECES]--;
Square lastSquare = pieceList[c][pt][--pieceCount[c][pt]];
index[lastSquare] = index[s];
pieceList[c][pt][index[lastSquare]] = lastSquare;
pieceList[c][pt][pieceCount[c][pt]] = SQ_NONE;
}
#endif // #ifndef POSITION_H_INCLUDED
+8 -8
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(PSQTAB_H_INCLUDED)
#ifndef PSQTAB_H_INCLUDED
#define PSQTAB_H_INCLUDED
#include "types.h"
@@ -33,12 +33,12 @@ static const Score PSQT[][SQUARE_NB] = {
{ },
{ // Pawn
S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S(0, 0), S( 0, 0), S( 0, 0), S( 0, 0),
S(-28,-8), S(-6,-8), S( 4,-8), S(14,-8), S(14,-8), S( 4,-8), S(-6,-8), S(-28,-8),
S(-28,-8), S(-6,-8), S( 9,-8), S(36,-8), S(36,-8), S( 9,-8), S(-6,-8), S(-28,-8),
S(-28,-8), S(-6,-8), S(17,-8), S(58,-8), S(58,-8), S(17,-8), S(-6,-8), S(-28,-8),
S(-28,-8), S(-6,-8), S(17,-8), S(36,-8), S(36,-8), S(17,-8), S(-6,-8), S(-28,-8),
S(-28,-8), S(-6,-8), S( 9,-8), S(14,-8), S(14,-8), S( 9,-8), S(-6,-8), S(-28,-8),
S(-28,-8), S(-6,-8), S( 4,-8), S(14,-8), S(14,-8), S( 4,-8), S(-6,-8), S(-28,-8),
S(-20,-8), S(-6,-8), S( 4,-8), S(14,-8), S(14,-8), S( 4,-8), S(-6,-8), S(-20,-8),
S(-20,-8), S(-6,-8), S( 9,-8), S(34,-8), S(34,-8), S( 9,-8), S(-6,-8), S(-20,-8),
S(-20,-8), S(-6,-8), S(17,-8), S(54,-8), S(54,-8), S(17,-8), S(-6,-8), S(-20,-8),
S(-20,-8), S(-6,-8), S(17,-8), S(34,-8), S(34,-8), S(17,-8), S(-6,-8), S(-20,-8),
S(-20,-8), S(-6,-8), S( 9,-8), S(14,-8), S(14,-8), S( 9,-8), S(-6,-8), S(-20,-8),
S(-20,-8), S(-6,-8), S( 4,-8), S(14,-8), S(14,-8), S( 4,-8), S(-6,-8), S(-20,-8),
S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S(0, 0), S( 0, 0), S( 0, 0), S( 0, 0)
},
{ // Knight
@@ -95,4 +95,4 @@ static const Score PSQT[][SQUARE_NB] = {
#undef S
#endif // !defined(PSQTAB_H_INCLUDED)
#endif // #ifndef PSQTAB_H_INCLUDED
+11 -14
View File
@@ -22,7 +22,7 @@
(at your option) any later version.
*/
#if !defined(RKISS_H_INCLUDED)
#ifndef RKISS_H_INCLUDED
#define RKISS_H_INCLUDED
#include "types.h"
@@ -43,34 +43,31 @@
class RKISS {
// Keep variables always together
struct S { uint64_t a, b, c, d; } s;
uint64_t a, b, c, d;
uint64_t rotate(uint64_t x, uint64_t k) const {
return (x << k) | (x >> (64 - k));
}
// Return 64 bit unsigned integer in between [0, 2^64 - 1]
uint64_t rand64() {
const uint64_t
e = s.a - rotate(s.b, 7);
s.a = s.b ^ rotate(s.c, 13);
s.b = s.c + rotate(s.d, 37);
s.c = s.d + e;
return s.d = e + s.a;
const uint64_t e = a - rotate(b, 7);
a = b ^ rotate(c, 13);
b = c + rotate(d, 37);
c = d + e;
return d = e + a;
}
public:
RKISS(int seed = 73) {
s.a = 0xf1ea5eed;
s.b = s.c = s.d = 0xd4e12c77;
for (int i = 0; i < seed; i++) // Scramble a few rounds
a = 0xF1EA5EED, b = c = d = 0xD4E12C77;
for (int i = 0; i < seed; ++i) // Scramble a few rounds
rand64();
}
template<typename T> T rand() { return T(rand64()); }
};
#endif // !defined(RKISS_H_INCLUDED)
#endif // #ifndef RKISS_H_INCLUDED
+292 -348
View File
File diff suppressed because it is too large Load Diff
+3 -5
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(SEARCH_H_INCLUDED)
#ifndef SEARCH_H_INCLUDED
#define SEARCH_H_INCLUDED
#include <cstring>
@@ -45,9 +45,7 @@ struct Stack {
Move killers[2];
Depth reduction;
Value staticEval;
Value evalMargin;
int skipNullMove;
int futilityMoveCount;
};
@@ -79,7 +77,7 @@ struct RootMove {
struct LimitsType {
LimitsType() { memset(this, 0, sizeof(LimitsType)); }
LimitsType() { std::memset(this, 0, sizeof(LimitsType)); }
bool use_time_management() const { return !(mate | movetime | depth | nodes | infinite); }
int time[COLOR_NB], inc[COLOR_NB], movestogo, depth, nodes, movetime, mate, infinite, ponder;
@@ -109,4 +107,4 @@ extern void think();
} // namespace Search
#endif // !defined(SEARCH_H_INCLUDED)
#endif // #ifndef SEARCH_H_INCLUDED
+84 -71
View File
@@ -19,7 +19,6 @@
#include <algorithm> // For std::count
#include <cassert>
#include <iostream>
#include "movegen.h"
#include "search.h"
@@ -30,42 +29,64 @@ using namespace Search;
ThreadPool Threads; // Global object
namespace { extern "C" {
namespace {
// start_routine() is the C function which is called when a new thread
// is launched. It is a wrapper to the virtual function idle_loop().
long start_routine(Thread* th) { th->idle_loop(); return 0; }
} }
extern "C" { long start_routine(ThreadBase* th) { th->idle_loop(); return 0; } }
// Thread c'tor starts a newly-created thread of execution that will call
// the the virtual function idle_loop(), going immediately to sleep.
// Helpers to launch a thread after creation and joining before delete. Must be
// outside Thread c'tor and d'tor because object shall be fully initialized
// when start_routine (and hence virtual idle_loop) is called and when joining.
template<typename T> T* new_thread() {
T* th = new T();
thread_create(th->handle, start_routine, th); // Will go to sleep
return th;
}
void delete_thread(ThreadBase* th) {
th->exit = true; // Search must be already finished
th->notify_one();
thread_join(th->handle); // Wait for thread termination
delete th;
}
}
// ThreadBase::notify_one() wakes up the thread when there is some work to do
void ThreadBase::notify_one() {
mutex.lock();
sleepCondition.notify_one();
mutex.unlock();
}
// ThreadBase::wait_for() set the thread to sleep until condition 'b' turns true
void ThreadBase::wait_for(volatile const bool& b) {
mutex.lock();
while (!b) sleepCondition.wait(mutex);
mutex.unlock();
}
// Thread c'tor just inits data but does not launch any thread of execution that
// instead will be started only upon c'tor returns.
Thread::Thread() /* : splitPoints() */ { // Value-initialization bug in MSVC
searching = exit = false;
searching = false;
maxPly = splitPointsSize = 0;
activeSplitPoint = NULL;
activePosition = NULL;
idx = Threads.size();
if (!thread_create(handle, start_routine, this))
{
std::cerr << "Failed to create thread number " << idx << std::endl;
::exit(EXIT_FAILURE);
}
}
// Thread d'tor waits for thread termination before to return
Thread::~Thread() {
exit = true; // Search must be already finished
notify_one();
thread_join(handle); // Wait for thread termination
}
@@ -80,11 +101,11 @@ void TimerThread::idle_loop() {
mutex.lock();
if (!exit)
sleepCondition.wait_for(mutex, msec ? msec : INT_MAX);
sleepCondition.wait_for(mutex, run ? Resolution : INT_MAX);
mutex.unlock();
if (msec)
if (run)
check_time();
}
}
@@ -123,26 +144,6 @@ void MainThread::idle_loop() {
}
// Thread::notify_one() wakes up the thread when there is some search to do
void Thread::notify_one() {
mutex.lock();
sleepCondition.notify_one();
mutex.unlock();
}
// Thread::wait_for() set the thread to sleep until condition 'b' turns true
void Thread::wait_for(volatile const bool& b) {
mutex.lock();
while (!b) sleepCondition.wait(mutex);
mutex.unlock();
}
// Thread::cutoff_occurred() checks whether a beta cutoff has occurred in the
// current active split point, or in some ancestor of the split point.
@@ -156,14 +157,14 @@ bool Thread::cutoff_occurred() const {
}
// Thread::is_available_to() checks whether the thread is available to help the
// Thread::available_to() checks whether the thread is available to help the
// thread 'master' at a split point. An obvious requirement is that thread must
// be idle. With more than two threads, this is not sufficient: If the thread is
// the master of some split point, it is only available as a slave to the slaves
// which are busy searching the split point at the top of slaves split point
// stack (the "helpful master concept" in YBWC terminology).
bool Thread::is_available_to(Thread* master) const {
bool Thread::available_to(const Thread* master) const {
if (searching)
return false;
@@ -186,8 +187,8 @@ bool Thread::is_available_to(Thread* master) const {
void ThreadPool::init() {
sleepWhileIdle = true;
timer = new TimerThread();
push_back(new MainThread());
timer = new_thread<TimerThread>();
push_back(new_thread<MainThread>());
read_uci_options();
}
@@ -196,10 +197,10 @@ void ThreadPool::init() {
void ThreadPool::exit() {
delete timer; // As first because check_time() accesses threads data
delete_thread(timer); // As first because check_time() accesses threads data
for (iterator it = begin(); it != end(); ++it)
delete *it;
delete_thread(*it);
}
@@ -216,12 +217,19 @@ void ThreadPool::read_uci_options() {
assert(requested > 0);
// Value 0 has a special meaning: We determine the optimal minimum split depth
// automatically. Anyhow the minimumSplitDepth should never be under 4 plies.
if (!minimumSplitDepth)
minimumSplitDepth = (requested < 8 ? 4 : 7) * ONE_PLY;
else
minimumSplitDepth = std::max(4 * ONE_PLY, minimumSplitDepth);
while (size() < requested)
push_back(new Thread());
push_back(new_thread<Thread>());
while (size() > requested)
{
delete back();
delete_thread(back());
pop_back();
}
}
@@ -230,10 +238,10 @@ void ThreadPool::read_uci_options() {
// slave_available() tries to find an idle thread which is available as a slave
// for the thread 'master'.
Thread* ThreadPool::available_slave(Thread* master) const {
Thread* ThreadPool::available_slave(const Thread* master) const {
for (const_iterator it = begin(); it != end(); ++it)
if ((*it)->is_available_to(master))
if ((*it)->available_to(master))
return *it;
return NULL;
@@ -250,9 +258,9 @@ Thread* ThreadPool::available_slave(Thread* master) const {
// search() then split() returns.
template <bool Fake>
void Thread::split(Position& pos, Stack* ss, Value alpha, Value beta, Value* bestValue,
void Thread::split(Position& pos, const Stack* ss, Value alpha, Value beta, Value* bestValue,
Move* bestMove, Depth depth, Move threatMove, int moveCount,
MovePicker* movePicker, int nodeType) {
MovePicker* movePicker, int nodeType, bool cutNode) {
assert(pos.pos_is_ok());
assert(*bestValue <= alpha && alpha < beta && beta <= VALUE_INFINITE);
@@ -274,6 +282,7 @@ void Thread::split(Position& pos, Stack* ss, Value alpha, Value beta, Value* bes
sp.alpha = alpha;
sp.beta = beta;
sp.nodeType = nodeType;
sp.cutNode = cutNode;
sp.movePicker = movePicker;
sp.moveCount = moveCount;
sp.pos = &pos;
@@ -287,7 +296,7 @@ void Thread::split(Position& pos, Stack* ss, Value alpha, Value beta, Value* bes
Threads.mutex.lock();
sp.mutex.lock();
splitPointsSize++;
++splitPointsSize;
activeSplitPoint = &sp;
activePosition = NULL;
@@ -321,13 +330,13 @@ void Thread::split(Position& pos, Stack* ss, Value alpha, Value beta, Value* bes
// We have returned from the idle loop, which means that all threads are
// finished. Note that setting 'searching' and decreasing splitPointsSize is
// done under lock protection to avoid a race with Thread::is_available_to().
// done under lock protection to avoid a race with Thread::available_to().
Threads.mutex.lock();
sp.mutex.lock();
}
searching = true;
splitPointsSize--;
--splitPointsSize;
activeSplitPoint = sp.parentSplitPoint;
activePosition = &pos;
pos.set_nodes_searched(pos.nodes_searched() + sp.nodes);
@@ -339,15 +348,15 @@ void Thread::split(Position& pos, Stack* ss, Value alpha, Value beta, Value* bes
}
// Explicit template instantiations
template void Thread::split<false>(Position&, Stack*, Value, Value, Value*, Move*, Depth, Move, int, MovePicker*, int);
template void Thread::split< true>(Position&, Stack*, Value, Value, Value*, Move*, Depth, Move, int, MovePicker*, int);
template void Thread::split<false>(Position&, const Stack*, Value, Value, Value*, Move*, Depth, Move, int, MovePicker*, int, bool);
template void Thread::split< true>(Position&, const Stack*, Value, Value, Value*, Move*, Depth, Move, int, MovePicker*, int, bool);
// wait_for_think_finished() waits for main thread to go to sleep then returns
void ThreadPool::wait_for_think_finished() {
MainThread* t = main_thread();
MainThread* t = main();
t->mutex.lock();
while (t->thinking) sleepCondition.wait(t->mutex);
t->mutex.unlock();
@@ -366,16 +375,20 @@ void ThreadPool::start_thinking(const Position& pos, const LimitsType& limits,
Signals.stopOnPonderhit = Signals.firstRootMove = false;
Signals.stop = Signals.failedLowAtRoot = false;
RootMoves.clear();
RootPos = pos;
Limits = limits;
SetupStates = states; // Ownership transfer here
RootMoves.clear();
if (states.get()) // If we don't set a new position, preserve current state
{
SetupStates = states; // Ownership transfer here
assert(!states.get());
}
for (MoveList<LEGAL> ml(pos); !ml.end(); ++ml)
for (MoveList<LEGAL> it(pos); *it; ++it)
if ( searchMoves.empty()
|| std::count(searchMoves.begin(), searchMoves.end(), ml.move()))
RootMoves.push_back(RootMove(ml.move()));
|| std::count(searchMoves.begin(), searchMoves.end(), *it))
RootMoves.push_back(RootMove(*it));
main_thread()->thinking = true;
main_thread()->notify_one(); // Starts main thread
main()->thinking = true;
main()->notify_one(); // Starts main thread
}
+32 -20
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(THREAD_H_INCLUDED)
#ifndef THREAD_H_INCLUDED
#define THREAD_H_INCLUDED
#include <vector>
@@ -68,6 +68,7 @@ struct SplitPoint {
Value beta;
int nodeType;
Move threatMove;
bool cutNode;
// Const pointers to shared data
MovePicker* movePicker;
@@ -85,25 +86,39 @@ struct SplitPoint {
};
/// ThreadBase struct is the base of the hierarchy from where we derive all the
/// specialized thread classes.
struct ThreadBase {
ThreadBase() : exit(false) {}
virtual ~ThreadBase() {}
virtual void idle_loop() = 0;
void notify_one();
void wait_for(volatile const bool& b);
Mutex mutex;
ConditionVariable sleepCondition;
NativeHandle handle;
volatile bool exit;
};
/// Thread struct keeps together all the thread related stuff like locks, state
/// and especially split points. We also use per-thread pawn and material hash
/// tables so that once we get a pointer to an entry its life time is unlimited
/// and we don't have to care about someone changing the entry under our feet.
struct Thread {
struct Thread : public ThreadBase {
Thread();
virtual ~Thread();
virtual void idle_loop();
void notify_one();
bool cutoff_occurred() const;
bool is_available_to(Thread* master) const;
void wait_for(volatile const bool& b);
bool available_to(const Thread* master) const;
template <bool Fake>
void split(Position& pos, Search::Stack* ss, Value alpha, Value beta, Value* bestValue, Move* bestMove,
Depth depth, Move threatMove, int moveCount, MovePicker* movePicker, int nodeType);
void split(Position& pos, const Search::Stack* ss, Value alpha, Value beta, Value* bestValue, Move* bestMove,
Depth depth, Move threatMove, int moveCount, MovePicker* movePicker, int nodeType, bool cutNode);
SplitPoint splitPoints[MAX_SPLITPOINTS_PER_THREAD];
Material::Table materialTable;
@@ -112,17 +127,13 @@ struct Thread {
Position* activePosition;
size_t idx;
int maxPly;
Mutex mutex;
ConditionVariable sleepCondition;
NativeHandle handle;
SplitPoint* volatile activeSplitPoint;
volatile int splitPointsSize;
volatile bool searching;
volatile bool exit;
};
/// MainThread and TimerThread are sublassed from Thread to characterize the two
/// MainThread and TimerThread are derived classes used to characterize the two
/// special threads: the main one and the recurring timer.
struct MainThread : public Thread {
@@ -131,10 +142,11 @@ struct MainThread : public Thread {
volatile bool thinking;
};
struct TimerThread : public Thread {
TimerThread() : msec(0) {}
struct TimerThread : public ThreadBase {
TimerThread() : run(false) {}
virtual void idle_loop();
int msec;
bool run;
static const int Resolution = 5; // msec between two check_time() calls
};
@@ -147,9 +159,9 @@ struct ThreadPool : public std::vector<Thread*> {
void init(); // No c'tor and d'tor, threads rely on globals that should
void exit(); // be initialized and valid during the whole thread lifetime.
MainThread* main_thread() { return static_cast<MainThread*>((*this)[0]); }
MainThread* main() { return static_cast<MainThread*>((*this)[0]); }
void read_uci_options();
Thread* available_slave(Thread* master) const;
Thread* available_slave(const Thread* master) const;
void wait_for_think_finished();
void start_thinking(const Position&, const Search::LimitsType&,
const std::vector<Move>&, Search::StateStackPtr&);
@@ -164,4 +176,4 @@ struct ThreadPool : public std::vector<Thread*> {
extern ThreadPool Threads;
#endif // !defined(THREAD_H_INCLUDED)
#endif // #ifndef THREAD_H_INCLUDED
+12 -13
View File
@@ -17,8 +17,8 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cmath>
#include <algorithm>
#include <cmath>
#include "search.h"
#include "timeman.h"
@@ -29,8 +29,8 @@ namespace {
/// Constants
const int MoveHorizon = 50; // Plan time management at most this many moves ahead
const float MaxRatio = 7.0f; // When in trouble, we can step over reserved time with this ratio
const float StealRatio = 0.33f; // However we must not steal time from remaining moves over this ratio
const double MaxRatio = 7.0; // When in trouble, we can step over reserved time with this ratio
const double StealRatio = 0.33; // However we must not steal time from remaining moves over this ratio
// MoveImportance[] is based on naive statistical analysis of "how many games are still undecided
@@ -76,10 +76,9 @@ namespace {
}
void TimeManager::pv_instability(int curChanges, int prevChanges) {
void TimeManager::pv_instability(double bestMoveChanges) {
unstablePVExtraTime = curChanges * (optimumSearchTime / 2)
+ prevChanges * (optimumSearchTime / 3);
unstablePVExtraTime = int(bestMoveChanges * optimumSearchTime / 1.4);
}
@@ -115,7 +114,7 @@ void TimeManager::init(const Search::LimitsType& limits, int currentPly, Color u
// We calculate optimum time usage for different hypothetic "moves to go"-values and choose the
// minimum of calculated search time values. Usually the greatest hypMTG gives the minimum values.
for (hypMTG = 1; hypMTG <= (limits.movestogo ? std::min(limits.movestogo, MoveHorizon) : MoveHorizon); hypMTG++)
for (hypMTG = 1; hypMTG <= (limits.movestogo ? std::min(limits.movestogo, MoveHorizon) : MoveHorizon); ++hypMTG)
{
// Calculate thinking time for hypothetic "moves to go"-value
hypMyTime = limits.time[us]
@@ -145,17 +144,17 @@ namespace {
template<TimeType T>
int remaining(int myTime, int movesToGo, int currentPly, int slowMover)
{
const float TMaxRatio = (T == OptimumTime ? 1 : MaxRatio);
const float TStealRatio = (T == OptimumTime ? 0 : StealRatio);
const double TMaxRatio = (T == OptimumTime ? 1 : MaxRatio);
const double TStealRatio = (T == OptimumTime ? 0 : StealRatio);
int thisMoveImportance = move_importance(currentPly) * slowMover / 100;
double thisMoveImportance = double(move_importance(currentPly) * slowMover) / 100;
int otherMovesImportance = 0;
for (int i = 1; i < movesToGo; i++)
for (int i = 1; i < movesToGo; ++i)
otherMovesImportance += move_importance(currentPly + 2 * i);
float ratio1 = (TMaxRatio * thisMoveImportance) / float(TMaxRatio * thisMoveImportance + otherMovesImportance);
float ratio2 = (thisMoveImportance + TStealRatio * otherMovesImportance) / float(thisMoveImportance + otherMovesImportance);
double ratio1 = (TMaxRatio * thisMoveImportance) / (TMaxRatio * thisMoveImportance + otherMovesImportance);
double ratio2 = (thisMoveImportance + TStealRatio * otherMovesImportance) / (thisMoveImportance + otherMovesImportance);
return int(floor(myTime * std::min(ratio1, ratio2)));
}
+3 -3
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(TIMEMAN_H_INCLUDED)
#ifndef TIMEMAN_H_INCLUDED
#define TIMEMAN_H_INCLUDED
/// The TimeManager class computes the optimal time to think depending on the
@@ -26,7 +26,7 @@
class TimeManager {
public:
void init(const Search::LimitsType& limits, int currentPly, Color us);
void pv_instability(int curChanges, int prevChanges);
void pv_instability(double bestMoveChanges);
int available_time() const { return optimumSearchTime + unstablePVExtraTime; }
int maximum_time() const { return maximumSearchTime; }
@@ -36,4 +36,4 @@ private:
int unstablePVExtraTime;
};
#endif // !defined(TIMEMAN_H_INCLUDED)
#endif // #ifndef TIMEMAN_H_INCLUDED
+31 -31
View File
@@ -40,17 +40,17 @@ void TranspositionTable::set_size(size_t mbSize) {
return;
hashMask = size - ClusterSize;
delete [] table;
table = new (std::nothrow) TTEntry[size];
free(mem);
mem = calloc(size * sizeof(TTEntry) + CACHE_LINE_SIZE - 1, 1);
if (!table)
if (!mem)
{
std::cerr << "Failed to allocate " << mbSize
<< "MB for transposition table." << std::endl;
exit(EXIT_FAILURE);
}
clear(); // Operator new is not guaranteed to initialize memory to zero
table = (TTEntry*)((uintptr_t(mem) + CACHE_LINE_SIZE - 1) & ~(CACHE_LINE_SIZE - 1));
}
@@ -60,7 +60,24 @@ void TranspositionTable::set_size(size_t mbSize) {
void TranspositionTable::clear() {
memset(table, 0, (hashMask + ClusterSize) * sizeof(TTEntry));
std::memset(table, 0, (hashMask + ClusterSize) * sizeof(TTEntry));
}
/// TranspositionTable::probe() looks up the current position in the
/// transposition table. Returns a pointer to the TTEntry or NULL if
/// position is not found.
const TTEntry* TranspositionTable::probe(const Key key) const {
const TTEntry* tte = first_entry(key);
uint32_t key32 = key >> 32;
for (unsigned i = 0; i < ClusterSize; ++i, ++tte)
if (tte->key() == key32)
return tte;
return NULL;
}
@@ -72,7 +89,7 @@ void TranspositionTable::clear() {
/// more valuable than a TTEntry t2 if t1 is from the current search and t2 is from
/// a previous search, or if the depth of t1 is bigger than the depth of t2.
void TranspositionTable::store(const Key key, Value v, Bound t, Depth d, Move m, Value statV, Value kingD) {
void TranspositionTable::store(const Key key, Value v, Bound b, Depth d, Move m, Value statV) {
int c1, c2, c3;
TTEntry *tte, *replace;
@@ -80,42 +97,25 @@ void TranspositionTable::store(const Key key, Value v, Bound t, Depth d, Move m,
tte = replace = first_entry(key);
for (unsigned i = 0; i < ClusterSize; i++, tte++)
for (unsigned i = 0; i < ClusterSize; ++i, ++tte)
{
if (!tte->key() || tte->key() == key32) // Empty or overwrite old
{
// Preserve any existing ttMove
if (m == MOVE_NONE)
m = tte->move();
if (!m)
m = tte->move(); // Preserve any existing ttMove
tte->save(key32, v, t, d, m, generation, statV, kingD);
return;
replace = tte;
break;
}
// Implement replace strategy
c1 = (replace->generation() == generation ? 2 : 0);
c2 = (tte->generation() == generation || tte->type() == BOUND_EXACT ? -2 : 0);
c2 = (tte->generation() == generation || tte->bound() == BOUND_EXACT ? -2 : 0);
c3 = (tte->depth() < replace->depth() ? 1 : 0);
if (c1 + c2 + c3 > 0)
replace = tte;
}
replace->save(key32, v, t, d, m, generation, statV, kingD);
}
/// TranspositionTable::probe() looks up the current position in the
/// transposition table. Returns a pointer to the TTEntry or NULL if
/// position is not found.
TTEntry* TranspositionTable::probe(const Key key) const {
TTEntry* tte = first_entry(key);
uint32_t key32 = key >> 32;
for (unsigned i = 0; i < ClusterSize; i++, tte++)
if (tte->key() == key32)
return tte;
return NULL;
replace->save(key32, v, b, d, m, generation, statV);
}
+24 -33
View File
@@ -17,60 +17,50 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(TT_H_INCLUDED)
#ifndef TT_H_INCLUDED
#define TT_H_INCLUDED
#include "misc.h"
#include "types.h"
/// The TTEntry is the class of transposition table entries
/// The TTEntry is the 128 bit transposition table entry, defined as below:
///
/// A TTEntry needs 128 bits to be stored
///
/// bit 0-31: key
/// bit 32-63: data
/// bit 64-79: value
/// bit 80-95: depth
/// bit 96-111: static value
/// bit 112-127: margin of static value
///
/// the 32 bits of the data field are so defined
///
/// bit 0-15: move
/// bit 16-20: not used
/// bit 21-22: value type
/// bit 23-31: generation
/// key: 32 bit
/// move: 16 bit
/// bound type: 8 bit
/// generation: 8 bit
/// value: 16 bit
/// depth: 16 bit
/// static value: 16 bit
/// static margin: 16 bit
class TTEntry {
struct TTEntry {
public:
void save(uint32_t k, Value v, Bound b, Depth d, Move m, int g, Value ev, Value em) {
void save(uint32_t k, Value v, Bound b, Depth d, Move m, int g, Value ev) {
key32 = (uint32_t)k;
move16 = (uint16_t)m;
bound = (uint8_t)b;
bound8 = (uint8_t)b;
generation8 = (uint8_t)g;
value16 = (int16_t)v;
depth16 = (int16_t)d;
evalValue = (int16_t)ev;
evalMargin = (int16_t)em;
}
void set_generation(int g) { generation8 = (uint8_t)g; }
void set_generation(uint8_t g) { generation8 = g; }
uint32_t key() const { return key32; }
Depth depth() const { return (Depth)depth16; }
Move move() const { return (Move)move16; }
Value value() const { return (Value)value16; }
Bound type() const { return (Bound)bound; }
Bound bound() const { return (Bound)bound8; }
int generation() const { return (int)generation8; }
Value eval_value() const { return (Value)evalValue; }
Value eval_margin() const { return (Value)evalMargin; }
private:
uint32_t key32;
uint16_t move16;
uint8_t bound, generation8;
int16_t value16, depth16, evalValue, evalMargin;
uint8_t bound8, generation8;
int16_t value16, depth16, evalValue;
};
@@ -85,20 +75,21 @@ class TranspositionTable {
static const unsigned ClusterSize = 4; // A cluster is 64 Bytes
public:
~TranspositionTable() { delete [] table; }
void new_search() { generation++; }
~TranspositionTable() { free(mem); }
void new_search() { ++generation; }
TTEntry* probe(const Key key) const;
const TTEntry* probe(const Key key) const;
TTEntry* first_entry(const Key key) const;
void refresh(const TTEntry* tte) const;
void set_size(size_t mbSize);
void clear();
void store(const Key key, Value v, Bound type, Depth d, Move m, Value statV, Value kingD);
void store(const Key key, Value v, Bound type, Depth d, Move m, Value statV);
private:
uint32_t hashMask;
TTEntry* table;
uint8_t generation; // Size must be not bigger then TTEntry::generation8
void* mem;
uint8_t generation; // Size must be not bigger than TTEntry::generation8
};
extern TranspositionTable TT;
@@ -122,4 +113,4 @@ inline void TranspositionTable::refresh(const TTEntry* tte) const {
const_cast<TTEntry*>(tte)->set_generation(generation);
}
#endif // !defined(TT_H_INCLUDED)
#endif // #ifndef TT_H_INCLUDED
+45 -98
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(TYPES_H_INCLUDED)
#ifndef TYPES_H_INCLUDED
#define TYPES_H_INCLUDED
/// For Linux and OSX configuration is done automatically using Makefile. To get
@@ -42,6 +42,8 @@
#include "platform.h"
#define unlikely(x) (x) // For code annotation purposes
#if defined(_WIN64) && !defined(IS_64BIT)
# include <intrin.h> // MSVC popcnt and bsfq instrinsics
# define IS_64BIT
@@ -56,13 +58,14 @@
# include <xmmintrin.h> // Intel and Microsoft header for _mm_prefetch()
# endif
#define CACHE_LINE_SIZE 64
#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
# define CACHE_LINE_ALIGNMENT __declspec(align(64))
# define CACHE_LINE_ALIGNMENT __declspec(align(CACHE_LINE_SIZE))
#else
# define CACHE_LINE_ALIGNMENT __attribute__ ((aligned(64)))
# define CACHE_LINE_ALIGNMENT __attribute__ ((aligned(CACHE_LINE_SIZE)))
#endif
#if defined(_MSC_VER)
#ifdef _MSC_VER
# define FORCE_INLINE __forceinline
#elif defined(__GNUC__)
# define FORCE_INLINE inline __attribute__((always_inline))
@@ -70,13 +73,13 @@
# define FORCE_INLINE inline
#endif
#if defined(USE_POPCNT)
#ifdef USE_POPCNT
const bool HasPopCnt = true;
#else
const bool HasPopCnt = false;
#endif
#if defined(IS_64BIT)
#ifdef IS_64BIT
const bool Is64Bit = true;
#else
const bool Is64Bit = false;
@@ -85,28 +88,9 @@ const bool Is64Bit = false;
typedef uint64_t Key;
typedef uint64_t Bitboard;
const int MAX_MOVES = 192;
const int MAX_MOVES = 256;
const int MAX_PLY = 100;
const int MAX_PLY_PLUS_2 = MAX_PLY + 2;
const Bitboard FileABB = 0x0101010101010101ULL;
const Bitboard FileBBB = FileABB << 1;
const Bitboard FileCBB = FileABB << 2;
const Bitboard FileDBB = FileABB << 3;
const Bitboard FileEBB = FileABB << 4;
const Bitboard FileFBB = FileABB << 5;
const Bitboard FileGBB = FileABB << 6;
const Bitboard FileHBB = FileABB << 7;
const Bitboard Rank1BB = 0xFF;
const Bitboard Rank2BB = Rank1BB << (8 * 1);
const Bitboard Rank3BB = Rank1BB << (8 * 2);
const Bitboard Rank4BB = Rank1BB << (8 * 3);
const Bitboard Rank5BB = Rank1BB << (8 * 4);
const Bitboard Rank6BB = Rank1BB << (8 * 5);
const Bitboard Rank7BB = Rank1BB << (8 * 6);
const Bitboard Rank8BB = Rank1BB << (8 * 7);
const int MAX_PLY_PLUS_6 = MAX_PLY + 6;
/// A move needs 16 bits to be stored
///
@@ -120,24 +104,24 @@ const Bitboard Rank8BB = Rank1BB << (8 * 7);
/// while MOVE_NONE and MOVE_NULL have the same origin and destination square.
enum Move {
MOVE_NONE = 0,
MOVE_NONE,
MOVE_NULL = 65
};
enum MoveType {
NORMAL = 0,
NORMAL,
PROMOTION = 1 << 14,
ENPASSANT = 2 << 14,
CASTLE = 3 << 14
};
enum CastleRight { // Defined as in PolyGlot book hash key
CASTLES_NONE = 0,
WHITE_OO = 1,
WHITE_OOO = 2,
BLACK_OO = 4,
BLACK_OOO = 8,
ALL_CASTLES = 15,
CASTLES_NONE,
WHITE_OO,
WHITE_OOO = WHITE_OO << 1,
BLACK_OO = WHITE_OO << 2,
BLACK_OOO = WHITE_OO << 3,
ALL_CASTLES = WHITE_OO | WHITE_OOO | BLACK_OO | BLACK_OOO,
CASTLE_RIGHT_NB = 16
};
@@ -148,7 +132,7 @@ enum CastlingSide {
};
enum Phase {
PHASE_ENDGAME = 0,
PHASE_ENDGAME,
PHASE_MIDGAME = 128,
MG = 0, EG = 1, PHASE_NB = 2
};
@@ -161,9 +145,9 @@ enum ScaleFactor {
};
enum Bound {
BOUND_NONE = 0,
BOUND_UPPER = 1,
BOUND_LOWER = 2,
BOUND_NONE,
BOUND_UPPER,
BOUND_LOWER,
BOUND_EXACT = BOUND_UPPER | BOUND_LOWER
};
@@ -189,15 +173,15 @@ enum Value {
};
enum PieceType {
NO_PIECE_TYPE = 0, ALL_PIECES = 0,
PAWN = 1, KNIGHT = 2, BISHOP = 3, ROOK = 4, QUEEN = 5, KING = 6,
NO_PIECE_TYPE, PAWN, KNIGHT, BISHOP, ROOK, QUEEN, KING,
ALL_PIECES = 0,
PIECE_TYPE_NB = 8
};
enum Piece {
NO_PIECE = 0,
W_PAWN = 1, W_KNIGHT = 2, W_BISHOP = 3, W_ROOK = 4, W_QUEEN = 5, W_KING = 6,
B_PAWN = 9, B_KNIGHT = 10, B_BISHOP = 11, B_ROOK = 12, B_QUEEN = 13, B_KING = 14,
NO_PIECE,
W_PAWN = 1, W_KNIGHT, W_BISHOP, W_ROOK, W_QUEEN, W_KING,
B_PAWN = 9, B_KNIGHT, B_BISHOP, B_ROOK, B_QUEEN, B_KING,
PIECE_NB = 16
};
@@ -210,8 +194,8 @@ enum Depth {
ONE_PLY = 2,
DEPTH_ZERO = 0 * ONE_PLY,
DEPTH_QS_CHECKS = -1 * ONE_PLY,
DEPTH_QS_NO_CHECKS = -2 * ONE_PLY,
DEPTH_QS_CHECKS = 0 * ONE_PLY,
DEPTH_QS_NO_CHECKS = -1 * ONE_PLY,
DEPTH_QS_RECAPTURES = -5 * ONE_PLY,
DEPTH_NONE = -127 * ONE_PLY
@@ -244,11 +228,11 @@ enum Square {
};
enum File {
FILE_A, FILE_B, FILE_C, FILE_D, FILE_E, FILE_F, FILE_G, FILE_H, FILE_NB = 8
FILE_A, FILE_B, FILE_C, FILE_D, FILE_E, FILE_F, FILE_G, FILE_H, FILE_NB
};
enum Rank {
RANK_1, RANK_2, RANK_3, RANK_4, RANK_5, RANK_6, RANK_7, RANK_8, RANK_NB = 8
RANK_1, RANK_2, RANK_3, RANK_4, RANK_5, RANK_6, RANK_7, RANK_8, RANK_NB
};
@@ -257,7 +241,7 @@ enum Rank {
/// 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 {
SCORE_ZERO = 0,
SCORE_ZERO,
SCORE_ENSURE_INTEGER_SIZE_P = INT_MAX,
SCORE_ENSURE_INTEGER_SIZE_N = INT_MIN
};
@@ -267,7 +251,7 @@ inline Score make_score(int mg, int eg) { return Score((mg << 16) + eg); }
/// 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(((s + 32768) & ~0xffff) / 0x10000); }
inline Value mg_value(Score s) { return Value(((s + 0x8000) & ~0xffff) / 0x10000); }
/// On Intel 64 bit we have a small speed regression with the standard conforming
/// version, so use a faster code in this case that, although not 100% standard
@@ -290,15 +274,15 @@ inline T operator-(const T d1, const T d2) { return T(int(d1) - int(d2)); } \
inline T operator*(int i, const T d) { return T(i * int(d)); } \
inline T operator*(const T d, int i) { return T(int(d) * i); } \
inline T operator-(const T d) { return T(-int(d)); } \
inline T& operator+=(T& d1, const T d2) { d1 = d1 + d2; return d1; } \
inline T& operator-=(T& d1, const T d2) { d1 = d1 - d2; return d1; } \
inline T& operator*=(T& d, int i) { d = T(int(d) * i); return d; }
inline T& operator+=(T& d1, const T d2) { return d1 = d1 + d2; } \
inline T& operator-=(T& d1, const T d2) { return d1 = d1 - d2; } \
inline T& operator*=(T& d, int i) { return d = T(int(d) * i); }
#define ENABLE_OPERATORS_ON(T) ENABLE_SAFE_OPERATORS_ON(T) \
inline T operator++(T& d, int) { d = T(int(d) + 1); return d; } \
inline T operator--(T& d, int) { d = T(int(d) - 1); return d; } \
inline T& operator++(T& d) { return d = T(int(d) + 1); } \
inline T& operator--(T& d) { return d = T(int(d) - 1); } \
inline T operator/(const T d, int i) { return T(int(d) / i); } \
inline T& operator/=(T& d, int i) { d = T(int(d) / i); return d; }
inline T& operator/=(T& d, int i) { return d = T(int(d) / i); }
ENABLE_OPERATORS_ON(Value)
ENABLE_OPERATORS_ON(PieceType)
@@ -324,47 +308,26 @@ inline Score operator/(Score s, int i) {
return make_score(mg_value(s) / i, eg_value(s) / i);
}
/// Weight score v by score w trying to prevent overflow
inline Score apply_weight(Score v, Score w) {
return make_score((int(mg_value(v)) * mg_value(w)) / 0x100,
(int(eg_value(v)) * eg_value(w)) / 0x100);
}
#undef ENABLE_OPERATORS_ON
#undef ENABLE_SAFE_OPERATORS_ON
namespace Zobrist {
extern Key psq[COLOR_NB][PIECE_TYPE_NB][SQUARE_NB];
extern Key enpassant[FILE_NB];
extern Key castle[CASTLE_RIGHT_NB];
extern Key side;
extern Key exclusion;
void init();
}
CACHE_LINE_ALIGNMENT
extern Score pieceSquareTable[PIECE_NB][SQUARE_NB];
extern Value PieceValue[PHASE_NB][PIECE_NB];
extern int SquareDistance[SQUARE_NB][SQUARE_NB];
struct MoveStack {
struct ExtMove {
Move move;
int score;
};
inline bool operator<(const MoveStack& f, const MoveStack& s) {
inline bool operator<(const ExtMove& f, const ExtMove& s) {
return f.score < s.score;
}
inline Color operator~(Color c) {
return Color(c ^ 1);
return Color(c ^ BLACK);
}
inline Square operator~(Square s) {
return Square(s ^ 56); // Vertical flip SQ_A1 -> SQ_A8
return Square(s ^ SQ_A8); // Vertical flip SQ_A1 -> SQ_A8
}
inline Square operator|(File f, Rank r) {
@@ -408,10 +371,6 @@ inline Rank rank_of(Square s) {
return Rank(s >> 3);
}
inline Square mirror(Square s) {
return Square(s ^ 7); // Horizontal flip SQ_A1 -> SQ_H1
}
inline Square relative_square(Color c, Square s) {
return Square(s ^ (c * 56));
}
@@ -429,18 +388,6 @@ inline bool opposite_colors(Square s1, Square s2) {
return ((s >> 3) ^ s) & 1;
}
inline int file_distance(Square s1, Square s2) {
return abs(file_of(s1) - file_of(s2));
}
inline int rank_distance(Square s1, Square s2) {
return abs(rank_of(s1) - rank_of(s2));
}
inline int square_distance(Square s1, Square s2) {
return SquareDistance[s1][s2];
}
inline char file_to_char(File f, bool tolower = true) {
return char(f - FILE_A + (tolower ? 'a' : 'A'));
}
@@ -489,4 +436,4 @@ inline const std::string square_to_string(Square s) {
return ch;
}
#endif // !defined(TYPES_H_INCLUDED)
#endif // #ifndef TYPES_H_INCLUDED
+16 -12
View File
@@ -42,8 +42,8 @@ namespace {
// position just before to start searching). Needed by repetition draw detection.
Search::StateStackPtr SetupStates;
void set_option(istringstream& up);
void set_position(Position& pos, istringstream& up);
void setoption(istringstream& up);
void position(Position& pos, istringstream& up);
void go(const Position& pos, istringstream& up);
}
@@ -55,7 +55,7 @@ namespace {
void UCI::loop(const string& args) {
Position pos(StartFEN, false, Threads.main_thread()); // The root position
Position pos(StartFEN, false, Threads.main()); // The root position
string token, cmd = args;
do {
@@ -76,7 +76,7 @@ void UCI::loop(const string& args) {
if (token != "ponderhit" || Search::Signals.stopOnPonderhit)
{
Search::Signals.stop = true;
Threads.main_thread()->notify_one(); // Could be sleeping
Threads.main()->notify_one(); // Could be sleeping
}
else
Search::Limits.ponder = false;
@@ -102,15 +102,19 @@ void UCI::loop(const string& args) {
<< "\n" << Options
<< "\nuciok" << sync_endl;
else if (token == "eval")
{
Search::RootColor = pos.side_to_move(); // Ensure it is set
sync_cout << Eval::trace(pos) << sync_endl;
}
else if (token == "ucinewgame") { /* Avoid returning "Unknown command" */ }
else if (token == "go") go(pos, is);
else if (token == "position") set_position(pos, is);
else if (token == "setoption") set_option(is);
else if (token == "position") position(pos, is);
else if (token == "setoption") setoption(is);
else if (token == "flip") pos.flip();
else if (token == "bench") benchmark(pos, is);
else if (token == "d") sync_cout << pos.pretty() << sync_endl;
else if (token == "isready") sync_cout << "readyok" << sync_endl;
else if (token == "eval") sync_cout << Eval::trace(pos) << sync_endl;
else
sync_cout << "Unknown command: " << cmd << sync_endl;
@@ -122,12 +126,12 @@ void UCI::loop(const string& args) {
namespace {
// set_position() is called when engine receives the "position" UCI command.
// position() is called when engine receives the "position" UCI command.
// The function sets up the position described in the given fen string ("fen")
// or the starting position ("startpos") and then makes the moves given in the
// following move list ("moves").
void set_position(Position& pos, istringstream& is) {
void position(Position& pos, istringstream& is) {
Move m;
string token, fen;
@@ -145,7 +149,7 @@ namespace {
else
return;
pos.set(fen, Options["UCI_Chess960"], Threads.main_thread());
pos.set(fen, Options["UCI_Chess960"], Threads.main());
SetupStates = Search::StateStackPtr(new std::stack<StateInfo>());
// Parse move list (if any)
@@ -157,10 +161,10 @@ namespace {
}
// set_option() is called when engine receives the "setoption" UCI command. The
// setoption() is called when engine receives the "setoption" UCI command. The
// function updates the UCI option ("name") to the given value ("value").
void set_option(istringstream& is) {
void setoption(istringstream& is) {
string token, name, value;
+14 -17
View File
@@ -51,31 +51,28 @@ bool CaseInsensitiveLess::operator() (const string& s1, const string& s2) const
/// init() initializes the UCI options to their hard coded default values
/// and initializes the default value of "Threads" and "Min Split Depth"
/// parameters according to the number of CPU cores detected.
void init(OptionsMap& o) {
int cpus = std::min(cpu_count(), MAX_THREADS);
int msd = cpus < 8 ? 4 : 7;
o["Use Debug Log"] = Option(false, on_logger);
o["Use Search Log"] = Option(false);
o["Write Debug Log"] = Option(false, on_logger);
o["Write Search Log"] = Option(false);
o["Search Log Filename"] = Option("SearchLog.txt");
o["Book File"] = Option("book.bin");
o["Best Book Move"] = Option(false);
o["Contempt Factor"] = Option(0, -50, 50);
o["Mobility (Middle Game)"] = Option(100, 0, 200, on_eval);
o["Mobility (Midgame)"] = Option(100, 0, 200, on_eval);
o["Mobility (Endgame)"] = Option(100, 0, 200, on_eval);
o["Passed Pawns (Middle Game)"] = Option(100, 0, 200, on_eval);
o["Pawn Structure (Midgame)"] = Option(100, 0, 200, on_eval);
o["Pawn Structure (Endgame)"] = Option(100, 0, 200, on_eval);
o["Passed Pawns (Midgame)"] = Option(100, 0, 200, on_eval);
o["Passed Pawns (Endgame)"] = Option(100, 0, 200, on_eval);
o["Space"] = Option(100, 0, 200, on_eval);
o["Aggressiveness"] = Option(100, 0, 200, on_eval);
o["Cowardice"] = Option(100, 0, 200, on_eval);
o["Min Split Depth"] = Option(msd, 4, 12, on_threads);
o["Max Threads per Split Point"] = Option(5, 4, 8, on_threads);
o["Threads"] = Option(cpus, 1, MAX_THREADS, on_threads);
o["Use Sleeping Threads"] = Option(false);
o["Min Split Depth"] = Option(0, 0, 12, on_threads);
o["Max Threads per Split Point"] = Option(5, 4, 8, on_threads);
o["Threads"] = Option(1, 1, MAX_THREADS, on_threads);
o["Idle Threads Sleep"] = Option(false);
o["Hash"] = Option(32, 1, 8192, on_hash_size);
o["Clear Hash"] = Option(on_clear_hash);
o["Ponder"] = Option(true);
@@ -83,10 +80,10 @@ void init(OptionsMap& o) {
o["MultiPV"] = Option(1, 1, 500);
o["Skill Level"] = Option(20, 0, 20);
o["Emergency Move Horizon"] = Option(40, 0, 50);
o["Emergency Base Time"] = Option(200, 0, 30000);
o["Emergency Move Time"] = Option(70, 0, 5000);
o["Emergency Base Time"] = Option(60, 0, 30000);
o["Emergency Move Time"] = Option(30, 0, 5000);
o["Minimum Thinking Time"] = Option(20, 0, 5000);
o["Slow Mover"] = Option(100, 10, 1000);
o["Slow Mover"] = Option(70, 10, 1000);
o["UCI_Chess960"] = Option(false);
o["UCI_AnalyseMode"] = Option(false, on_eval);
}
@@ -97,7 +94,7 @@ void init(OptionsMap& o) {
std::ostream& operator<<(std::ostream& os, const OptionsMap& om) {
for (size_t idx = 0; idx < om.size(); idx++)
for (size_t idx = 0; idx < om.size(); ++idx)
for (OptionsMap::const_iterator it = om.begin(); it != om.end(); ++it)
if (it->second.idx == idx)
{
+2 -2
View File
@@ -17,7 +17,7 @@
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#if !defined(UCIOPTION_H_INCLUDED)
#ifndef UCIOPTION_H_INCLUDED
#define UCIOPTION_H_INCLUDED
#include <map>
@@ -66,4 +66,4 @@ void loop(const std::string&);
extern UCI::OptionsMap Options;
#endif // !defined(UCIOPTION_H_INCLUDED)
#endif // #ifndef UCIOPTION_H_INCLUDED