Enable NEON for armv8

USE_NEON instead of IS_ARM
New platform apple-silicon with default USE_NEON
nnue_common.h includes arm_neon.h for USE_NEON

.travis.yml

@@ -1,5 +1,5 @@
 language: cpp
-dist: xenial
+dist: bionic
 
 matrix:
   include:
@@ -7,7 +7,6 @@ matrix:
       compiler: gcc
       addons:
         apt:
-          sources: ['ubuntu-toolchain-r-test']
           packages: ['g++-8', 'g++-8-multilib', 'g++-multilib', 'valgrind', 'expect', 'curl']
       env:
         - COMPILER=g++-8
@@ -17,23 +16,23 @@ matrix:
       compiler: clang
       addons:
         apt:
-          sources: ['ubuntu-toolchain-r-test', 'llvm-toolchain-xenial-6.0']
+          packages: ['clang-10', 'llvm-10-dev', 'g++-multilib', 'valgrind', 'expect', 'curl']
-          packages: ['clang-6.0', 'llvm-6.0-dev', 'g++-multilib', 'valgrind', 'expect', 'curl']
       env:
-        - COMPILER=clang++-6.0
+        - COMPILER=clang++-10
         - COMP=clang
-        - LDFLAGS=-fuse-ld=lld
 
     - os: osx
+      osx_image: xcode12
       compiler: gcc
       env:
         - COMPILER=g++
         - COMP=gcc
 
     - os: osx
+      osx_image: xcode12
       compiler: clang
       env:
-        - COMPILER=clang++ V='Apple LLVM 9.4.1' # Apple LLVM version 9.1.0 (clang-902.0.39.2)
+        - COMPILER=clang++
         - COMP=clang
 
 branches:
@@ -48,26 +47,34 @@ script:
   - git log HEAD | grep "\b[Bb]ench[ :]\+[0-9]\{7\}" | head -n 1 | sed "s/[^0-9]*\([0-9]*\).*/\1/g" > git_sig
   - export benchref=$(cat git_sig)
   - echo "Reference bench:" $benchref
+
+  #
+  # Compiler version string
+  - $COMPILER -v
+
   #
   # Verify bench number against various builds
   - export CXXFLAGS="-Werror -D_GLIBCXX_DEBUG"
   - make clean && make -j2 ARCH=x86-64 optimize=no debug=yes build && ../tests/signature.sh $benchref
-  - make clean && make -j2 ARCH=x86-32 optimize=no debug=yes build && ../tests/signature.sh $benchref
+  - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32 optimize=no debug=yes build && ../tests/signature.sh $benchref; fi
-  - make clean && make -j2 ARCH=x86-32 build && ../tests/signature.sh $benchref
+  - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32 build && ../tests/signature.sh $benchref; fi
 
   #
   # Check perft and reproducible search
+  - export CXXFLAGS="-Werror"
+  - make clean && make -j2 ARCH=x86-64 build
   - ../tests/perft.sh
   - ../tests/reprosearch.sh
+
   #
   # Valgrind
   #
   - export CXXFLAGS="-O1 -fno-inline"
   - if [ -x "$(command -v valgrind )" ]; then make clean && make -j2 ARCH=x86-64 debug=yes optimize=no build > /dev/null && ../tests/instrumented.sh --valgrind; fi
   - if [ -x "$(command -v valgrind )" ]; then ../tests/instrumented.sh --valgrind-thread; fi
+
   #
   # Sanitizer
   #
-  # Use g++-8 as a proxy for having sanitizers, might need revision as they become available for more recent versions of clang/gcc
+  - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-64 sanitize=undefined optimize=no debug=yes build > /dev/null && ../tests/instrumented.sh --sanitizer-undefined; fi
-  - if [[ "$COMPILER" == "g++-8" ]]; then make clean && make -j2 ARCH=x86-64 sanitize=undefined optimize=no debug=yes build > /dev/null && ../tests/instrumented.sh --sanitizer-undefined; fi
+  - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-64 sanitize=thread    optimize=no debug=yes build > /dev/null && ../tests/instrumented.sh --sanitizer-thread; fi
-  - if [[ "$COMPILER" == "g++-8" ]]; then make clean && make -j2 ARCH=x86-64 sanitize=thread    optimize=no debug=yes build > /dev/null && ../tests/instrumented.sh --sanitizer-thread; fi

AUTHORS

@@ -1,10 +1,17 @@
-# List of authors for Stockfish, as of March 30, 2020
+# List of authors for Stockfish, as of August 4, 2020
 
+# Founders of the Stockfish project and fishtest infrastructure
 Tord Romstad (romstad)
 Marco Costalba (mcostalba)
 Joona Kiiski (zamar)
 Gary Linscott (glinscott)
 
+# Authors and inventors of NNUE, training, NNUE port
+Yu Nasu (ynasu87)
+Motohiro Isozaki (yaneurao)
+Hisayori Noda (nodchip)
+
+# all other authors of the code in alphabetical order
 Aditya (absimaldata)
 Adrian Petrescu (apetresc)
 Ajith Chandy Jose (ajithcj)
@@ -36,6 +43,7 @@ Dariusz Orzechowski
 David Zar
 Daylen Yang (daylen)
 DiscanX
+Dominik Schlösser (domschl)
 double-beep
 Eduardo Cáceres (eduherminio)
 Eelco de Groot (KingDefender)
@@ -115,7 +123,8 @@ Nick Pelling (nickpelling)
 Nicklas Persson (NicklasPersson)
 Niklas Fiekas (niklasf)
 Nikolay Kostov (NikolayIT)
-Nguyen Pham
+Nguyen Pham (nguyenpham)
+Norman Schmidt (FireFather)
 Ondrej Mosnáček (WOnder93)
 Oskar Werkelin Ahlin
 Pablo Vazquez
@@ -135,6 +144,7 @@ Richard Lloyd
 Rodrigo Exterckötter Tjäder
 Ron Britvich (Britvich)
 Ronald de Man (syzygy1, syzygy)
+rqs
 Ryan Schmitt
 Ryan Takker
 Sami Kiminki (skiminki)
@@ -143,6 +153,7 @@ Sergei Antonov (saproj)
 Sergei Ivanov (svivanov72)
 sf-x
 Shane Booth (shane31)
+Shawn Varghese (xXH4CKST3RXx)
 Stefan Geschwentner (locutus2)
 Stefano Cardanobile (Stefano80)
 Steinar Gunderson (sesse)
@@ -155,9 +166,11 @@ Tom Vijlbrief (tomtor)
 Tomasz Sobczyk (Sopel97)
 Torsten Franz (torfranz, tfranzer)
 Tracey Emery (basepr1me)
+tttak
 Unai Corzo (unaiic)
 Uri Blass (uriblass)
 Vince Negri (cuddlestmonkey)
+zz4032
 
 
 # Additionally, we acknowledge the authors and maintainers of fishtest,

README.md

@@ -4,7 +4,13 @@
 [![Build Status](https://ci.appveyor.com/api/projects/status/github/official-stockfish/Stockfish?branch=master&svg=true)](https://ci.appveyor.com/project/mcostalba/stockfish/branch/master)
 
 [Stockfish](https://stockfishchess.org) is a free, powerful UCI chess engine
-derived from Glaurung 2.1. It is not a complete chess program and requires a
+derived from Glaurung 2.1. It features two evaluation functions, the classical
+evaluation based on handcrafted terms, and the NNUE evaluation based on
+efficiently updateable neural networks. The classical evaluation runs efficiently
+on most 64bit CPU architectures, while the NNUE evaluation benefits strongly from the
+vector intrinsics available on modern CPUs (avx2 or similar).
+
+Stockfish is not a complete chess program and requires a
 UCI-compatible GUI (e.g. XBoard with PolyGlot, Scid, Cute Chess, eboard, Arena,
 Sigma Chess, Shredder, Chess Partner or Fritz) in order to be used comfortably.
 Read the documentation for your GUI of choice for information about how to use
@@ -22,21 +28,20 @@ This distribution of Stockfish consists of the following files:
   * src, a subdirectory containing the full source code, including a Makefile
     that can be used to compile Stockfish on Unix-like systems.
 
+To use the NNUE evaluation an additional data file with neural network parameters
+needs to be downloaded. The filename for the default set can be found as the default
+value of the `EvalFile` UCI option, with the format
+`nn-[SHA256 first 12 digits].nnue` (e.g. nn-c157e0a5755b.nnue). This file can be downloaded from
+```
+https://tests.stockfishchess.org/api/nn/[filename]
+```
+replacing `[filename]` as needed.
 
-## UCI parameters
+
+## UCI options
 
 Currently, Stockfish has the following UCI options:
 
-  * #### Debug Log File
-    Write all communication to and from the engine into a text file.
-
-  * #### Contempt
-    A positive value for contempt favors middle game positions and avoids draws.
-
-  * #### Analysis Contempt
-    By default, contempt is set to prefer the side to move. Set this option to "White"
-    or "Black" to analyse with contempt for that side, or "Off" to disable contempt.
-
   * #### Threads
     The number of CPU threads used for searching a position. For best performance, set
     this equal to the number of CPU cores available.
@@ -44,9 +49,6 @@ Currently, Stockfish has the following UCI options:
   * #### Hash
     The size of the hash table in MB. It is recommended to set Hash after setting Threads.
 
-  * #### Clear Hash
-    Clear the hash table.
-
   * #### Ponder
     Let Stockfish ponder its next move while the opponent is thinking.
 
@@ -54,10 +56,32 @@ Currently, Stockfish has the following UCI options:
     Output the N best lines (principal variations, PVs) when searching.
     Leave at 1 for best performance.
 
-  * #### Skill Level
+  * #### Use NNUE
-    Lower the Skill Level in order to make Stockfish play weaker (see also UCI_LimitStrength).
+    Toggle between the NNUE and classical evaluation functions. If set to "true",
-    Internally, MultiPV is enabled, and with a certain probability depending on the Skill Level a
+    the network parameters must be availabe to load from file (see also EvalFile).
-    weaker move will be played.
+
+  * #### EvalFile
+    The name of the file of the NNUE evaluation parameters. Depending on the GUI the
+    filename should include the full path to the folder/directory that contains the file.
+
+  * #### Contempt
+    A positive value for contempt favors middle game positions and avoids draws,
+    effective for the classical evaluation only.
+
+  * #### Analysis Contempt
+    By default, contempt is set to prefer the side to move. Set this option to "White"
+    or "Black" to analyse with contempt for that side, or "Off" to disable contempt.
+
+  * #### UCI_AnalyseMode
+    An option handled by your GUI.
+
+  * #### UCI_Chess960
+    An option handled by your GUI. If true, Stockfish will play Chess960.
+
+  * #### UCI_ShowWDL
+    If enabled, show approximate WDL statistics as part of the engine output.
+    These WDL numbers model expected game outcomes for a given evaluation and
+    game ply for engine self-play at fishtest LTC conditions (60+0.6s per game).
 
   * #### UCI_LimitStrength
     Enable weaker play aiming for an Elo rating as set by UCI_Elo. This option overrides Skill Level.
@@ -66,28 +90,10 @@ Currently, Stockfish has the following UCI options:
     If enabled by UCI_LimitStrength, aim for an engine strength of the given Elo.
     This Elo rating has been calibrated at a time control of 60s+0.6s and anchored to CCRL 40/4.
 
-  * #### UCI_ShowWDL
+  * #### Skill Level
-    If enabled, show approximate WDL statistics as part of the engine output.
+    Lower the Skill Level in order to make Stockfish play weaker (see also UCI_LimitStrength).
-    These WDL numbers model expected game outcomes for a given evaluation and
+    Internally, MultiPV is enabled, and with a certain probability depending on the Skill Level a
-    game ply for engine self-play at fishtest LTC conditions (60+0.6s per game).
+    weaker move will be played.
-
-  * #### Move Overhead
-    Assume a time delay of x ms due to network and GUI overheads. This is useful to
-    avoid losses on time in those cases.
-
-  * #### Slow Mover
-    Lower values will make Stockfish take less time in games, higher values will
-    make it think longer.
-
-  * #### nodestime
-    Tells the engine to use nodes searched instead of wall time to account for
-    elapsed time. Useful for engine testing.
-
-  * #### UCI_Chess960
-    An option handled by your GUI. If true, Stockfish will play Chess960.
-
-  * #### UCI_AnalyseMode
-    An option handled by your GUI.
 
   * #### SyzygyPath
     Path to the folders/directories storing the Syzygy tablebase files. Multiple
@@ -114,6 +120,47 @@ Currently, Stockfish has the following UCI options:
     Limit Syzygy tablebase probing to positions with at most this many pieces left
     (including kings and pawns).
 
+  * #### Move Overhead
+    Assume a time delay of x ms due to network and GUI overheads. This is useful to
+    avoid losses on time in those cases.
+
+  * #### Slow Mover
+    Lower values will make Stockfish take less time in games, higher values will
+    make it think longer.
+
+  * #### nodestime
+    Tells the engine to use nodes searched instead of wall time to account for
+    elapsed time. Useful for engine testing.
+
+  * #### Clear Hash
+    Clear the hash table.
+
+  * #### Debug Log File
+    Write all communication to and from the engine into a text file.
+
+## classical and NNUE evaluation
+
+Both approaches assign a value to a position that is used in alpha-beta (PVS) search
+to find the best move. The classical evaluation computes this value as a function
+of various chess concepts, handcrafted by experts, tested and tuned using fishtest.
+The NNUE evaluation computes this value with a neural network based on basic
+inputs (e.g. piece positions only). The network is optimized and trained
+on the evalutions of millions of positions at moderate search depth.
+
+The NNUE evaluation was first introduced in shogi, and ported to Stockfish afterward.
+It can be evaluated efficiently on CPUs, and exploits the fact that only parts
+of the neural network need to be updated after a typical chess move.
+[The nodchip repository](https://github.com/nodchip/Stockfish) provides additional
+tools to train and develop the NNUE networks.
+
+On CPUs supporting modern vector instructions (avx2 and similar), the NNUE evaluation
+results in stronger playing strength, even if the nodes per second computed by the engine
+is somewhat lower (roughly 60% of nps is typical).
+
+Note that the NNUE evaluation depends on the Stockfish binary and the network parameter
+file (see EvalFile). Not every parameter file is compatible with a given Stockfish binary.
+The default value of the EvalFile UCI option is the name of a network that is guaranteed
+to be compatible with that binary.
 
 ## What to expect from Syzygybases?
 

appveyor.yml

@@ -4,10 +4,10 @@ clone_depth: 50
 branches:
   only:
     - master
-    - appveyor
+    - nnue-player-wip
 
 # Operating system (build VM template)
-os: Visual Studio 2017
+os: Visual Studio 2019
 
 # Build platform, i.e. x86, x64, AnyCPU. This setting is optional.
 platform:
@@ -36,8 +36,11 @@ before_build:
       $src = $src.Replace("\", "/")
 
       # Build CMakeLists.txt
-      $t = 'cmake_minimum_required(VERSION 3.8)',
+      $t = 'cmake_minimum_required(VERSION 3.17)',
            'project(Stockfish)',
+           'set(CMAKE_CXX_STANDARD 17)',
+           'set(CMAKE_CXX_STANDARD_REQUIRED ON)',
+           'set (CMAKE_CXX_EXTENSIONS OFF)',
            'set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_SOURCE_DIR}/src)',
            'set(source_files', $src, ')',
            'add_executable(stockfish ${source_files})'
@@ -51,10 +54,11 @@ before_build:
       $b = git log HEAD | sls "\b[Bb]ench[ :]+[0-9]{7}" | select -first 1
       $bench = $b -match '\D+(\d+)' | % { $matches[1] }
       Write-Host "Reference bench:" $bench
-      $g = "Visual Studio 15 2017"
+      $g = "Visual Studio 16 2019"
-      If (${env:PLATFORM} -eq 'x64') { $g = $g + ' Win64' }
+      If (${env:PLATFORM} -eq 'x64') { $a = "x64" }
-      cmake -G "${g}" .
+      If (${env:PLATFORM} -eq 'x86') { $a = "Win32" }
-      Write-Host "Generated files for: " $g
+      cmake -G "${g}" -A ${a} .
+      Write-Host "Generated files for: " $g $a
 
 build_script:
   - cmake --build . --config %CONFIGURATION% -- /verbosity:minimal

src/Makefile

@@ -38,11 +38,12 @@ PGOBENCH = ./$(EXE) bench
 ### Source and object files
 SRCS = benchmark.cpp bitbase.cpp bitboard.cpp endgame.cpp evaluate.cpp main.cpp \
 	material.cpp misc.cpp movegen.cpp movepick.cpp pawns.cpp position.cpp psqt.cpp \
-	search.cpp thread.cpp timeman.cpp tt.cpp uci.cpp ucioption.cpp tune.cpp syzygy/tbprobe.cpp
+	search.cpp thread.cpp timeman.cpp tt.cpp uci.cpp ucioption.cpp tune.cpp syzygy/tbprobe.cpp \
+	nnue/evaluate_nnue.cpp nnue/features/half_kp.cpp
 
 OBJS = $(notdir $(SRCS:.cpp=.o))
 
-VPATH = syzygy
+VPATH = syzygy:nnue:nnue/features
 
 ### Establish the operating system name
 KERNEL = $(shell uname -s)
@@ -67,7 +68,14 @@ endif
 # prefetch = yes/no   --- -DUSE_PREFETCH   --- Use prefetch asm-instruction
 # popcnt = yes/no     --- -DUSE_POPCNT     --- Use popcnt asm-instruction
 # sse = yes/no        --- -msse            --- Use Intel Streaming SIMD Extensions
+# sse3 = yes/no       --- -msse3           --- Use Intel Streaming SIMD Extensions 3
+# ssse3 = yes/no      --- -mssse3          --- Use Intel Supplemental Streaming SIMD Extensions 3
+# sse41 = yes/no      --- -msse4.1         --- Use Intel Streaming SIMD Extensions 4.1
+# sse42 = yes/no      --- -msse4.2         --- Use Intel Streaming SIMD Extensions 4.2
+# avx2 = yes/no       --- -mavx2           --- Use Intel Advanced Vector Extensions 2
 # pext = yes/no       --- -DUSE_PEXT       --- Use pext x86_64 asm-instruction
+# avx512 = yes/no     --- -mavx512bw       --- Use Intel Advanced Vector Extensions 512
+# neon = yes/no       --- -DUSE_NEON       --- Use ARM SIMD architecture
 #
 # Note that Makefile is space sensitive, so when adding new architectures
 # or modifying existing flags, you have to make sure there are no extra spaces
@@ -81,7 +89,14 @@ bits = 64
 prefetch = no
 popcnt = no
 sse = no
+sse3 = no
+ssse3 = no
+sse41 = no
+sse42 = no
+avx2 = no
 pext = no
+avx512 = no
+neon = no
 
 ### 2.2 Architecture specific
 ifeq ($(ARCH),general-32)
@@ -111,11 +126,70 @@ ifeq ($(ARCH),x86-64)
 	sse = yes
 endif
 
+ifeq ($(ARCH),x86-64-sse3)
+	arch = x86_64
+	prefetch = yes
+	sse = yes
+	sse3 = yes
+endif
+
+ifeq ($(ARCH),x86-64-sse3-popcnt)
+	arch = x86_64
+	prefetch = yes
+	sse = yes
+	sse3 = yes
+	popcnt = yes
+endif
+
+ifeq ($(ARCH),x86-64-ssse3)
+	arch = x86_64
+	prefetch = yes
+	sse = yes
+	sse3 = yes
+	ssse3 = yes
+endif
+
+ifeq ($(ARCH),x86-64-sse41)
+	arch = x86_64
+	prefetch = yes
+	popcnt = yes
+	sse = yes
+	sse3 = yes
+	ssse3 = yes
+	sse41 = yes
+endif
+
 ifeq ($(ARCH),x86-64-modern)
 	arch = x86_64
 	prefetch = yes
 	popcnt = yes
 	sse = yes
+	sse3 = yes
+	ssse3 = yes
+	sse41 = yes
+endif
+
+ifeq ($(ARCH),x86-64-sse42)
+	arch = x86_64
+	prefetch = yes
+	popcnt = yes
+	sse = yes
+	sse3 = yes
+	ssse3 = yes
+	sse41 = yes
+	sse42 = yes
+endif
+
+ifeq ($(ARCH),x86-64-avx2)
+	arch = x86_64
+	prefetch = yes
+	popcnt = yes
+	sse = yes
+	sse3 = yes
+	ssse3 = yes
+	sse41 = yes
+	sse42 = yes
+	avx2 = yes
 endif
 
 ifeq ($(ARCH),x86-64-bmi2)
@@ -123,9 +197,28 @@ ifeq ($(ARCH),x86-64-bmi2)
 	prefetch = yes
 	popcnt = yes
 	sse = yes
+	sse3 = yes
+	ssse3 = yes
+	sse41 = yes
+	sse42 = yes
+	avx2 = yes
 	pext = yes
 endif
 
+ifeq ($(ARCH),x86-64-avx512)
+	arch = x86_64
+	prefetch = yes
+	popcnt = yes
+	sse = yes
+	sse3 = yes
+	ssse3 = yes
+	sse41 = yes
+	sse42 = yes
+	avx2 = yes
+	pext = yes
+	avx512 = yes
+endif
+
 ifeq ($(ARCH),armv7)
 	arch = armv7
 	prefetch = yes
@@ -136,6 +229,14 @@ ifeq ($(ARCH),armv8)
 	arch = armv8-a
 	prefetch = yes
 	popcnt = yes
+	neon = yes
+endif
+
+ifeq ($(ARCH),apple-silicon)
+	arch = arm64
+	prefetch = yes
+	popcnt = yes
+	neon = yes
 endif
 
 ifeq ($(ARCH),ppc-32)
@@ -154,8 +255,8 @@ endif
 ### ==========================================================================
 
 ### 3.1 Selecting compiler (default = gcc)
-CXXFLAGS += -Wall -Wcast-qual -fno-exceptions -std=c++11 $(EXTRACXXFLAGS)
+CXXFLAGS += -Wall -Wcast-qual -fno-exceptions -std=c++17 $(EXTRACXXFLAGS)
-DEPENDFLAGS += -std=c++11
+DEPENDFLAGS += -std=c++17
 LDFLAGS += $(EXTRALDFLAGS)
 
 ifeq ($(COMP),)
@@ -249,8 +350,8 @@ endif
 endif
 
 ifeq ($(KERNEL),Darwin)
-	CXXFLAGS += -arch $(arch) -mmacosx-version-min=10.9
+	CXXFLAGS += -arch $(arch) -mmacosx-version-min=10.15
-	LDFLAGS += -arch $(arch) -mmacosx-version-min=10.9
+	LDFLAGS += -arch $(arch) -mmacosx-version-min=10.15
 endif
 
 ### Travis CI script uses COMPILER to overwrite CXX
@@ -283,8 +384,8 @@ endif
 
 ### 3.2.2 Debugging with undefined behavior sanitizers
 ifneq ($(sanitize),no)
-        CXXFLAGS += -g3 -fsanitize=$(sanitize) -fuse-ld=gold
+        CXXFLAGS += -g3 -fsanitize=$(sanitize)
-        LDFLAGS += -fsanitize=$(sanitize) -fuse-ld=gold
+        LDFLAGS += -fsanitize=$(sanitize)
 endif
 
 ### 3.3 Optimization
@@ -322,7 +423,7 @@ endif
 
 ### 3.6 popcnt
 ifeq ($(popcnt),yes)
-	ifeq ($(arch),$(filter $(arch),ppc64 armv8-a))
+	ifeq ($(arch),$(filter $(arch),ppc64 armv8-a arm64))
 		CXXFLAGS += -DUSE_POPCNT
 	else ifeq ($(comp),icc)
 		CXXFLAGS += -msse3 -DUSE_POPCNT
@@ -331,11 +432,61 @@ ifeq ($(popcnt),yes)
 	endif
 endif
 
+ifeq ($(avx2),yes)
+	CXXFLAGS += -DUSE_AVX2
+	ifeq ($(comp),$(filter $(comp),gcc clang mingw))
+		CXXFLAGS += -mavx2
+	endif
+endif
+
+ifeq ($(avx512),yes)
+	CXXFLAGS += -DUSE_AVX512
+	ifeq ($(comp),$(filter $(comp),gcc clang mingw))
+		CXXFLAGS += -mavx512bw
+	endif
+endif
+
+ifeq ($(sse42),yes)
+	CXXFLAGS += -DUSE_SSE42
+	ifeq ($(comp),$(filter $(comp),gcc clang mingw))
+		CXXFLAGS += -msse4.2
+	endif
+endif
+
+ifeq ($(sse41),yes)
+	CXXFLAGS += -DUSE_SSE41
+	ifeq ($(comp),$(filter $(comp),gcc clang mingw))
+		CXXFLAGS += -msse4.1
+	endif
+endif
+
+ifeq ($(ssse3),yes)
+	CXXFLAGS += -DUSE_SSSE3
+	ifeq ($(comp),$(filter $(comp),gcc clang mingw))
+		CXXFLAGS += -mssse3
+	endif
+endif
+
+ifeq ($(sse3),yes)
+	CXXFLAGS += -DUSE_SSE3
+	ifeq ($(comp),$(filter $(comp),gcc clang mingw))
+		CXXFLAGS += -msse3
+	endif
+endif
+
+ifeq ($(neon),yes)
+	CXXFLAGS += -DUSE_NEON
+endif
+
+ifeq ($(arch),x86_64)
+	CXXFLAGS += -DUSE_SSE2
+endif
+
 ### 3.7 pext
 ifeq ($(pext),yes)
 	CXXFLAGS += -DUSE_PEXT
 	ifeq ($(comp),$(filter $(comp),gcc clang mingw))
-		CXXFLAGS += -msse4 -mbmi2
+		CXXFLAGS += -mbmi2
 	endif
 endif
 
@@ -381,15 +532,23 @@ help:
 	@echo "Supported targets:"
 	@echo ""
 	@echo "build                   > Standard build"
-	@echo "profile-build           > PGO build"
+	@echo "profile-build           > Standard build with PGO"
 	@echo "strip                   > Strip executable"
 	@echo "install                 > Install executable"
 	@echo "clean                   > Clean up"
+	@echo "net                     > Download the default nnue net"
 	@echo ""
 	@echo "Supported archs:"
 	@echo ""
-	@echo "x86-64-bmi2             > x86 64-bit with pext support (also enables SSE4)"
+	@echo "x86-64-avx512           > x86 64-bit with avx512 support"
-	@echo "x86-64-modern           > x86 64-bit with popcnt support (also enables SSE3)"
+	@echo "x86-64-bmi2             > x86 64-bit with bmi2 support"
+	@echo "x86-64-avx2             > x86 64-bit with avx2 support"
+	@echo "x86-64-sse42            > x86 64-bit with sse42 support"
+	@echo "x86-64-modern           > x86 64-bit with sse41 support (x86-64-sse41)"
+	@echo "x86-64-sse41            > x86 64-bit with sse41 support"
+	@echo "x86-64-ssse3            > x86 64-bit with ssse3 support"
+	@echo "x86-64-sse3-popcnt      > x86 64-bit with sse3 and popcnt support"
+	@echo "x86-64-sse3             > x86 64-bit with sse3 support"
 	@echo "x86-64                  > x86 64-bit generic"
 	@echo "x86-32                  > x86 32-bit (also enables SSE)"
 	@echo "x86-32-old              > x86 32-bit fall back for old hardware"
@@ -397,6 +556,7 @@ help:
 	@echo "ppc-32                  > PPC 32-bit"
 	@echo "armv7                   > ARMv7 32-bit"
 	@echo "armv8                   > ARMv8 64-bit"
+	@echo "apple-silicon           > Apple silicon ARM64"
 	@echo "general-64              > unspecified 64-bit"
 	@echo "general-32              > unspecified 32-bit"
 	@echo ""
@@ -419,7 +579,7 @@ help:
 	@echo ""
 
 
-.PHONY: help build profile-build strip install clean objclean profileclean \
+.PHONY: help build profile-build strip install clean net objclean profileclean \
         config-sanity icc-profile-use icc-profile-make gcc-profile-use gcc-profile-make \
         clang-profile-use clang-profile-make
 
@@ -453,14 +613,21 @@ install:
 clean: objclean profileclean
 	@rm -f .depend *~ core
 
+net:
+	$(eval nnuenet := $(shell grep EvalFile ucioption.cpp | grep Option | sed 's/.*\(nn-[a-z0-9]\{12\}.nnue\).*/\1/'))
+	@echo "Default net: $(nnuenet)"
+	$(eval nnuedownloadurl := https://tests.stockfishchess.org/api/nn/$(nnuenet))
+	$(eval curl_or_wget := $(shell if hash curl 2>/dev/null; then echo "curl -sL"; elif hash wget 2>/dev/null; then echo "wget -qO-"; fi))
+	@if test -f "$(nnuenet)"; then echo "Already available."; else echo "Downloading $(nnuedownloadurl)"; $(curl_or_wget) $(nnuedownloadurl) > $(nnuenet); fi
+
 # clean binaries and objects
 objclean:
-	@rm -f $(EXE) *.o ./syzygy/*.o
+	@rm -f $(EXE) *.o ./syzygy/*.o ./nnue/*.o ./nnue/features/*.o
 
 # clean auxiliary profiling files
 profileclean:
 	@rm -rf profdir
-	@rm -f bench.txt *.gcda *.gcno
+	@rm -f bench.txt *.gcda *.gcno ./syzygy/*.gcda ./nnue/*.gcda ./nnue/features/*.gcda
 	@rm -f stockfish.profdata *.profraw
 
 default:
@@ -485,7 +652,14 @@ config-sanity:
 	@echo "prefetch: '$(prefetch)'"
 	@echo "popcnt: '$(popcnt)'"
 	@echo "sse: '$(sse)'"
+	@echo "sse3: '$(sse3)'"
+	@echo "ssse3: '$(ssse3)'"
+	@echo "sse41: '$(sse41)'"
+	@echo "sse42: '$(sse42)'"
+	@echo "avx2: '$(avx2)'"
 	@echo "pext: '$(pext)'"
+	@echo "avx512: '$(avx512)'"
+	@echo "neon: '$(neon)'"
 	@echo ""
 	@echo "Flags:"
 	@echo "CXX: $(CXX)"
@@ -499,12 +673,19 @@ config-sanity:
 	@test "$(optimize)" = "yes" || test "$(optimize)" = "no"
 	@test "$(arch)" = "any" || test "$(arch)" = "x86_64" || test "$(arch)" = "i386" || \
 	 test "$(arch)" = "ppc64" || test "$(arch)" = "ppc" || \
-	 test "$(arch)" = "armv7" || test "$(arch)" = "armv8-a"
+	 test "$(arch)" = "armv7" || test "$(arch)" = "armv8-a" || test "$(arch)" = "arm64"
 	@test "$(bits)" = "32" || test "$(bits)" = "64"
 	@test "$(prefetch)" = "yes" || test "$(prefetch)" = "no"
 	@test "$(popcnt)" = "yes" || test "$(popcnt)" = "no"
 	@test "$(sse)" = "yes" || test "$(sse)" = "no"
+	@test "$(sse3)" = "yes" || test "$(sse3)" = "no"
+	@test "$(ssse3)" = "yes" || test "$(ssse3)" = "no"
+	@test "$(sse41)" = "yes" || test "$(sse41)" = "no"
+	@test "$(sse42)" = "yes" || test "$(sse42)" = "no"
+	@test "$(avx2)" = "yes" || test "$(avx2)" = "no"
 	@test "$(pext)" = "yes" || test "$(pext)" = "no"
+	@test "$(avx512)" = "yes" || test "$(avx512)" = "no"
+	@test "$(neon)" = "yes" || test "$(neon)" = "no"
 	@test "$(comp)" = "gcc" || test "$(comp)" = "icc" || test "$(comp)" = "mingw" || test "$(comp)" = "clang"
 
 $(EXE): $(OBJS)

src/benchmark.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/bitbase.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/bitboard.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/bitboard.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/endgame.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/endgame.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/evaluate.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -20,15 +18,50 @@
 
 #include <algorithm>
 #include <cassert>
+#include <cstdlib>
 #include <cstring>   // For std::memset
 #include <iomanip>
 #include <sstream>
+#include <iostream>
 
 #include "bitboard.h"
 #include "evaluate.h"
 #include "material.h"
 #include "pawns.h"
 #include "thread.h"
+#include "uci.h"
+
+namespace Eval {
+
+  bool useNNUE;
+  std::string eval_file_loaded="None";
+
+  void init_NNUE() {
+
+    useNNUE = Options["Use NNUE"];
+    std::string eval_file = std::string(Options["EvalFile"]);
+    if (useNNUE && eval_file_loaded != eval_file)
+        if (Eval::NNUE::load_eval_file(eval_file))
+            eval_file_loaded = eval_file;
+  }
+
+  void verify_NNUE() {
+
+    std::string eval_file = std::string(Options["EvalFile"]);
+    if (useNNUE && eval_file_loaded != eval_file)
+    {
+        std::cerr << "Use of NNUE evaluation, but the file " << eval_file << " was not loaded successfully. "
+                  << "These network evaluation parameters must be available, compatible with this version of the code. "
+                  << "The UCI option EvalFile might need to specify the full path, including the directory/folder name, to the file." << std::endl;
+        std::exit(EXIT_FAILURE);
+    }
+
+    if (useNNUE)
+        sync_cout << "info string NNUE evaluation using " << eval_file << " enabled." << sync_endl;
+    else
+        sync_cout << "info string classical evaluation enabled." << sync_endl;
+  }
+}
 
 namespace Trace {
 
@@ -906,28 +939,40 @@ make_v:
 /// evaluation of the position from the point of view of the side to move.
 
 Value Eval::evaluate(const Position& pos) {
+
+  if (Eval::useNNUE)
+      return NNUE::evaluate(pos);
+  else
       return Evaluation<NO_TRACE>(pos).value();
 }
 
-
 /// trace() is like evaluate(), but instead of returning a value, it returns
 /// a string (suitable for outputting to stdout) that contains the detailed
 /// descriptions and values of each evaluation term. Useful for debugging.
+/// Trace scores are from white's point of view
 
 std::string Eval::trace(const Position& pos) {
 
   if (pos.checkers())
-      return "Total evaluation: none (in check)";
+      return "Final evaluation: none (in check)";
 
+  std::stringstream ss;
+  ss << std::showpoint << std::noshowpos << std::fixed << std::setprecision(2);
+
+  Value v;
+
+  if (Eval::useNNUE)
+  {
+      v = NNUE::evaluate(pos);
+  }
+  else
+  {
       std::memset(scores, 0, sizeof(scores));
 
       pos.this_thread()->contempt = SCORE_ZERO; // Reset any dynamic contempt
 
-  Value v = Evaluation<TRACE>(pos).value();
+      v = Evaluation<TRACE>(pos).value();
 
-  v = pos.side_to_move() == WHITE ? v : -v; // Trace scores are from white's point of view
-
-  std::stringstream ss;
       ss << std::showpoint << std::noshowpos << std::fixed << std::setprecision(2)
          << "     Term    |    White    |    Black    |    Total   \n"
          << "             |   MG    EG  |   MG    EG  |   MG    EG \n"
@@ -947,6 +992,9 @@ std::string Eval::trace(const Position& pos) {
          << "    Winnable | " << Term(WINNABLE)
          << " ------------+-------------+-------------+------------\n"
          << "       Total | " << Term(TOTAL);
+  }
+
+  v = pos.side_to_move() == WHITE ? v : -v;
 
   ss << "\nFinal evaluation: " << to_cp(v) << " (white side)\n";
 

src/evaluate.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -30,8 +28,22 @@ class Position;
 namespace Eval {
 
   std::string trace(const Position& pos);
+  Value evaluate(const Position& pos);
+
+  extern bool useNNUE;
+  extern std::string eval_file_loaded;
+  void init_NNUE();
+  void verify_NNUE();
+
+  namespace NNUE {
 
     Value evaluate(const Position& pos);
-}
+    Value compute_eval(const Position& pos);
+    void  update_eval(const Position& pos);
+    bool  load_eval_file(const std::string& evalFile);
+
+  } // namespace NNUE
+
+} // namespace Eval
 
 #endif // #ifndef EVALUATE_H_INCLUDED

src/main.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -46,6 +44,7 @@ int main(int argc, char* argv[]) {
   Endgames::init();
   Threads.set(size_t(Options["Threads"]));
   Search::clear(); // After threads are up
+  Eval::init_NNUE();
 
   UCI::loop(argc, argv);
 

src/material.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/material.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/misc.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -46,6 +44,7 @@ typedef bool(*fun3_t)(HANDLE, CONST GROUP_AFFINITY*, PGROUP_AFFINITY);
 #include <iostream>
 #include <sstream>
 #include <vector>
+#include <cstdlib>
 
 #if defined(__linux__) && !defined(__ANDROID__)
 #include <stdlib.h>
@@ -147,10 +146,8 @@ const string engine_info(bool to_uci) {
       ss << setw(2) << day << setw(2) << (1 + months.find(month) / 4) << year.substr(2);
   }
 
-  ss << (Is64Bit ? " 64" : "")
+  ss << (to_uci  ? "\nid author ": " by ")
-     << (HasPext ? " BMI2" : (HasPopCnt ? " POPCNT" : ""))
+     << "the Stockfish developers (see AUTHORS file)";
-     << (to_uci  ? "\nid author ": " by ")
-     << "T. Romstad, M. Costalba, J. Kiiski, G. Linscott";
 
   return ss.str();
 }
@@ -215,6 +212,32 @@ const std::string compiler_info() {
      compiler += " on unknown system";
   #endif
 
+  compiler += "\nCompilation settings include: ";
+  compiler += (Is64Bit ? " 64bit" : " 32bit");
+  #if defined(USE_AVX512)
+    compiler += " AVX512";
+  #endif
+  #if defined(USE_AVX2)
+    compiler += " AVX2";
+  #endif
+  #if defined(USE_SSE42)
+    compiler += " SSE42";
+  #endif
+  #if defined(USE_SSE41)
+    compiler += " SSE41";
+  #endif
+  #if defined(USE_SSSE3)
+    compiler += " SSSE3";
+  #endif
+  #if defined(USE_SSE3)
+    compiler += " SSE3";
+  #endif
+    compiler += (HasPext ? " BMI2" : "");
+    compiler += (HasPopCnt ? " POPCNT" : "");
+  #if !defined(NDEBUG)
+    compiler += " DEBUG";
+  #endif
+
   compiler += "\n__VERSION__ macro expands to: ";
   #ifdef __VERSION__
      compiler += __VERSION__;
@@ -293,6 +316,29 @@ void prefetch(void* addr) {
 
 #endif
 
+/// Wrappers for systems where the c++17 implementation doesn't guarantee the availability of aligned_alloc.
+/// Memory allocated with std_aligned_alloc must be freed with std_aligned_free.
+///
+
+void* std_aligned_alloc(size_t alignment, size_t size) {
+#if defined(__APPLE__)
+  return aligned_alloc(alignment, size);
+#elif defined(_WIN32)
+  return _mm_malloc(size, alignment);
+#else
+  return std::aligned_alloc(alignment, size);
+#endif
+}
+
+void std_aligned_free(void* ptr) {
+#if defined(__APPLE__)
+  free(ptr);
+#elif defined(_WIN32)
+  _mm_free(ptr);
+#else
+  free(ptr);
+#endif
+}
 
 /// aligned_ttmem_alloc() will return suitably aligned memory, and if possible use large pages.
 /// The returned pointer is the aligned one, while the mem argument is the one that needs

src/misc.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -33,6 +31,8 @@ const std::string engine_info(bool to_uci = false);
 const std::string compiler_info();
 void prefetch(void* addr);
 void start_logger(const std::string& fname);
+void* std_aligned_alloc(size_t alignment, size_t size);
+void std_aligned_free(void* ptr);
 void* aligned_ttmem_alloc(size_t size, void*& mem);
 void aligned_ttmem_free(void* mem); // nop if mem == nullptr
 

src/movegen.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/movegen.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/movepick.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/movepick.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/nnue/architectures/halfkp_256x2-32-32.h

@@ -0,0 +1,54 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Definition of input features and network structure used in NNUE evaluation function
+
+#ifndef NNUE_HALFKP_256X2_32_32_H_INCLUDED
+#define NNUE_HALFKP_256X2_32_32_H_INCLUDED
+
+#include "../features/feature_set.h"
+#include "../features/half_kp.h"
+
+#include "../layers/input_slice.h"
+#include "../layers/affine_transform.h"
+#include "../layers/clipped_relu.h"
+
+namespace Eval::NNUE {
+
+// Input features used in evaluation function
+using RawFeatures = Features::FeatureSet<
+    Features::HalfKP<Features::Side::kFriend>>;
+
+// Number of input feature dimensions after conversion
+constexpr IndexType kTransformedFeatureDimensions = 256;
+
+namespace Layers {
+
+// Define network structure
+using InputLayer = InputSlice<kTransformedFeatureDimensions * 2>;
+using HiddenLayer1 = ClippedReLU<AffineTransform<InputLayer, 32>>;
+using HiddenLayer2 = ClippedReLU<AffineTransform<HiddenLayer1, 32>>;
+using OutputLayer = AffineTransform<HiddenLayer2, 1>;
+
+}  // namespace Layers
+
+using Network = Layers::OutputLayer;
+
+}  // namespace Eval::NNUE
+
+#endif // #ifndef NNUE_HALFKP_256X2_32_32_H_INCLUDED

src/nnue/evaluate_nnue.cpp

@@ -0,0 +1,178 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Code for calculating NNUE evaluation function
+
+#include <fstream>
+#include <iostream>
+#include <set>
+
+#include "../evaluate.h"
+#include "../position.h"
+#include "../misc.h"
+#include "../uci.h"
+
+#include "evaluate_nnue.h"
+
+ExtPieceSquare kpp_board_index[PIECE_NB] = {
+ // convention: W - us, B - them
+ // viewed from other side, W and B are reversed
+    { PS_NONE,     PS_NONE     },
+    { PS_W_PAWN,   PS_B_PAWN   },
+    { PS_W_KNIGHT, PS_B_KNIGHT },
+    { PS_W_BISHOP, PS_B_BISHOP },
+    { PS_W_ROOK,   PS_B_ROOK   },
+    { PS_W_QUEEN,  PS_B_QUEEN  },
+    { PS_W_KING,   PS_B_KING   },
+    { PS_NONE,     PS_NONE     },
+    { PS_NONE,     PS_NONE     },
+    { PS_B_PAWN,   PS_W_PAWN   },
+    { PS_B_KNIGHT, PS_W_KNIGHT },
+    { PS_B_BISHOP, PS_W_BISHOP },
+    { PS_B_ROOK,   PS_W_ROOK   },
+    { PS_B_QUEEN,  PS_W_QUEEN  },
+    { PS_B_KING,   PS_W_KING   },
+    { PS_NONE,     PS_NONE     }
+};
+
+
+namespace Eval::NNUE {
+
+  // Input feature converter
+  AlignedPtr<FeatureTransformer> feature_transformer;
+
+  // Evaluation function
+  AlignedPtr<Network> network;
+
+  // Evaluation function file name
+  std::string fileName;
+
+  namespace Detail {
+
+  // Initialize the evaluation function parameters
+  template <typename T>
+  void Initialize(AlignedPtr<T>& pointer) {
+
+    pointer.reset(reinterpret_cast<T*>(std_aligned_alloc(alignof(T), sizeof(T))));
+    std::memset(pointer.get(), 0, sizeof(T));
+  }
+
+  // Read evaluation function parameters
+  template <typename T>
+  bool ReadParameters(std::istream& stream, const AlignedPtr<T>& pointer) {
+
+    std::uint32_t header;
+    stream.read(reinterpret_cast<char*>(&header), sizeof(header));
+    if (!stream || header != T::GetHashValue()) return false;
+    return pointer->ReadParameters(stream);
+  }
+
+  }  // namespace Detail
+
+  // Initialize the evaluation function parameters
+  void Initialize() {
+
+    Detail::Initialize(feature_transformer);
+    Detail::Initialize(network);
+  }
+
+  // Read network header
+  bool ReadHeader(std::istream& stream,
+    std::uint32_t* hash_value, std::string* architecture) {
+
+    std::uint32_t version, size;
+    stream.read(reinterpret_cast<char*>(&version), sizeof(version));
+    stream.read(reinterpret_cast<char*>(hash_value), sizeof(*hash_value));
+    stream.read(reinterpret_cast<char*>(&size), sizeof(size));
+    if (!stream || version != kVersion) return false;
+    architecture->resize(size);
+    stream.read(&(*architecture)[0], size);
+    return !stream.fail();
+  }
+
+  // Read network parameters
+  bool ReadParameters(std::istream& stream) {
+
+    std::uint32_t hash_value;
+    std::string architecture;
+    if (!ReadHeader(stream, &hash_value, &architecture)) return false;
+    if (hash_value != kHashValue) return false;
+    if (!Detail::ReadParameters(stream, feature_transformer)) return false;
+    if (!Detail::ReadParameters(stream, network)) return false;
+    return stream && stream.peek() == std::ios::traits_type::eof();
+  }
+
+  // Proceed with the difference calculation if possible
+  static void UpdateAccumulatorIfPossible(const Position& pos) {
+
+    feature_transformer->UpdateAccumulatorIfPossible(pos);
+  }
+
+  // Calculate the evaluation value
+  static Value ComputeScore(const Position& pos, bool refresh) {
+
+    auto& accumulator = pos.state()->accumulator;
+    if (!refresh && accumulator.computed_score) {
+      return accumulator.score;
+    }
+
+    alignas(kCacheLineSize) TransformedFeatureType
+        transformed_features[FeatureTransformer::kBufferSize];
+    feature_transformer->Transform(pos, transformed_features, refresh);
+    alignas(kCacheLineSize) char buffer[Network::kBufferSize];
+    const auto output = network->Propagate(transformed_features, buffer);
+
+    auto score = static_cast<Value>(output[0] / FV_SCALE);
+
+    accumulator.score = score;
+    accumulator.computed_score = true;
+    return accumulator.score;
+  }
+
+  // Load the evaluation function file
+  bool load_eval_file(const std::string& evalFile) {
+
+    Initialize();
+    fileName = evalFile;
+
+    std::ifstream stream(evalFile, std::ios::binary);
+
+    const bool result = ReadParameters(stream);
+
+    return result;
+  }
+
+  // Evaluation function. Perform differential calculation.
+  Value evaluate(const Position& pos) {
+    Value v = ComputeScore(pos, false);
+    v = Utility::clamp(v, VALUE_TB_LOSS_IN_MAX_PLY + 1, VALUE_TB_WIN_IN_MAX_PLY - 1);
+
+    return v;
+  }
+
+  // Evaluation function. Perform full calculation.
+  Value compute_eval(const Position& pos) {
+    return ComputeScore(pos, true);
+  }
+
+  // Proceed with the difference calculation if possible
+  void update_eval(const Position& pos) {
+    UpdateAccumulatorIfPossible(pos);
+  }
+
+} // namespace Eval::NNUE

src/nnue/evaluate_nnue.h

@@ -0,0 +1,48 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// header used in NNUE evaluation function
+
+#ifndef NNUE_EVALUATE_NNUE_H_INCLUDED
+#define NNUE_EVALUATE_NNUE_H_INCLUDED
+
+#include "nnue_feature_transformer.h"
+
+#include <memory>
+
+namespace Eval::NNUE {
+
+  // Hash value of evaluation function structure
+  constexpr std::uint32_t kHashValue =
+      FeatureTransformer::GetHashValue() ^ Network::GetHashValue();
+
+  // Deleter for automating release of memory area
+  template <typename T>
+  struct AlignedDeleter {
+    void operator()(T* ptr) const {
+      ptr->~T();
+      std_aligned_free(ptr);
+    }
+  };
+
+  template <typename T>
+  using AlignedPtr = std::unique_ptr<T, AlignedDeleter<T>>;
+
+}  // namespace Eval::NNUE
+
+#endif // #ifndef NNUE_EVALUATE_NNUE_H_INCLUDED

src/nnue/features/feature_set.h

@@ -0,0 +1,135 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// A class template that represents the input feature set of the NNUE evaluation function
+
+#ifndef NNUE_FEATURE_SET_H_INCLUDED
+#define NNUE_FEATURE_SET_H_INCLUDED
+
+#include "features_common.h"
+#include <array>
+
+namespace Eval::NNUE::Features {
+
+  // Class template that represents a list of values
+  template <typename T, T... Values>
+  struct CompileTimeList;
+
+  template <typename T, T First, T... Remaining>
+  struct CompileTimeList<T, First, Remaining...> {
+    static constexpr bool Contains(T value) {
+      return value == First || CompileTimeList<T, Remaining...>::Contains(value);
+    }
+    static constexpr std::array<T, sizeof...(Remaining) + 1>
+        kValues = {{First, Remaining...}};
+  };
+
+  // Base class of feature set
+  template <typename Derived>
+  class FeatureSetBase {
+
+   public:
+    // Get a list of indices for active features
+    template <typename IndexListType>
+    static void AppendActiveIndices(
+        const Position& pos, TriggerEvent trigger, IndexListType active[2]) {
+
+      for (Color perspective : { WHITE, BLACK }) {
+        Derived::CollectActiveIndices(
+            pos, trigger, perspective, &active[perspective]);
+      }
+    }
+
+    // Get a list of indices for recently changed features
+    template <typename PositionType, typename IndexListType>
+    static void AppendChangedIndices(
+        const PositionType& pos, TriggerEvent trigger,
+        IndexListType removed[2], IndexListType added[2], bool reset[2]) {
+
+      const auto& dp = pos.state()->dirtyPiece;
+      if (dp.dirty_num == 0) return;
+
+      for (Color perspective : { WHITE, BLACK }) {
+        reset[perspective] = false;
+        switch (trigger) {
+          case TriggerEvent::kFriendKingMoved:
+            reset[perspective] =
+                dp.pieceId[0] == PIECE_ID_KING + perspective;
+            break;
+          default:
+            assert(false);
+            break;
+        }
+        if (reset[perspective]) {
+          Derived::CollectActiveIndices(
+              pos, trigger, perspective, &added[perspective]);
+        } else {
+          Derived::CollectChangedIndices(
+              pos, trigger, perspective,
+              &removed[perspective], &added[perspective]);
+        }
+      }
+    }
+  };
+
+  // Class template that represents the feature set
+  template <typename FeatureType>
+  class FeatureSet<FeatureType> : public FeatureSetBase<FeatureSet<FeatureType>> {
+
+   public:
+    // Hash value embedded in the evaluation file
+    static constexpr std::uint32_t kHashValue = FeatureType::kHashValue;
+    // Number of feature dimensions
+    static constexpr IndexType kDimensions = FeatureType::kDimensions;
+    // Maximum number of simultaneously active features
+    static constexpr IndexType kMaxActiveDimensions =
+        FeatureType::kMaxActiveDimensions;
+    // Trigger for full calculation instead of difference calculation
+    using SortedTriggerSet =
+        CompileTimeList<TriggerEvent, FeatureType::kRefreshTrigger>;
+    static constexpr auto kRefreshTriggers = SortedTriggerSet::kValues;
+
+   private:
+    // Get a list of indices for active features
+    static void CollectActiveIndices(
+        const Position& pos, const TriggerEvent trigger, const Color perspective,
+        IndexList* const active) {
+      if (FeatureType::kRefreshTrigger == trigger) {
+        FeatureType::AppendActiveIndices(pos, perspective, active);
+      }
+    }
+
+    // Get a list of indices for recently changed features
+    static void CollectChangedIndices(
+        const Position& pos, const TriggerEvent trigger, const Color perspective,
+        IndexList* const removed, IndexList* const added) {
+
+      if (FeatureType::kRefreshTrigger == trigger) {
+        FeatureType::AppendChangedIndices(pos, perspective, removed, added);
+      }
+    }
+
+    // Make the base class and the class template that recursively uses itself a friend
+    friend class FeatureSetBase<FeatureSet>;
+    template <typename... FeatureTypes>
+    friend class FeatureSet;
+  };
+
+}  // namespace Eval::NNUE::Features
+
+#endif // #ifndef NNUE_FEATURE_SET_H_INCLUDED

src/nnue/features/features_common.h

@@ -0,0 +1,45 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+//Common header of input features of NNUE evaluation function
+
+#ifndef NNUE_FEATURES_COMMON_H_INCLUDED
+#define NNUE_FEATURES_COMMON_H_INCLUDED
+
+#include "../../evaluate.h"
+#include "../nnue_common.h"
+
+namespace Eval::NNUE::Features {
+
+  class IndexList;
+
+  template <typename... FeatureTypes>
+  class FeatureSet;
+
+  // Trigger to perform full calculations instead of difference only
+  enum class TriggerEvent {
+    kFriendKingMoved // calculate full evaluation when own king moves
+  };
+
+  enum class Side {
+    kFriend // side to move
+  };
+
+}  // namespace Eval::NNUE::Features
+
+#endif // #ifndef NNUE_FEATURES_COMMON_H_INCLUDED

src/nnue/features/half_kp.cpp

@@ -0,0 +1,92 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+//Definition of input features HalfKP of NNUE evaluation function
+
+#include "half_kp.h"
+#include "index_list.h"
+
+namespace Eval::NNUE::Features {
+
+  // Find the index of the feature quantity from the king position and PieceSquare
+  template <Side AssociatedKing>
+  inline IndexType HalfKP<AssociatedKing>::MakeIndex(Square sq_k, PieceSquare p) {
+    return static_cast<IndexType>(PS_END) * static_cast<IndexType>(sq_k) + p;
+  }
+
+  // Get pieces information
+  template <Side AssociatedKing>
+  inline void HalfKP<AssociatedKing>::GetPieces(
+      const Position& pos, Color perspective,
+      PieceSquare** pieces, Square* sq_target_k) {
+
+    *pieces = (perspective == BLACK) ?
+        pos.eval_list()->piece_list_fb() :
+        pos.eval_list()->piece_list_fw();
+    const PieceId target = (AssociatedKing == Side::kFriend) ?
+        static_cast<PieceId>(PIECE_ID_KING + perspective) :
+        static_cast<PieceId>(PIECE_ID_KING + ~perspective);
+    *sq_target_k = static_cast<Square>(((*pieces)[target] - PS_W_KING) % SQUARE_NB);
+  }
+
+  // Get a list of indices for active features
+  template <Side AssociatedKing>
+  void HalfKP<AssociatedKing>::AppendActiveIndices(
+      const Position& pos, Color perspective, IndexList* active) {
+
+    // Do nothing if array size is small to avoid compiler warning
+    if (RawFeatures::kMaxActiveDimensions < kMaxActiveDimensions) return;
+
+    PieceSquare* pieces;
+    Square sq_target_k;
+    GetPieces(pos, perspective, &pieces, &sq_target_k);
+    for (PieceId i = PIECE_ID_ZERO; i < PIECE_ID_KING; ++i) {
+      if (pieces[i] != PS_NONE) {
+        active->push_back(MakeIndex(sq_target_k, pieces[i]));
+      }
+    }
+  }
+
+  // Get a list of indices for recently changed features
+  template <Side AssociatedKing>
+  void HalfKP<AssociatedKing>::AppendChangedIndices(
+      const Position& pos, Color perspective,
+      IndexList* removed, IndexList* added) {
+
+    PieceSquare* pieces;
+    Square sq_target_k;
+    GetPieces(pos, perspective, &pieces, &sq_target_k);
+    const auto& dp = pos.state()->dirtyPiece;
+    for (int i = 0; i < dp.dirty_num; ++i) {
+      if (dp.pieceId[i] >= PIECE_ID_KING) continue;
+      const auto old_p = static_cast<PieceSquare>(
+          dp.old_piece[i].from[perspective]);
+      if (old_p != PS_NONE) {
+        removed->push_back(MakeIndex(sq_target_k, old_p));
+      }
+      const auto new_p = static_cast<PieceSquare>(
+          dp.new_piece[i].from[perspective]);
+      if (new_p != PS_NONE) {
+        added->push_back(MakeIndex(sq_target_k, new_p));
+      }
+    }
+  }
+
+  template class HalfKP<Side::kFriend>;
+
+}  // namespace Eval::NNUE::Features

src/nnue/features/half_kp.h

@@ -0,0 +1,67 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+//Definition of input features HalfKP of NNUE evaluation function
+
+#ifndef NNUE_FEATURES_HALF_KP_H_INCLUDED
+#define NNUE_FEATURES_HALF_KP_H_INCLUDED
+
+#include "../../evaluate.h"
+#include "features_common.h"
+
+namespace Eval::NNUE::Features {
+
+  // Feature HalfKP: Combination of the position of own king
+  // and the position of pieces other than kings
+  template <Side AssociatedKing>
+  class HalfKP {
+
+   public:
+    // Feature name
+    static constexpr const char* kName = "HalfKP(Friend)";
+    // Hash value embedded in the evaluation file
+    static constexpr std::uint32_t kHashValue =
+        0x5D69D5B9u ^ (AssociatedKing == Side::kFriend);
+    // Number of feature dimensions
+    static constexpr IndexType kDimensions =
+        static_cast<IndexType>(SQUARE_NB) * static_cast<IndexType>(PS_END);
+    // Maximum number of simultaneously active features
+    static constexpr IndexType kMaxActiveDimensions = PIECE_ID_KING;
+    // Trigger for full calculation instead of difference calculation
+    static constexpr TriggerEvent kRefreshTrigger = TriggerEvent::kFriendKingMoved;
+
+    // Get a list of indices for active features
+    static void AppendActiveIndices(const Position& pos, Color perspective,
+                                    IndexList* active);
+
+    // Get a list of indices for recently changed features
+    static void AppendChangedIndices(const Position& pos, Color perspective,
+                                     IndexList* removed, IndexList* added);
+
+    // Index of a feature for a given king position and another piece on some square
+    static IndexType MakeIndex(Square sq_k, PieceSquare p);
+
+   private:
+    // Get pieces information
+    static void GetPieces(const Position& pos, Color perspective,
+                          PieceSquare** pieces, Square* sq_target_k);
+  };
+
+}  // namespace Eval::NNUE::Features
+
+#endif // #ifndef NNUE_FEATURES_HALF_KP_H_INCLUDED

src/nnue/features/index_list.h

@@ -0,0 +1,64 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Definition of index list of input features
+
+#ifndef NNUE_FEATURES_INDEX_LIST_H_INCLUDED
+#define NNUE_FEATURES_INDEX_LIST_H_INCLUDED
+
+#include "../../position.h"
+#include "../nnue_architecture.h"
+
+namespace Eval::NNUE::Features {
+
+  // Class template used for feature index list
+  template <typename T, std::size_t MaxSize>
+  class ValueList {
+
+   public:
+    std::size_t size() const { return size_; }
+    void resize(std::size_t size) { size_ = size; }
+    void push_back(const T& value) { values_[size_++] = value; }
+    T& operator[](std::size_t index) { return values_[index]; }
+    T* begin() { return values_; }
+    T* end() { return values_ + size_; }
+    const T& operator[](std::size_t index) const { return values_[index]; }
+    const T* begin() const { return values_; }
+    const T* end() const { return values_ + size_; }
+
+    void swap(ValueList& other) {
+      const std::size_t max_size = std::max(size_, other.size_);
+      for (std::size_t i = 0; i < max_size; ++i) {
+        std::swap(values_[i], other.values_[i]);
+      }
+      std::swap(size_, other.size_);
+    }
+
+   private:
+    T values_[MaxSize];
+    std::size_t size_ = 0;
+  };
+
+  //Type of feature index list
+  class IndexList
+      : public ValueList<IndexType, RawFeatures::kMaxActiveDimensions> {
+  };
+
+}  // namespace Eval::NNUE::Features
+
+#endif // NNUE_FEATURES_INDEX_LIST_H_INCLUDED

src/nnue/layers/affine_transform.h

@@ -0,0 +1,215 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Definition of layer AffineTransform of NNUE evaluation function
+
+#ifndef NNUE_LAYERS_AFFINE_TRANSFORM_H_INCLUDED
+#define NNUE_LAYERS_AFFINE_TRANSFORM_H_INCLUDED
+
+#include <iostream>
+#include "../nnue_common.h"
+
+namespace Eval::NNUE::Layers {
+
+  // Affine transformation layer
+  template <typename PreviousLayer, IndexType OutputDimensions>
+  class AffineTransform {
+   public:
+    // Input/output type
+    using InputType = typename PreviousLayer::OutputType;
+    using OutputType = std::int32_t;
+    static_assert(std::is_same<InputType, std::uint8_t>::value, "");
+
+    // Number of input/output dimensions
+    static constexpr IndexType kInputDimensions =
+        PreviousLayer::kOutputDimensions;
+    static constexpr IndexType kOutputDimensions = OutputDimensions;
+    static constexpr IndexType kPaddedInputDimensions =
+        CeilToMultiple<IndexType>(kInputDimensions, kMaxSimdWidth);
+
+    // Size of forward propagation buffer used in this layer
+    static constexpr std::size_t kSelfBufferSize =
+        CeilToMultiple(kOutputDimensions * sizeof(OutputType), kCacheLineSize);
+
+    // Size of the forward propagation buffer used from the input layer to this layer
+    static constexpr std::size_t kBufferSize =
+        PreviousLayer::kBufferSize + kSelfBufferSize;
+
+    // Hash value embedded in the evaluation file
+    static constexpr std::uint32_t GetHashValue() {
+      std::uint32_t hash_value = 0xCC03DAE4u;
+      hash_value += kOutputDimensions;
+      hash_value ^= PreviousLayer::GetHashValue() >> 1;
+      hash_value ^= PreviousLayer::GetHashValue() << 31;
+      return hash_value;
+    }
+
+   // Read network parameters
+    bool ReadParameters(std::istream& stream) {
+      if (!previous_layer_.ReadParameters(stream)) return false;
+      stream.read(reinterpret_cast<char*>(biases_),
+                  kOutputDimensions * sizeof(BiasType));
+      stream.read(reinterpret_cast<char*>(weights_),
+                  kOutputDimensions * kPaddedInputDimensions *
+                  sizeof(WeightType));
+      return !stream.fail();
+    }
+
+    // Forward propagation
+    const OutputType* Propagate(
+        const TransformedFeatureType* transformed_features, char* buffer) const {
+      const auto input = previous_layer_.Propagate(
+          transformed_features, buffer + kSelfBufferSize);
+      const auto output = reinterpret_cast<OutputType*>(buffer);
+
+  #if defined(USE_AVX512)
+      constexpr IndexType kNumChunks = kPaddedInputDimensions / (kSimdWidth * 2);
+      const __m512i kOnes = _mm512_set1_epi16(1);
+      const auto input_vector = reinterpret_cast<const __m512i*>(input);
+
+  #elif defined(USE_AVX2)
+      constexpr IndexType kNumChunks = kPaddedInputDimensions / kSimdWidth;
+      const __m256i kOnes = _mm256_set1_epi16(1);
+      const auto input_vector = reinterpret_cast<const __m256i*>(input);
+
+  #elif defined(USE_SSSE3)
+      constexpr IndexType kNumChunks = kPaddedInputDimensions / kSimdWidth;
+      const __m128i kOnes = _mm_set1_epi16(1);
+      const auto input_vector = reinterpret_cast<const __m128i*>(input);
+
+  #elif defined(USE_NEON)
+      constexpr IndexType kNumChunks = kPaddedInputDimensions / kSimdWidth;
+      const auto input_vector = reinterpret_cast<const int8x8_t*>(input);
+  #endif
+
+      for (IndexType i = 0; i < kOutputDimensions; ++i) {
+        const IndexType offset = i * kPaddedInputDimensions;
+
+  #if defined(USE_AVX512)
+        __m512i sum = _mm512_setzero_si512();
+        const auto row = reinterpret_cast<const __m512i*>(&weights_[offset]);
+        for (IndexType j = 0; j < kNumChunks; ++j) {
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+            __m512i product = _mm512_maddubs_epi16(_mm512_loadu_si512(&input_vector[j]), _mm512_load_si512(&row[j]));
+  #else
+            __m512i product = _mm512_maddubs_epi16(_mm512_load_si512(&input_vector[j]), _mm512_load_si512(&row[j]));
+  #endif
+
+            product = _mm512_madd_epi16(product, kOnes);
+            sum = _mm512_add_epi32(sum, product);
+        }
+        output[i] = _mm512_reduce_add_epi32(sum) + biases_[i];
+
+        // Note: Changing kMaxSimdWidth from 32 to 64 breaks loading existing networks.
+        // As a result kPaddedInputDimensions may not be an even multiple of 64(512bit)
+        // and we have to do one more 256bit chunk.
+        if (kPaddedInputDimensions != kNumChunks * kSimdWidth * 2)
+        {
+            const auto iv_256  = reinterpret_cast<const __m256i*>(input);
+            const auto row_256 = reinterpret_cast<const __m256i*>(&weights_[offset]);
+            int j = kNumChunks * 2;
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)  // See HACK comment below in AVX2.
+            __m256i sum256 = _mm256_maddubs_epi16(_mm256_loadu_si256(&iv_256[j]), _mm256_load_si256(&row_256[j]));
+  #else
+            __m256i sum256 = _mm256_maddubs_epi16(_mm256_load_si256(&iv_256[j]), _mm256_load_si256(&row_256[j]));
+  #endif
+
+            sum256 = _mm256_madd_epi16(sum256, _mm256_set1_epi16(1));
+            sum256 = _mm256_hadd_epi32(sum256, sum256);
+            sum256 = _mm256_hadd_epi32(sum256, sum256);
+            const __m128i lo = _mm256_extracti128_si256(sum256, 0);
+            const __m128i hi = _mm256_extracti128_si256(sum256, 1);
+            output[i] += _mm_cvtsi128_si32(lo) + _mm_cvtsi128_si32(hi);
+        }
+
+  #elif defined(USE_AVX2)
+        __m256i sum = _mm256_setzero_si256();
+        const auto row = reinterpret_cast<const __m256i*>(&weights_[offset]);
+        for (IndexType j = 0; j < kNumChunks; ++j) {
+          __m256i product = _mm256_maddubs_epi16(
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+            // HACK: Use _mm256_loadu_si256() instead of _mm256_load_si256. Because the binary
+            //       compiled with g++ in MSYS2 crashes here because the output memory is not aligned
+            //       even though alignas is specified.
+            _mm256_loadu_si256
+  #else
+            _mm256_load_si256
+  #endif
+
+            (&input_vector[j]), _mm256_load_si256(&row[j]));
+          product = _mm256_madd_epi16(product, kOnes);
+          sum = _mm256_add_epi32(sum, product);
+        }
+        sum = _mm256_hadd_epi32(sum, sum);
+        sum = _mm256_hadd_epi32(sum, sum);
+        const __m128i lo = _mm256_extracti128_si256(sum, 0);
+        const __m128i hi = _mm256_extracti128_si256(sum, 1);
+        output[i] = _mm_cvtsi128_si32(lo) + _mm_cvtsi128_si32(hi) + biases_[i];
+
+  #elif defined(USE_SSSE3)
+        __m128i sum = _mm_cvtsi32_si128(biases_[i]);
+        const auto row = reinterpret_cast<const __m128i*>(&weights_[offset]);
+        for (IndexType j = 0; j < kNumChunks; ++j) {
+          __m128i product = _mm_maddubs_epi16(
+              _mm_load_si128(&input_vector[j]), _mm_load_si128(&row[j]));
+          product = _mm_madd_epi16(product, kOnes);
+          sum = _mm_add_epi32(sum, product);
+        }
+        sum = _mm_hadd_epi32(sum, sum);
+        sum = _mm_hadd_epi32(sum, sum);
+        output[i] = _mm_cvtsi128_si32(sum);
+
+  #elif defined(USE_NEON)
+        int32x4_t sum = {biases_[i]};
+        const auto row = reinterpret_cast<const int8x8_t*>(&weights_[offset]);
+        for (IndexType j = 0; j < kNumChunks; ++j) {
+          int16x8_t product = vmull_s8(input_vector[j * 2], row[j * 2]);
+          product = vmlal_s8(product, input_vector[j * 2 + 1], row[j * 2 + 1]);
+          sum = vpadalq_s16(sum, product);
+        }
+        output[i] = sum[0] + sum[1] + sum[2] + sum[3];
+
+  #else
+        OutputType sum = biases_[i];
+        for (IndexType j = 0; j < kInputDimensions; ++j) {
+          sum += weights_[offset + j] * input[j];
+        }
+        output[i] = sum;
+  #endif
+
+      }
+      return output;
+    }
+
+   private:
+    using BiasType = OutputType;
+    using WeightType = std::int8_t;
+
+    PreviousLayer previous_layer_;
+
+    alignas(kCacheLineSize) BiasType biases_[kOutputDimensions];
+    alignas(kCacheLineSize)
+        WeightType weights_[kOutputDimensions * kPaddedInputDimensions];
+  };
+
+}  // namespace Eval::NNUE::Layers
+
+#endif // #ifndef NNUE_LAYERS_AFFINE_TRANSFORM_H_INCLUDED

src/nnue/layers/clipped_relu.h

@@ -0,0 +1,186 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Definition of layer ClippedReLU of NNUE evaluation function
+
+#ifndef NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED
+#define NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED
+
+#include "../nnue_common.h"
+
+namespace Eval::NNUE::Layers {
+
+  // Clipped ReLU
+  template <typename PreviousLayer>
+  class ClippedReLU {
+   public:
+    // Input/output type
+    using InputType = typename PreviousLayer::OutputType;
+    using OutputType = std::uint8_t;
+    static_assert(std::is_same<InputType, std::int32_t>::value, "");
+
+    // Number of input/output dimensions
+    static constexpr IndexType kInputDimensions =
+        PreviousLayer::kOutputDimensions;
+    static constexpr IndexType kOutputDimensions = kInputDimensions;
+
+    // Size of forward propagation buffer used in this layer
+    static constexpr std::size_t kSelfBufferSize =
+        CeilToMultiple(kOutputDimensions * sizeof(OutputType), kCacheLineSize);
+
+    // Size of the forward propagation buffer used from the input layer to this layer
+    static constexpr std::size_t kBufferSize =
+        PreviousLayer::kBufferSize + kSelfBufferSize;
+
+    // Hash value embedded in the evaluation file
+    static constexpr std::uint32_t GetHashValue() {
+      std::uint32_t hash_value = 0x538D24C7u;
+      hash_value += PreviousLayer::GetHashValue();
+      return hash_value;
+    }
+
+    // Read network parameters
+    bool ReadParameters(std::istream& stream) {
+      return previous_layer_.ReadParameters(stream);
+    }
+
+    // Forward propagation
+    const OutputType* Propagate(
+        const TransformedFeatureType* transformed_features, char* buffer) const {
+      const auto input = previous_layer_.Propagate(
+          transformed_features, buffer + kSelfBufferSize);
+      const auto output = reinterpret_cast<OutputType*>(buffer);
+
+  #if defined(USE_AVX2)
+      constexpr IndexType kNumChunks = kInputDimensions / kSimdWidth;
+      const __m256i kZero = _mm256_setzero_si256();
+      const __m256i kOffsets = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0);
+      const auto in = reinterpret_cast<const __m256i*>(input);
+      const auto out = reinterpret_cast<__m256i*>(output);
+      for (IndexType i = 0; i < kNumChunks; ++i) {
+        const __m256i words0 = _mm256_srai_epi16(_mm256_packs_epi32(
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+          // HACK: Use _mm256_loadu_si256() instead of _mm256_load_si256. Because the binary
+          //       compiled with g++ in MSYS2 crashes here because the output memory is not aligned
+          //       even though alignas is specified.
+          _mm256_loadu_si256
+  #else
+          _mm256_load_si256
+  #endif
+
+          (&in[i * 4 + 0]),
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+          _mm256_loadu_si256
+  #else
+          _mm256_load_si256
+  #endif
+
+          (&in[i * 4 + 1])), kWeightScaleBits);
+        const __m256i words1 = _mm256_srai_epi16(_mm256_packs_epi32(
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+          _mm256_loadu_si256
+  #else
+          _mm256_load_si256
+  #endif
+
+          (&in[i * 4 + 2]),
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+          _mm256_loadu_si256
+  #else
+          _mm256_load_si256
+  #endif
+
+          (&in[i * 4 + 3])), kWeightScaleBits);
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+        _mm256_storeu_si256
+  #else
+        _mm256_store_si256
+  #endif
+
+          (&out[i], _mm256_permutevar8x32_epi32(_mm256_max_epi8(
+            _mm256_packs_epi16(words0, words1), kZero), kOffsets));
+      }
+      constexpr IndexType kStart = kNumChunks * kSimdWidth;
+
+  #elif defined(USE_SSSE3)
+      constexpr IndexType kNumChunks = kInputDimensions / kSimdWidth;
+
+  #ifdef USE_SSE41
+      const __m128i kZero = _mm_setzero_si128();
+  #else
+      const __m128i k0x80s = _mm_set1_epi8(-128);
+  #endif
+
+      const auto in = reinterpret_cast<const __m128i*>(input);
+      const auto out = reinterpret_cast<__m128i*>(output);
+      for (IndexType i = 0; i < kNumChunks; ++i) {
+        const __m128i words0 = _mm_srai_epi16(_mm_packs_epi32(
+            _mm_load_si128(&in[i * 4 + 0]),
+            _mm_load_si128(&in[i * 4 + 1])), kWeightScaleBits);
+        const __m128i words1 = _mm_srai_epi16(_mm_packs_epi32(
+            _mm_load_si128(&in[i * 4 + 2]),
+            _mm_load_si128(&in[i * 4 + 3])), kWeightScaleBits);
+        const __m128i packedbytes = _mm_packs_epi16(words0, words1);
+        _mm_store_si128(&out[i],
+
+  #ifdef USE_SSE41
+          _mm_max_epi8(packedbytes, kZero)
+  #else
+          _mm_subs_epi8(_mm_adds_epi8(packedbytes, k0x80s), k0x80s)
+  #endif
+
+        );
+      }
+      constexpr IndexType kStart = kNumChunks * kSimdWidth;
+
+  #elif defined(USE_NEON)
+      constexpr IndexType kNumChunks = kInputDimensions / (kSimdWidth / 2);
+      const int8x8_t kZero = {0};
+      const auto in = reinterpret_cast<const int32x4_t*>(input);
+      const auto out = reinterpret_cast<int8x8_t*>(output);
+      for (IndexType i = 0; i < kNumChunks; ++i) {
+        int16x8_t shifted;
+        const auto pack = reinterpret_cast<int16x4_t*>(&shifted);
+        pack[0] = vqshrn_n_s32(in[i * 2 + 0], kWeightScaleBits);
+        pack[1] = vqshrn_n_s32(in[i * 2 + 1], kWeightScaleBits);
+        out[i] = vmax_s8(vqmovn_s16(shifted), kZero);
+      }
+      constexpr IndexType kStart = kNumChunks * (kSimdWidth / 2);
+  #else
+      constexpr IndexType kStart = 0;
+  #endif
+
+      for (IndexType i = kStart; i < kInputDimensions; ++i) {
+        output[i] = static_cast<OutputType>(
+            std::max(0, std::min(127, input[i] >> kWeightScaleBits)));
+      }
+      return output;
+    }
+
+   private:
+    PreviousLayer previous_layer_;
+  };
+
+}  // namespace Eval::NNUE::Layers
+
+#endif // NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED

src/nnue/layers/input_slice.h

@@ -0,0 +1,68 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// NNUE evaluation function layer InputSlice definition
+
+#ifndef NNUE_LAYERS_INPUT_SLICE_H_INCLUDED
+#define NNUE_LAYERS_INPUT_SLICE_H_INCLUDED
+
+#include "../nnue_common.h"
+
+namespace Eval::NNUE::Layers {
+
+// Input layer
+template <IndexType OutputDimensions, IndexType Offset = 0>
+class InputSlice {
+ public:
+  // Need to maintain alignment
+  static_assert(Offset % kMaxSimdWidth == 0, "");
+
+  // Output type
+  using OutputType = TransformedFeatureType;
+
+  // Output dimensionality
+  static constexpr IndexType kOutputDimensions = OutputDimensions;
+
+  // Size of forward propagation buffer used from the input layer to this layer
+  static constexpr std::size_t kBufferSize = 0;
+
+  // Hash value embedded in the evaluation file
+  static constexpr std::uint32_t GetHashValue() {
+    std::uint32_t hash_value = 0xEC42E90Du;
+    hash_value ^= kOutputDimensions ^ (Offset << 10);
+    return hash_value;
+  }
+
+  // Read network parameters
+  bool ReadParameters(std::istream& /*stream*/) {
+    return true;
+  }
+
+  // Forward propagation
+  const OutputType* Propagate(
+      const TransformedFeatureType* transformed_features,
+      char* /*buffer*/) const {
+    return transformed_features + Offset;
+  }
+
+ private:
+};
+
+}  // namespace Layers
+
+#endif // #ifndef NNUE_LAYERS_INPUT_SLICE_H_INCLUDED

src/nnue/nnue_accumulator.h

@@ -0,0 +1,39 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Class for difference calculation of NNUE evaluation function
+
+#ifndef NNUE_ACCUMULATOR_H_INCLUDED
+#define NNUE_ACCUMULATOR_H_INCLUDED
+
+#include "nnue_architecture.h"
+
+namespace Eval::NNUE {
+
+  // Class that holds the result of affine transformation of input features
+  struct alignas(32) Accumulator {
+    std::int16_t
+        accumulation[2][kRefreshTriggers.size()][kTransformedFeatureDimensions];
+    Value score;
+    bool computed_accumulation;
+    bool computed_score;
+  };
+
+}  // namespace Eval::NNUE
+
+#endif // NNUE_ACCUMULATOR_H_INCLUDED

src/nnue/nnue_architecture.h

@@ -0,0 +1,38 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Input features and network structure used in NNUE evaluation function
+
+#ifndef NNUE_ARCHITECTURE_H_INCLUDED
+#define NNUE_ARCHITECTURE_H_INCLUDED
+
+// Defines the network structure
+#include "architectures/halfkp_256x2-32-32.h"
+
+namespace Eval::NNUE {
+
+  static_assert(kTransformedFeatureDimensions % kMaxSimdWidth == 0, "");
+  static_assert(Network::kOutputDimensions == 1, "");
+  static_assert(std::is_same<Network::OutputType, std::int32_t>::value, "");
+
+  // Trigger for full calculation instead of difference calculation
+  constexpr auto kRefreshTriggers = RawFeatures::kRefreshTriggers;
+
+}  // namespace Eval::NNUE
+
+#endif // #ifndef NNUE_ARCHITECTURE_H_INCLUDED

src/nnue/nnue_common.h

@@ -0,0 +1,77 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// Constants used in NNUE evaluation function
+
+#ifndef NNUE_COMMON_H_INCLUDED
+#define NNUE_COMMON_H_INCLUDED
+
+#if defined(USE_AVX2)
+#include <immintrin.h>
+
+#elif defined(USE_SSE41)
+#include <smmintrin.h>
+
+#elif defined(USE_SSSE3)
+#include <tmmintrin.h>
+
+#elif defined(USE_SSE2)
+#include <emmintrin.h>
+
+#elif defined(USE_NEON)
+#include <arm_neon.h>
+#endif
+
+namespace Eval::NNUE {
+
+  // Version of the evaluation file
+  constexpr std::uint32_t kVersion = 0x7AF32F16u;
+
+  // Constant used in evaluation value calculation
+  constexpr int FV_SCALE = 16;
+  constexpr int kWeightScaleBits = 6;
+
+  // Size of cache line (in bytes)
+  constexpr std::size_t kCacheLineSize = 64;
+
+  // SIMD width (in bytes)
+  #if defined(USE_AVX2)
+  constexpr std::size_t kSimdWidth = 32;
+
+  #elif defined(USE_SSE2)
+  constexpr std::size_t kSimdWidth = 16;
+
+  #elif defined(USE_NEON)
+  constexpr std::size_t kSimdWidth = 16;
+  #endif
+
+  constexpr std::size_t kMaxSimdWidth = 32;
+
+  // Type of input feature after conversion
+  using TransformedFeatureType = std::uint8_t;
+  using IndexType = std::uint32_t;
+
+  // Round n up to be a multiple of base
+  template <typename IntType>
+  constexpr IntType CeilToMultiple(IntType n, IntType base) {
+    return (n + base - 1) / base * base;
+  }
+
+}  // namespace Eval::NNUE
+
+#endif // #ifndef NNUE_COMMON_H_INCLUDED

src/nnue/nnue_feature_transformer.h

@@ -0,0 +1,355 @@
+/*
+  Stockfish, a UCI chess playing engine derived from Glaurung 2.1
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+
+  Stockfish is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  Stockfish is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+// A class that converts the input features of the NNUE evaluation function
+
+#ifndef NNUE_FEATURE_TRANSFORMER_H_INCLUDED
+#define NNUE_FEATURE_TRANSFORMER_H_INCLUDED
+
+#include "nnue_common.h"
+#include "nnue_architecture.h"
+#include "features/index_list.h"
+
+#include <cstring> // std::memset()
+
+namespace Eval::NNUE {
+
+  // Input feature converter
+  class FeatureTransformer {
+
+   private:
+    // Number of output dimensions for one side
+    static constexpr IndexType kHalfDimensions = kTransformedFeatureDimensions;
+
+   public:
+    // Output type
+    using OutputType = TransformedFeatureType;
+
+    // Number of input/output dimensions
+    static constexpr IndexType kInputDimensions = RawFeatures::kDimensions;
+    static constexpr IndexType kOutputDimensions = kHalfDimensions * 2;
+
+    // Size of forward propagation buffer
+    static constexpr std::size_t kBufferSize =
+        kOutputDimensions * sizeof(OutputType);
+
+    // Hash value embedded in the evaluation file
+    static constexpr std::uint32_t GetHashValue() {
+      return RawFeatures::kHashValue ^ kOutputDimensions;
+    }
+
+    // Read network parameters
+    bool ReadParameters(std::istream& stream) {
+      stream.read(reinterpret_cast<char*>(biases_),
+                  kHalfDimensions * sizeof(BiasType));
+      stream.read(reinterpret_cast<char*>(weights_),
+                  kHalfDimensions * kInputDimensions * sizeof(WeightType));
+      return !stream.fail();
+    }
+
+    // Proceed with the difference calculation if possible
+    bool UpdateAccumulatorIfPossible(const Position& pos) const {
+      const auto now = pos.state();
+      if (now->accumulator.computed_accumulation) {
+        return true;
+      }
+      const auto prev = now->previous;
+      if (prev && prev->accumulator.computed_accumulation) {
+        UpdateAccumulator(pos);
+        return true;
+      }
+      return false;
+    }
+
+    // Convert input features
+    void Transform(const Position& pos, OutputType* output, bool refresh) const {
+      if (refresh || !UpdateAccumulatorIfPossible(pos)) {
+        RefreshAccumulator(pos);
+      }
+      const auto& accumulation = pos.state()->accumulator.accumulation;
+
+  #if defined(USE_AVX2)
+      constexpr IndexType kNumChunks = kHalfDimensions / kSimdWidth;
+      constexpr int kControl = 0b11011000;
+      const __m256i kZero = _mm256_setzero_si256();
+
+  #elif defined(USE_SSSE3)
+      constexpr IndexType kNumChunks = kHalfDimensions / kSimdWidth;
+
+  #ifdef USE_SSE41
+      const __m128i kZero = _mm_setzero_si128();
+  #else
+      const __m128i k0x80s = _mm_set1_epi8(-128);
+  #endif
+
+  #elif defined(USE_NEON)
+      constexpr IndexType kNumChunks = kHalfDimensions / (kSimdWidth / 2);
+      const int8x8_t kZero = {0};
+  #endif
+
+      const Color perspectives[2] = {pos.side_to_move(), ~pos.side_to_move()};
+      for (IndexType p = 0; p < 2; ++p) {
+        const IndexType offset = kHalfDimensions * p;
+
+  #if defined(USE_AVX2)
+        auto out = reinterpret_cast<__m256i*>(&output[offset]);
+        for (IndexType j = 0; j < kNumChunks; ++j) {
+          __m256i sum0 =
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+            // HACK: Use _mm256_loadu_si256() instead of _mm256_load_si256. Because the binary
+            //       compiled with g++ in MSYS2 crashes here because the output memory is not aligned
+            //       even though alignas is specified.
+            _mm256_loadu_si256
+  #else
+            _mm256_load_si256
+  #endif
+
+            (&reinterpret_cast<const __m256i*>(
+              accumulation[perspectives[p]][0])[j * 2 + 0]);
+          __m256i sum1 =
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+            _mm256_loadu_si256
+  #else
+            _mm256_load_si256
+  #endif
+
+            (&reinterpret_cast<const __m256i*>(
+              accumulation[perspectives[p]][0])[j * 2 + 1]);
+
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+          _mm256_storeu_si256
+  #else
+          _mm256_store_si256
+  #endif
+
+          (&out[j], _mm256_permute4x64_epi64(_mm256_max_epi8(
+              _mm256_packs_epi16(sum0, sum1), kZero), kControl));
+        }
+
+  #elif defined(USE_SSSE3)
+        auto out = reinterpret_cast<__m128i*>(&output[offset]);
+        for (IndexType j = 0; j < kNumChunks; ++j) {
+          __m128i sum0 = _mm_load_si128(&reinterpret_cast<const __m128i*>(
+              accumulation[perspectives[p]][0])[j * 2 + 0]);
+          __m128i sum1 = _mm_load_si128(&reinterpret_cast<const __m128i*>(
+              accumulation[perspectives[p]][0])[j * 2 + 1]);
+      const __m128i packedbytes = _mm_packs_epi16(sum0, sum1);
+
+          _mm_store_si128(&out[j],
+
+  #ifdef USE_SSE41
+            _mm_max_epi8(packedbytes, kZero)
+  #else
+            _mm_subs_epi8(_mm_adds_epi8(packedbytes, k0x80s), k0x80s)
+  #endif
+
+          );
+        }
+
+  #elif defined(USE_NEON)
+        const auto out = reinterpret_cast<int8x8_t*>(&output[offset]);
+        for (IndexType j = 0; j < kNumChunks; ++j) {
+          int16x8_t sum = reinterpret_cast<const int16x8_t*>(
+              accumulation[perspectives[p]][0])[j];
+          out[j] = vmax_s8(vqmovn_s16(sum), kZero);
+        }
+
+  #else
+        for (IndexType j = 0; j < kHalfDimensions; ++j) {
+          BiasType sum = accumulation[static_cast<int>(perspectives[p])][0][j];
+          output[offset + j] = static_cast<OutputType>(
+              std::max<int>(0, std::min<int>(127, sum)));
+        }
+  #endif
+
+      }
+    }
+
+   private:
+    // Calculate cumulative value without using difference calculation
+    void RefreshAccumulator(const Position& pos) const {
+      auto& accumulator = pos.state()->accumulator;
+      IndexType i = 0;
+      Features::IndexList active_indices[2];
+      RawFeatures::AppendActiveIndices(pos, kRefreshTriggers[i],
+                                       active_indices);
+      for (Color perspective : { WHITE, BLACK }) {
+        std::memcpy(accumulator.accumulation[perspective][i], biases_,
+                   kHalfDimensions * sizeof(BiasType));
+        for (const auto index : active_indices[perspective]) {
+          const IndexType offset = kHalfDimensions * index;
+
+  #if defined(USE_AVX2)
+          auto accumulation = reinterpret_cast<__m256i*>(
+              &accumulator.accumulation[perspective][i][0]);
+          auto column = reinterpret_cast<const __m256i*>(&weights_[offset]);
+          constexpr IndexType kNumChunks = kHalfDimensions / (kSimdWidth / 2);
+          for (IndexType j = 0; j < kNumChunks; ++j) {
+  #if defined(__MINGW32__) || defined(__MINGW64__)
+            _mm256_storeu_si256(&accumulation[j], _mm256_add_epi16(_mm256_loadu_si256(&accumulation[j]), column[j]));
+  #else
+            accumulation[j] = _mm256_add_epi16(accumulation[j], column[j]);
+  #endif
+          }
+
+  #elif defined(USE_SSE2)
+          auto accumulation = reinterpret_cast<__m128i*>(
+              &accumulator.accumulation[perspective][i][0]);
+          auto column = reinterpret_cast<const __m128i*>(&weights_[offset]);
+          constexpr IndexType kNumChunks = kHalfDimensions / (kSimdWidth / 2);
+          for (IndexType j = 0; j < kNumChunks; ++j) {
+            accumulation[j] = _mm_add_epi16(accumulation[j], column[j]);
+          }
+
+  #elif defined(USE_NEON)
+          auto accumulation = reinterpret_cast<int16x8_t*>(
+              &accumulator.accumulation[perspective][i][0]);
+          auto column = reinterpret_cast<const int16x8_t*>(&weights_[offset]);
+          constexpr IndexType kNumChunks = kHalfDimensions / (kSimdWidth / 2);
+          for (IndexType j = 0; j < kNumChunks; ++j) {
+            accumulation[j] = vaddq_s16(accumulation[j], column[j]);
+          }
+
+  #else
+          for (IndexType j = 0; j < kHalfDimensions; ++j) {
+            accumulator.accumulation[perspective][i][j] += weights_[offset + j];
+          }
+  #endif
+
+        }
+      }
+
+      accumulator.computed_accumulation = true;
+      accumulator.computed_score = false;
+    }
+
+    // Calculate cumulative value using difference calculation
+    void UpdateAccumulator(const Position& pos) const {
+      const auto prev_accumulator = pos.state()->previous->accumulator;
+      auto& accumulator = pos.state()->accumulator;
+      IndexType i = 0;
+      Features::IndexList removed_indices[2], added_indices[2];
+      bool reset[2];
+      RawFeatures::AppendChangedIndices(pos, kRefreshTriggers[i],
+                                        removed_indices, added_indices, reset);
+      for (Color perspective : { WHITE, BLACK }) {
+
+  #if defined(USE_AVX2)
+        constexpr IndexType kNumChunks = kHalfDimensions / (kSimdWidth / 2);
+        auto accumulation = reinterpret_cast<__m256i*>(
+            &accumulator.accumulation[perspective][i][0]);
+
+  #elif defined(USE_SSE2)
+        constexpr IndexType kNumChunks = kHalfDimensions / (kSimdWidth / 2);
+        auto accumulation = reinterpret_cast<__m128i*>(
+            &accumulator.accumulation[perspective][i][0]);
+
+  #elif defined(USE_NEON)
+        constexpr IndexType kNumChunks = kHalfDimensions / (kSimdWidth / 2);
+        auto accumulation = reinterpret_cast<int16x8_t*>(
+            &accumulator.accumulation[perspective][i][0]);
+  #endif
+
+        if (reset[perspective]) {
+          std::memcpy(accumulator.accumulation[perspective][i], biases_,
+                      kHalfDimensions * sizeof(BiasType));
+        } else {
+          std::memcpy(accumulator.accumulation[perspective][i],
+                      prev_accumulator.accumulation[perspective][i],
+                      kHalfDimensions * sizeof(BiasType));
+          // Difference calculation for the deactivated features
+          for (const auto index : removed_indices[perspective]) {
+            const IndexType offset = kHalfDimensions * index;
+
+  #if defined(USE_AVX2)
+            auto column = reinterpret_cast<const __m256i*>(&weights_[offset]);
+            for (IndexType j = 0; j < kNumChunks; ++j) {
+              accumulation[j] = _mm256_sub_epi16(accumulation[j], column[j]);
+            }
+
+  #elif defined(USE_SSE2)
+            auto column = reinterpret_cast<const __m128i*>(&weights_[offset]);
+            for (IndexType j = 0; j < kNumChunks; ++j) {
+              accumulation[j] = _mm_sub_epi16(accumulation[j], column[j]);
+            }
+
+  #elif defined(USE_NEON)
+            auto column = reinterpret_cast<const int16x8_t*>(&weights_[offset]);
+            for (IndexType j = 0; j < kNumChunks; ++j) {
+              accumulation[j] = vsubq_s16(accumulation[j], column[j]);
+            }
+
+  #else
+            for (IndexType j = 0; j < kHalfDimensions; ++j) {
+              accumulator.accumulation[perspective][i][j] -=
+                  weights_[offset + j];
+            }
+  #endif
+
+          }
+        }
+        { // Difference calculation for the activated features
+          for (const auto index : added_indices[perspective]) {
+            const IndexType offset = kHalfDimensions * index;
+
+  #if defined(USE_AVX2)
+            auto column = reinterpret_cast<const __m256i*>(&weights_[offset]);
+            for (IndexType j = 0; j < kNumChunks; ++j) {
+              accumulation[j] = _mm256_add_epi16(accumulation[j], column[j]);
+            }
+
+  #elif defined(USE_SSE2)
+            auto column = reinterpret_cast<const __m128i*>(&weights_[offset]);
+            for (IndexType j = 0; j < kNumChunks; ++j) {
+              accumulation[j] = _mm_add_epi16(accumulation[j], column[j]);
+            }
+
+  #elif defined(USE_NEON)
+            auto column = reinterpret_cast<const int16x8_t*>(&weights_[offset]);
+            for (IndexType j = 0; j < kNumChunks; ++j) {
+              accumulation[j] = vaddq_s16(accumulation[j], column[j]);
+            }
+
+  #else
+            for (IndexType j = 0; j < kHalfDimensions; ++j) {
+              accumulator.accumulation[perspective][i][j] +=
+                  weights_[offset + j];
+            }
+  #endif
+
+          }
+        }
+      }
+
+      accumulator.computed_accumulation = true;
+      accumulator.computed_score = false;
+    }
+
+    using BiasType = std::int16_t;
+    using WeightType = std::int16_t;
+
+    alignas(kCacheLineSize) BiasType biases_[kHalfDimensions];
+    alignas(kCacheLineSize)
+        WeightType weights_[kHalfDimensions * kInputDimensions];
+  };
+
+}  // namespace Eval::NNUE
+
+#endif // #ifndef NNUE_FEATURE_TRANSFORMER_H_INCLUDED

src/pawns.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/pawns.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/position.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -200,6 +198,9 @@ Position& Position::set(const string& fenStr, bool isChess960, StateInfo* si, Th
   std::fill_n(&pieceList[0][0], sizeof(pieceList) / sizeof(Square), SQ_NONE);
   st = si;
 
+  // Each piece on board gets a unique ID used to track the piece later
+  PieceId piece_id, next_piece_id = PIECE_ID_ZERO;
+
   ss >> std::noskipws;
 
   // 1. Piece placement
@@ -213,7 +214,19 @@ Position& Position::set(const string& fenStr, bool isChess960, StateInfo* si, Th
 
       else if ((idx = PieceToChar.find(token)) != string::npos)
       {
-          put_piece(Piece(idx), sq);
+          auto pc = Piece(idx);
+          put_piece(pc, sq);
+
+          if (Eval::useNNUE)
+          {
+              // Kings get a fixed ID, other pieces get ID in order of placement
+              piece_id =
+                (idx == W_KING) ? PIECE_ID_WKING :
+                (idx == B_KING) ? PIECE_ID_BKING :
+                next_piece_id++;
+              evalList.put_piece(piece_id, sq, pc);
+          }
+
           ++sq;
       }
   }
@@ -705,6 +718,14 @@ void Position::do_move(Move m, StateInfo& newSt, bool givesCheck) {
   ++st->rule50;
   ++st->pliesFromNull;
 
+  // Used by NNUE
+  st->accumulator.computed_accumulation = false;
+  st->accumulator.computed_score = false;
+  PieceId dp0 = PIECE_ID_NONE;
+  PieceId dp1 = PIECE_ID_NONE;
+  auto& dp = st->dirtyPiece;
+  dp.dirty_num = 1;
+
   Color us = sideToMove;
   Color them = ~us;
   Square from = from_sq(m);
@@ -752,6 +773,16 @@ void Position::do_move(Move m, StateInfo& newSt, bool givesCheck) {
       else
           st->nonPawnMaterial[them] -= PieceValue[MG][captured];
 
+      if (Eval::useNNUE)
+      {
+          dp.dirty_num = 2; // 2 pieces moved
+          dp1 = piece_id_on(capsq);
+          dp.pieceId[1] = dp1;
+          dp.old_piece[1] = evalList.piece_with_id(dp1);
+          evalList.put_piece(dp1, capsq, NO_PIECE);
+          dp.new_piece[1] = evalList.piece_with_id(dp1);
+      }
+
       // Update board and piece lists
       remove_piece(capsq);
 
@@ -787,7 +818,18 @@ void Position::do_move(Move m, StateInfo& newSt, bool givesCheck) {
 
   // Move the piece. The tricky Chess960 castling is handled earlier
   if (type_of(m) != CASTLING)
+  {
+      if (Eval::useNNUE)
+      {
+          dp0 = piece_id_on(from);
+          dp.pieceId[0] = dp0;
+          dp.old_piece[0] = evalList.piece_with_id(dp0);
+          evalList.put_piece(dp0, to, pc);
+          dp.new_piece[0] = evalList.piece_with_id(dp0);
+      }
+
       move_piece(from, to);
+  }
 
   // If the moving piece is a pawn do some special extra work
   if (type_of(pc) == PAWN)
@@ -810,6 +852,13 @@ void Position::do_move(Move m, StateInfo& newSt, bool givesCheck) {
           remove_piece(to);
           put_piece(promotion, to);
 
+          if (Eval::useNNUE)
+          {
+              dp0 = piece_id_on(to);
+              evalList.put_piece(dp0, to, promotion);
+              dp.new_piece[0] = evalList.piece_with_id(dp0);
+          }
+
           // Update hash keys
           k ^= Zobrist::psq[pc][to] ^ Zobrist::psq[promotion][to];
           st->pawnKey ^= Zobrist::psq[pc][to];
@@ -901,6 +950,12 @@ void Position::undo_move(Move m) {
   {
       move_piece(to, from); // Put the piece back at the source square
 
+      if (Eval::useNNUE)
+      {
+          PieceId dp0 = st->dirtyPiece.pieceId[0];
+          evalList.put_piece(dp0, from, pc);
+      }
+
       if (st->capturedPiece)
       {
           Square capsq = to;
@@ -917,6 +972,14 @@ void Position::undo_move(Move m) {
           }
 
           put_piece(st->capturedPiece, capsq); // Restore the captured piece
+
+          if (Eval::useNNUE)
+          {
+              PieceId dp1 = st->dirtyPiece.pieceId[1];
+              assert(evalList.piece_with_id(dp1).from[WHITE] == PS_NONE);
+              assert(evalList.piece_with_id(dp1).from[BLACK] == PS_NONE);
+              evalList.put_piece(dp1, capsq, st->capturedPiece);
+          }
       }
   }
 
@@ -938,6 +1001,34 @@ void Position::do_castling(Color us, Square from, Square& to, Square& rfrom, Squ
   rto = relative_square(us, kingSide ? SQ_F1 : SQ_D1);
   to = relative_square(us, kingSide ? SQ_G1 : SQ_C1);
 
+  if (Eval::useNNUE)
+  {
+      PieceId dp0, dp1;
+      auto& dp = st->dirtyPiece;
+      dp.dirty_num = 2; // 2 pieces moved
+
+      if (Do)
+      {
+          dp0 = piece_id_on(from);
+          dp1 = piece_id_on(rfrom);
+          dp.pieceId[0] = dp0;
+          dp.old_piece[0] = evalList.piece_with_id(dp0);
+          evalList.put_piece(dp0, to, make_piece(us, KING));
+          dp.new_piece[0] = evalList.piece_with_id(dp0);
+          dp.pieceId[1] = dp1;
+          dp.old_piece[1] = evalList.piece_with_id(dp1);
+          evalList.put_piece(dp1, rto, make_piece(us, ROOK));
+          dp.new_piece[1] = evalList.piece_with_id(dp1);
+      }
+      else
+      {
+          dp0 = piece_id_on(to);
+          dp1 = piece_id_on(rto);
+          evalList.put_piece(dp0, from, make_piece(us, KING));
+          evalList.put_piece(dp1, rfrom, make_piece(us, ROOK));
+      }
+  }
+
   // Remove both pieces first since squares could overlap in Chess960
   remove_piece(Do ? from : to);
   remove_piece(Do ? rfrom : rto);
@@ -955,7 +1046,14 @@ void Position::do_null_move(StateInfo& newSt) {
   assert(!checkers());
   assert(&newSt != st);
 
+  if (Eval::useNNUE)
+  {
       std::memcpy(&newSt, st, sizeof(StateInfo));
+      st->accumulator.computed_score = false;
+  }
+  else
+      std::memcpy(&newSt, st, offsetof(StateInfo, accumulator));
+
   newSt.previous = st;
   st = &newSt;
 

src/position.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -27,8 +25,11 @@
 #include <string>
 
 #include "bitboard.h"
+#include "evaluate.h"
 #include "types.h"
 
+#include "nnue/nnue_accumulator.h"
+
 
 /// 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
@@ -54,6 +55,10 @@ struct StateInfo {
   Bitboard   pinners[COLOR_NB];
   Bitboard   checkSquares[PIECE_TYPE_NB];
   int        repetition;
+
+  // Used by NNUE
+  Eval::NNUE::Accumulator accumulator;
+  DirtyPiece dirtyPiece;
 };
 
 
@@ -163,6 +168,10 @@ public:
   bool pos_is_ok() const;
   void flip();
 
+  // Used by NNUE
+  StateInfo* state() const;
+  const EvalList* eval_list() const;
+
 private:
   // Initialization helpers (used while setting up a position)
   void set_castling_right(Color c, Square rfrom);
@@ -176,6 +185,9 @@ private:
   template<bool Do>
   void do_castling(Color us, Square from, Square& to, Square& rfrom, Square& rto);
 
+  // ID of a piece on a given square
+  PieceId piece_id_on(Square sq) const;
+
   // Data members
   Piece board[SQUARE_NB];
   Bitboard byTypeBB[PIECE_TYPE_NB];
@@ -192,6 +204,9 @@ private:
   Thread* thisThread;
   StateInfo* st;
   bool chess960;
+
+  // List of pieces used in NNUE evaluation function
+  EvalList evalList;
 };
 
 namespace PSQT {
@@ -426,4 +441,25 @@ inline void Position::do_move(Move m, StateInfo& newSt) {
   do_move(m, newSt, gives_check(m));
 }
 
+inline StateInfo* Position::state() const {
+
+  return st;
+}
+
+inline const EvalList* Position::eval_list() const {
+
+  return &evalList;
+}
+
+inline PieceId Position::piece_id_on(Square sq) const
+{
+
+  assert(piece_on(sq) != NO_PIECE);
+
+  PieceId pid = evalList.piece_id_list[sq];
+  assert(is_ok(pid));
+
+  return pid;
+}
+
 #endif // #ifndef POSITION_H_INCLUDED

src/psqt.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/search.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -227,6 +225,8 @@ void MainThread::search() {
   Time.init(Limits, us, rootPos.game_ply());
   TT.new_search();
 
+  Eval::verify_NNUE();
+
   if (rootMoves.empty())
   {
       rootMoves.emplace_back(MOVE_NONE);

src/search.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/syzygy/tbprobe.cpp

@@ -1,7 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (c) 2013 Ronald de Man
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2016-2020 Marco Costalba, Lucas Braesch
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/syzygy/tbprobe.h

@@ -1,7 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (c) 2013 Ronald de Man
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2016-2020 Marco Costalba, Lucas Braesch
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/thread.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/thread.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/thread_win32_osx.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/timeman.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/timeman.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/tt.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/tt.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/tune.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/tune.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2017 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2018 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/types.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -203,6 +201,22 @@ enum Piece {
   PIECE_NB = 16
 };
 
+// An ID used to track the pieces. Max. 32 pieces on board.
+enum PieceId {
+  PIECE_ID_ZERO   = 0,
+  PIECE_ID_KING   = 30,
+  PIECE_ID_WKING  = 30,
+  PIECE_ID_BKING  = 31,
+  PIECE_ID_NONE   = 32
+};
+
+inline PieceId operator++(PieceId& d, int) {
+
+  PieceId x = d;
+  d = PieceId(int(d) + 1);
+  return x;
+}
+
 constexpr Value PieceValue[PHASE_NB][PIECE_NB] = {
   { VALUE_ZERO, PawnValueMg, KnightValueMg, BishopValueMg, RookValueMg, QueenValueMg, VALUE_ZERO, VALUE_ZERO,
     VALUE_ZERO, PawnValueMg, KnightValueMg, BishopValueMg, RookValueMg, QueenValueMg, VALUE_ZERO, VALUE_ZERO },
@@ -232,6 +246,7 @@ enum Square : int {
   SQ_A8, SQ_B8, SQ_C8, SQ_D8, SQ_E8, SQ_F8, SQ_G8, SQ_H8,
   SQ_NONE,
 
+  SQUARE_ZERO = 0,
   SQUARE_NB   = 64
 };
 
@@ -255,6 +270,94 @@ enum Rank : int {
   RANK_1, RANK_2, RANK_3, RANK_4, RANK_5, RANK_6, RANK_7, RANK_8, RANK_NB
 };
 
+// unique number for each piece type on each square
+enum PieceSquare : uint32_t {
+  PS_NONE     =  0,
+  PS_W_PAWN   =  1,
+  PS_B_PAWN   =  1 * SQUARE_NB + 1,
+  PS_W_KNIGHT =  2 * SQUARE_NB + 1,
+  PS_B_KNIGHT =  3 * SQUARE_NB + 1,
+  PS_W_BISHOP =  4 * SQUARE_NB + 1,
+  PS_B_BISHOP =  5 * SQUARE_NB + 1,
+  PS_W_ROOK   =  6 * SQUARE_NB + 1,
+  PS_B_ROOK   =  7 * SQUARE_NB + 1,
+  PS_W_QUEEN  =  8 * SQUARE_NB + 1,
+  PS_B_QUEEN  =  9 * SQUARE_NB + 1,
+  PS_W_KING   = 10 * SQUARE_NB + 1,
+  PS_END      = PS_W_KING, // pieces without kings (pawns included)
+  PS_B_KING   = 11 * SQUARE_NB + 1,
+  PS_END2     = 12 * SQUARE_NB + 1
+};
+
+struct ExtPieceSquare {
+  PieceSquare from[COLOR_NB];
+};
+
+// Array for finding the PieceSquare corresponding to the piece on the board
+extern ExtPieceSquare kpp_board_index[PIECE_NB];
+
+constexpr bool is_ok(PieceId pid);
+constexpr Square rotate180(Square sq);
+
+// Structure holding which tracked piece (PieceId) is where (PieceSquare)
+class EvalList {
+
+public:
+  // Max. number of pieces without kings is 30 but must be a multiple of 4 in AVX2
+  static const int MAX_LENGTH = 32;
+
+  // Array that holds the piece id for the pieces on the board
+  PieceId piece_id_list[SQUARE_NB];
+
+  // List of pieces, separate from White and Black POV
+  PieceSquare* piece_list_fw() const { return const_cast<PieceSquare*>(pieceListFw); }
+  PieceSquare* piece_list_fb() const { return const_cast<PieceSquare*>(pieceListFb); }
+
+  // Place the piece pc with piece_id on the square sq on the board
+  void put_piece(PieceId piece_id, Square sq, Piece pc)
+  {
+      assert(is_ok(piece_id));
+      if (pc != NO_PIECE)
+      {
+          pieceListFw[piece_id] = PieceSquare(kpp_board_index[pc].from[WHITE] + sq);
+          pieceListFb[piece_id] = PieceSquare(kpp_board_index[pc].from[BLACK] + rotate180(sq));
+          piece_id_list[sq] = piece_id;
+      }
+      else
+      {
+          pieceListFw[piece_id] = PS_NONE;
+          pieceListFb[piece_id] = PS_NONE;
+          piece_id_list[sq] = piece_id;
+      }
+  }
+
+  // Convert the specified piece_id piece to ExtPieceSquare type and return it
+  ExtPieceSquare piece_with_id(PieceId piece_id) const
+  {
+      ExtPieceSquare eps;
+      eps.from[WHITE] = pieceListFw[piece_id];
+      eps.from[BLACK] = pieceListFb[piece_id];
+      return eps;
+  }
+
+private:
+  PieceSquare pieceListFw[MAX_LENGTH];
+  PieceSquare pieceListFb[MAX_LENGTH];
+};
+
+// For differential evaluation of pieces that changed since last turn
+struct DirtyPiece {
+
+  // Number of changed pieces
+  int dirty_num;
+
+  // The ids of changed pieces, max. 2 pieces can change in one move
+  PieceId pieceId[2];
+
+  // What changed from the piece with that piece number
+  ExtPieceSquare old_piece[2];
+  ExtPieceSquare new_piece[2];
+};
 
 /// Score enum stores a middlegame and an endgame value in a single integer (enum).
 /// The least significant 16 bits are used to store the middlegame value and the
@@ -302,6 +405,9 @@ inline T& operator/=(T& d, int i) { return d = T(int(d) / i); }
 ENABLE_FULL_OPERATORS_ON(Value)
 ENABLE_FULL_OPERATORS_ON(Direction)
 
+ENABLE_INCR_OPERATORS_ON(Piece)
+ENABLE_INCR_OPERATORS_ON(PieceSquare)
+ENABLE_INCR_OPERATORS_ON(PieceId)
 ENABLE_INCR_OPERATORS_ON(PieceType)
 ENABLE_INCR_OPERATORS_ON(Square)
 ENABLE_INCR_OPERATORS_ON(File)
@@ -390,6 +496,10 @@ inline Color color_of(Piece pc) {
   return Color(pc >> 3);
 }
 
+constexpr bool is_ok(PieceId pid) {
+  return pid < PIECE_ID_NONE;
+}
+
 constexpr bool is_ok(Square s) {
   return s >= SQ_A1 && s <= SQ_H8;
 }
@@ -426,6 +536,11 @@ constexpr Square to_sq(Move m) {
   return Square(m & 0x3F);
 }
 
+// Return relative square when turning the board 180 degrees
+constexpr Square rotate180(Square sq) {
+  return (Square)(sq ^ 0x3F);
+}
+
 constexpr int from_to(Move m) {
  return m & 0xFFF;
 }

src/uci.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -78,6 +76,20 @@ namespace {
     }
   }
 
+  // trace_eval() prints the evaluation for the current position, consistent with the UCI
+  // options set so far.
+
+  void trace_eval(Position& pos) {
+
+    StateListPtr states(new std::deque<StateInfo>(1));
+    Position p;
+    p.set(pos.fen(), Options["UCI_Chess960"], &states->back(), Threads.main());
+
+    Eval::verify_NNUE();
+
+    sync_cout << "\n" << Eval::trace(p) << sync_endl;
+  }
+
 
   // setoption() is called when engine receives the "setoption" UCI command. The
   // function updates the UCI option ("name") to the given value ("value").
@@ -166,7 +178,7 @@ namespace {
                nodes += Threads.nodes_searched();
             }
             else
-               sync_cout << "\n" << Eval::trace(pos) << sync_endl;
+               trace_eval(pos);
         }
         else if (token == "setoption")  setoption(is);
         else if (token == "position")   position(pos, is, states);
@@ -261,7 +273,7 @@ void UCI::loop(int argc, char* argv[]) {
       else if (token == "flip")     pos.flip();
       else if (token == "bench")    bench(pos, is, states);
       else if (token == "d")        sync_cout << pos << sync_endl;
-      else if (token == "eval")     sync_cout << Eval::trace(pos) << sync_endl;
+      else if (token == "eval")     trace_eval(pos);
       else if (token == "compiler") sync_cout << compiler_info() << sync_endl;
       else
           sync_cout << "Unknown command: " << cmd << sync_endl;

src/uci.h

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by

src/ucioption.cpp

@@ -1,8 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2008 Tord Romstad (Glaurung author)
+  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
-  Copyright (C) 2008-2015 Marco Costalba, Joona Kiiski, Tord Romstad
-  Copyright (C) 2015-2020 Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad
 
   Stockfish is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -42,7 +40,8 @@ void on_hash_size(const Option& o) { TT.resize(size_t(o)); }
 void on_logger(const Option& o) { start_logger(o); }
 void on_threads(const Option& o) { Threads.set(size_t(o)); }
 void on_tb_path(const Option& o) { Tablebases::init(o); }
-
+void on_use_NNUE(const Option& ) { Eval::init_NNUE(); }
+void on_eval_file(const Option& ) { Eval::init_NNUE(); }
 
 /// Our case insensitive less() function as required by UCI protocol
 bool CaseInsensitiveLess::operator() (const string& s1, const string& s2) const {
@@ -79,6 +78,8 @@ void init(OptionsMap& o) {
   o["SyzygyProbeDepth"]      << Option(1, 1, 100);
   o["Syzygy50MoveRule"]      << Option(true);
   o["SyzygyProbeLimit"]      << Option(7, 0, 7);
+  o["Use NNUE"]              << Option(false, on_use_NNUE);
+  o["EvalFile"]              << Option("nn-97f742aaefcd.nnue", on_eval_file);
 }