# 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
+# vnni = yes/no --- -mavx512vnni --- Use Intel Vector Neural Network Instructions 512
# neon = yes/no --- -DUSE_NEON --- Use ARM SIMD architecture
#
# Note that Makefile is space sensitive, so when adding new architectures
# at the end of the line for flag values.
### 2.1. General and architecture defaults
+
+ifeq ($(ARCH),)
+ empty_arch = yes
+endif
+
optimize = yes
debug = no
sanitize = no
bits = 64
prefetch = no
popcnt = no
+mmx = no
sse = no
-sse3 = no
ssse3 = no
sse41 = no
-sse42 = no
avx2 = no
pext = no
avx512 = no
+vnni = no
neon = no
ARCH = x86-64-modern
### 2.2 Architecture specific
+
ifeq ($(ARCH),general-32)
arch = any
bits = 32
arch = i386
bits = 32
prefetch = yes
+ mmx = yes
sse = yes
endif
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
arch = x86_64
prefetch = yes
sse = yes
- sse3 = yes
ssse3 = yes
endif
-ifeq ($(ARCH),x86-64-sse41)
+ifeq ($(ARCH),$(filter $(ARCH),x86-64-sse41-popcnt x86-64-modern))
arch = x86_64
prefetch = yes
popcnt = yes
sse = yes
- sse3 = yes
ssse3 = yes
sse41 = yes
endif
-ifeq ($(ARCH),x86-64-modern)
- arch = x86_64
- prefetch = yes
- popcnt = yes
- sse = yes
- sse3 = yes
- ssse3 = yes
- sse41 = yes
-endif
-
-ifeq ($(ARCH),x86-64-sse42)
+ifeq ($(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 ($(ARCH),x86-64-avx2)
+ifeq ($(ARCH),x86-64-bmi2)
arch = x86_64
prefetch = yes
popcnt = yes
sse = yes
- sse3 = yes
ssse3 = yes
sse41 = yes
- sse42 = yes
avx2 = yes
+ pext = yes
endif
-ifeq ($(ARCH),x86-64-bmi2)
+ifeq ($(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 ($(ARCH),x86-64-avx512)
+ifeq ($(ARCH),x86-64-vnni)
arch = x86_64
prefetch = yes
popcnt = yes
sse = yes
- sse3 = yes
ssse3 = yes
sse41 = yes
- sse42 = yes
avx2 = yes
pext = yes
avx512 = yes
+ vnni = yes
endif
ifeq ($(ARCH),armv7)
ifneq ($(KERNEL),Darwin)
LDFLAGS += -Wl,--no-as-needed
endif
+
+ gccversion = $(shell $(CXX) --version)
+ gccisclang = $(findstring clang,$(gccversion))
endif
ifeq ($(COMP),mingw)
endif
ifeq ($(KERNEL),Darwin)
- CXXFLAGS += -arch $(arch) -mmacosx-version-min=10.15
- LDFLAGS += -arch $(arch) -mmacosx-version-min=10.15
+ CXXFLAGS += -arch $(arch) -mmacosx-version-min=10.14
+ LDFLAGS += -arch $(arch) -mmacosx-version-min=10.14
endif
### Travis CI script uses COMPILER to overwrite CXX
ifeq ($(avx512),yes)
CXXFLAGS += -DUSE_AVX512
ifeq ($(comp),$(filter $(comp),gcc clang mingw))
- CXXFLAGS += -mavx512bw
+ CXXFLAGS += -mavx512f -mavx512bw
endif
endif
-ifeq ($(sse42),yes)
- CXXFLAGS += -DUSE_SSE42
+ifeq ($(vnni),yes)
+ CXXFLAGS += -DUSE_VNNI
ifeq ($(comp),$(filter $(comp),gcc clang mingw))
- CXXFLAGS += -msse4.2
+ CXXFLAGS += -mavx512vnni -mavx512dq -mavx512vl
endif
endif
endif
endif
-ifeq ($(sse3),yes)
- CXXFLAGS += -DUSE_SSE3
+ifeq ($(mmx),yes)
+ CXXFLAGS += -DUSE_MMX
ifeq ($(comp),$(filter $(comp),gcc clang mingw))
- CXXFLAGS += -msse3
+ CXXFLAGS += -mmmx
endif
endif
endif
ifeq ($(arch),x86_64)
- CXXFLAGS += -DUSE_SSE2
+ CXXFLAGS += -msse2 -DUSE_SSE2
endif
### 3.7 pext
### needs access to the optimization flags.
ifeq ($(optimize),yes)
ifeq ($(debug), no)
- ifeq ($(comp),$(filter $(comp),gcc clang))
+ ifeq ($(comp),clang)
+ CXXFLAGS += -flto=thin
+ LDFLAGS += $(CXXFLAGS)
+
+# GCC and CLANG use different methods for parallelizing LTO and CLANG pretends to be
+# GCC on some systems.
+ else ifeq ($(comp),gcc)
+ ifeq ($(gccisclang),)
CXXFLAGS += -flto
+ LDFLAGS += $(CXXFLAGS) -flto=jobserver
+ ifneq ($(findstring MINGW,$(KERNEL)),)
+ LDFLAGS += -save-temps
+ else ifneq ($(findstring MSYS,$(KERNEL)),)
+ LDFLAGS += -save-temps
+ endif
+ else
+ CXXFLAGS += -flto=thin
LDFLAGS += $(CXXFLAGS)
endif
# To use LTO and static linking on windows, the tool chain requires a recent gcc:
# gcc version 10.1 in msys2 or TDM-GCC version 9.2 are know to work, older might not.
# So, only enable it for a cross from Linux by default.
- ifeq ($(comp),mingw)
+ else ifeq ($(comp),mingw)
ifeq ($(KERNEL),Linux)
CXXFLAGS += -flto
- LDFLAGS += $(CXXFLAGS)
+ LDFLAGS += $(CXXFLAGS) -flto=jobserver
endif
endif
endif
@echo ""
@echo "Supported targets:"
@echo ""
+ @echo "help > Display architecture details"
@echo "build > Standard build"
- @echo "profile-build > Standard build with PGO"
+ @echo "net > Download the default nnue net"
+ @echo "profile-build > Faster build (with profile-guided optimization)"
@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-vnni > x86 64-bit with vnni support"
@echo "x86-64-avx512 > x86 64-bit with avx512 support"
@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-sse41-popcnt > x86 64-bit with sse41 and popcnt support"
+ @echo "x86-64-modern > common modern CPU, currently x86-64-sse41-popcnt"
@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 > x86 32-bit (also enables MMX and SSE)"
@echo "x86-32-old > x86 32-bit fall back for old hardware"
@echo "ppc-64 > PPC 64-bit"
@echo "ppc-32 > PPC 32-bit"
@echo ""
@echo "Simple examples. If you don't know what to do, you likely want to run: "
@echo ""
- @echo "make -j build ARCH=x86-64 (This is for 64-bit systems)"
- @echo "make -j build ARCH=x86-32 (This is for 32-bit systems)"
+ @echo "make -j build ARCH=x86-64 (A portable, slow compile for 64-bit systems)"
+ @echo "make -j build ARCH=x86-32 (A portable, slow compile for 32-bit systems)"
@echo ""
- @echo "Advanced examples, for experienced users: "
+ @echo "Advanced examples, for experienced users looking for performance: "
@echo ""
- @echo "make -j build ARCH=x86-64-modern COMP=clang"
- @echo "make -j profile-build ARCH=x86-64-bmi2 COMP=gcc COMPCXX=g++-4.8"
- @echo ""
- @echo "The selected architecture $(ARCH) enables the following configuration: "
+ @echo "make help ARCH=x86-64-bmi2"
+ @echo "make -j profile-build ARCH=x86-64-bmi2 COMP=gcc COMPCXX=g++-9.0"
+ @echo "make -j build ARCH=x86-64-ssse3 COMP=clang"
@echo ""
+ifneq ($(empty_arch), yes)
+ @echo "-------------------------------\n"
+ @echo "The selected architecture $(ARCH) will enable the following configuration: "
@$(MAKE) ARCH=$(ARCH) COMP=$(COMP) config-sanity
+endif
.PHONY: help build profile-build strip install clean net objclean profileclean \
build: config-sanity
$(MAKE) ARCH=$(ARCH) COMP=$(COMP) all
-profile-build: config-sanity objclean profileclean
+profile-build: net config-sanity objclean profileclean
@echo ""
@echo "Step 1/4. Building instrumented executable ..."
$(MAKE) ARCH=$(ARCH) COMP=$(COMP) $(profile_make)
$(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))
+ $(eval curl_or_wget := $(shell if hash curl 2>/dev/null; then echo "curl -skL"; elif hash wget 2>/dev/null; then echo "wget -qO-"; fi))
@if test -f "$(nnuenet)"; then echo "Already available."; else echo "Downloading $(nnuedownloadurl)"; $(curl_or_wget) $(nnuedownloadurl) > $(nnuenet); fi
+ $(eval shasum_command := $(shell if hash shasum 2>/dev/null; then echo "shasum -a 256 "; elif hash sha256sum 2>/dev/null; then echo "sha256sum "; fi))
+ @if [ "$(nnuenet)" != "nn-"`$(shasum_command) $(nnuenet) | cut -c1-12`".nnue" ]; then echo "Failed download or $(nnuenet) corrupted, please delete!"; exit 1; fi
# clean binaries and objects
objclean:
# clean auxiliary profiling files
profileclean:
@rm -rf profdir
- @rm -f bench.txt *.gcda *.gcno ./syzygy/*.gcda ./nnue/*.gcda ./nnue/features/*.gcda
+ @rm -f bench.txt *.gcda *.gcno ./syzygy/*.gcda ./nnue/*.gcda ./nnue/features/*.gcda *.s
@rm -f stockfish.profdata *.profraw
default:
@echo "prefetch: '$(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: '$(avx512)'"
+ @echo "vnni: '$(vnni)'"
@echo "neon: '$(neon)'"
@echo ""
@echo "Flags:"
@test "$(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 "$(avx512)" = "no"
+ @test "$(vnni)" = "yes" || test "$(vnni)" = "no"
@test "$(neon)" = "yes" || test "$(neon)" = "no"
@test "$(comp)" = "gcc" || test "$(comp)" = "icc" || test "$(comp)" = "mingw" || test "$(comp)" = "clang"
$(EXE): $(OBJS)
- $(CXX) -o $@ $(OBJS) $(LDFLAGS)
+ +$(CXX) -o $@ $(OBJS) $(LDFLAGS)
clang-profile-make:
$(MAKE) ARCH=$(ARCH) COMP=$(COMP) \