Expanded support for x86-32 architectures.

add new ARCH targets

x86-32-sse41-popcnt     > x86 32-bit with sse41 and popcnt support
x86-32-sse2             > x86 32-bit with sse2 support
x86-32                  > x86 32-bit generic (with mmx and sse support)

retire x86-32-old (use general-32)

closes https://github.com/official-stockfish/Stockfish/pull/3022

No functional change.

diff --git a/.travis.yml b/.travis.yml
index 45f1bd3d34fe352a2c04a6d64a11aedbb8c565db..12596f1e32700f27e495476fcf104ef4cb48f0d0 100644 (file)
--- a/.travis.yml
+++ b/.travis.yml
@@ -67,9 +67,10 @@ script:
   - make clean && make -j2 ARCH=x86-64 build && ../tests/signature.sh $benchref
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=general-64 build && ../tests/signature.sh $benchref; fi
   - if [[ "$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=x86-64 build && ../tests/signature.sh $benchref
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=general-64 build && ../tests/signature.sh $benchref; fi
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32 optimize=no debug=yes build && ../tests/signature.sh $benchref; fi

+  - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32-sse41-popcnt build && ../tests/signature.sh $benchref; fi
+  - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32-sse2 build && ../tests/signature.sh $benchref; fi

   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32 build && ../tests/signature.sh $benchref; fi
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=general-32 build && ../tests/signature.sh $benchref; fi
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32 build && ../tests/signature.sh $benchref; fi
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=general-32 build && ../tests/signature.sh $benchref; fi

-  - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then make clean && make -j2 ARCH=x86-32-old build && ../tests/signature.sh $benchref; fi

   - if [[ "$TRAVIS_OS_NAME" == "linux" && "$COMP" == "gcc" ]]; then make clean && make -j2 ARCH=x86-64-modern profile-build && ../tests/signature.sh $benchref; fi
 
   # compile only for some more advanced architectures (might not run in travis)
   - if [[ "$TRAVIS_OS_NAME" == "linux" && "$COMP" == "gcc" ]]; then make clean && make -j2 ARCH=x86-64-modern profile-build && ../tests/signature.sh $benchref; fi
 
   # compile only for some more advanced architectures (might not run in travis)

diff --git a/src/Makefile b/src/Makefile
index a3feb68e3e349957c97155a4543ac9c31e97c5b6..79c7333ac26edc43fe73f3a44c8ad29e86296ca2 100644 (file)
--- a/src/Makefile
+++ b/src/Makefile
@@ -67,11 +67,13 @@ endif
 # bits = 64/32        --- -DIS_64BIT       --- 64-/32-bit operating system
 # prefetch = yes/no   --- -DUSE_PREFETCH   --- Use prefetch asm-instruction
 # popcnt = yes/no     --- -DUSE_POPCNT     --- Use popcnt asm-instruction
 # bits = 64/32        --- -DIS_64BIT       --- 64-/32-bit operating system
 # prefetch = yes/no   --- -DUSE_PREFETCH   --- Use prefetch asm-instruction
 # popcnt = yes/no     --- -DUSE_POPCNT     --- Use popcnt asm-instruction

+# pext = yes/no       --- -DUSE_PEXT       --- Use pext x86_64 asm-instruction

 # sse = yes/no        --- -msse            --- Use Intel Streaming SIMD Extensions
 # sse = yes/no        --- -msse            --- Use Intel Streaming SIMD Extensions

+# mmx = yes/no        --- -mmmx            --- Use Intel MMX instructions
+# sse2 = yes/no       --- -msse2           --- Use Intel Streaming SIMD Extensions 2

 # ssse3 = yes/no      --- -mssse3          --- Use Intel Supplemental Streaming SIMD Extensions 3
 # sse41 = yes/no      --- -msse4.1         --- Use Intel Streaming SIMD Extensions 4.1
 # avx2 = yes/no       --- -mavx2           --- Use Intel Advanced Vector Extensions 2
 # ssse3 = yes/no      --- -mssse3          --- Use Intel Supplemental Streaming SIMD Extensions 3
 # sse41 = yes/no      --- -msse4.1         --- Use Intel Streaming SIMD Extensions 4.1
 # 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
 # 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


@@ -92,12 +94,13 @@ sanitize = no
 bits = 64
 prefetch = no
 popcnt = no
 bits = 64
 prefetch = no
 popcnt = no

-mmx = no
+pext = no

 sse = no
 sse = no

+mmx = no
+sse2 = no

 ssse3 = no
 sse41 = no
 avx2 = no
 ssse3 = no
 sse41 = no
 avx2 = no

-pext = no

 avx512 = no
 vnni = no
 neon = no
 avx512 = no
 vnni = no
 neon = no


@@ -106,83 +109,82 @@ STRIP = strip
 
 ### 2.2 Architecture specific
 
 
 ### 2.2 Architecture specific
 

-ifeq ($(ARCH),general-32)
-       arch = any
-       bits = 32
-endif
+ifeq ($(findstring x86,$(ARCH)),x86)

 
 

-ifeq ($(ARCH),x86-32-old)
-       arch = i386
-       bits = 32
-endif
+# x86-32/64

 
 

-ifeq ($(ARCH),x86-32)
+ifeq ($(findstring x86-32,$(ARCH)),x86-32)

        arch = i386
        bits = 32
        arch = i386
        bits = 32

-       prefetch = yes
+       sse = yes

        mmx = yes
        mmx = yes

+else
+       arch = x86_64

        sse = yes
        sse = yes

+       sse2 = yes

 endif
 
 endif
 

-ifeq ($(ARCH),general-64)
-       arch = any
+ifeq ($(findstring -sse,$(ARCH)),-sse)
+       sse = yes

 endif
 
 endif
 

-ifeq ($(ARCH),x86-64)
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -popcnt,$(ARCH)),-popcnt)
+       popcnt = yes
+endif
+
+ifeq ($(findstring -mmx,$(ARCH)),-mmx)
+       mmx = yes
+endif
+
+ifeq ($(findstring -sse2,$(ARCH)),-sse2)

        sse = yes
        sse = yes

+       sse2 = yes

 endif
 
 endif
 

-ifeq ($(ARCH),x86-64-sse3-popcnt)
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -ssse3,$(ARCH)),-ssse3)

        sse = yes
        sse = yes

-       popcnt = yes
+       sse2 = yes
+       ssse3 = yes

 endif
 
 endif
 

-ifeq ($(ARCH),x86-64-ssse3)
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -sse41,$(ARCH)),-sse41)

        sse = yes
        sse = yes

+       sse2 = yes

        ssse3 = yes
        ssse3 = yes

+       sse41 = yes

 endif
 
 endif
 

-ifeq ($(ARCH),$(filter $(ARCH),x86-64-sse41-popcnt x86-64-modern))
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -modern,$(ARCH)),-modern)

        popcnt = yes
        sse = yes
        popcnt = yes
        sse = yes

+       sse2 = yes

        ssse3 = yes
        sse41 = yes
 endif
 
        ssse3 = yes
        sse41 = yes
 endif
 

-ifeq ($(ARCH),x86-64-avx2)
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -avx2,$(ARCH)),-avx2)

        popcnt = yes
        sse = yes
        popcnt = yes
        sse = yes

+       sse2 = yes

        ssse3 = yes
        sse41 = yes
        avx2 = yes
 endif
 
        ssse3 = yes
        sse41 = yes
        avx2 = yes
 endif
 

-ifeq ($(ARCH),x86-64-bmi2)
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -bmi2,$(ARCH)),-bmi2)

        popcnt = yes
        sse = yes
        popcnt = yes
        sse = yes

+       sse2 = yes

        ssse3 = yes
        sse41 = yes
        avx2 = yes
        pext = yes
 endif
 
        ssse3 = yes
        sse41 = yes
        avx2 = yes
        pext = yes
 endif
 

-ifeq ($(ARCH),x86-64-avx512)
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -avx512,$(ARCH)),-avx512)

        popcnt = yes
        sse = yes
        popcnt = yes
        sse = yes

+       sse2 = yes

        ssse3 = yes
        sse41 = yes
        avx2 = yes
        ssse3 = yes
        sse41 = yes
        avx2 = yes


@@ -190,11 +192,10 @@ ifeq ($(ARCH),x86-64-avx512)
        avx512 = yes
 endif
 
        avx512 = yes
 endif
 

-ifeq ($(ARCH),x86-64-vnni)
-       arch = x86_64
-       prefetch = yes
+ifeq ($(findstring -vnni,$(ARCH)),-vnni)

        popcnt = yes
        sse = yes
        popcnt = yes
        sse = yes

+       sse2 = yes

        ssse3 = yes
        sse41 = yes
        avx2 = yes
        ssse3 = yes
        sse41 = yes
        avx2 = yes


@@ -203,6 +204,28 @@ ifeq ($(ARCH),x86-64-vnni)
        vnni = yes
 endif
 
        vnni = yes
 endif
 

+ifeq ($(sse),yes)
+       prefetch = yes
+endif
+
+# 64-bit pext is not available on x86-32
+ifeq ($(bits),32)
+       pext = no
+endif
+
+else
+
+# all other architectures
+
+ifeq ($(ARCH),general-32)
+       arch = any
+       bits = 32
+endif
+
+ifeq ($(ARCH),general-64)
+       arch = any
+endif
+

 ifeq ($(ARCH),armv7)
        arch = armv7
        prefetch = yes
 ifeq ($(ARCH),armv7)
        arch = armv7
        prefetch = yes


@@ -242,6 +265,8 @@ ifeq ($(ARCH),ppc-64)
        prefetch = yes
 endif
 
        prefetch = yes
 endif
 

+endif
+

 ### ==========================================================================
 ### Section 3. Low-level Configuration
 ### ==========================================================================
 ### ==========================================================================
 ### Section 3. Low-level Configuration
 ### ==========================================================================


@@ -487,6 +512,13 @@ ifeq ($(ssse3),yes)
        endif
 endif
 
        endif
 endif
 

+ifeq ($(sse2),yes)
+       CXXFLAGS += -DUSE_SSE2
+       ifeq ($(comp),$(filter $(comp),gcc clang mingw))
+               CXXFLAGS += -msse2
+       endif
+endif
+

 ifeq ($(mmx),yes)
        CXXFLAGS += -DUSE_MMX
        ifeq ($(comp),$(filter $(comp),gcc clang mingw))
 ifeq ($(mmx),yes)
        CXXFLAGS += -DUSE_MMX
        ifeq ($(comp),$(filter $(comp),gcc clang mingw))


@@ -503,10 +535,6 @@ ifeq ($(neon),yes)
        endif
 endif
 
        endif
 endif
 

-ifeq ($(arch),x86_64)
-       CXXFLAGS += -msse2 -DUSE_SSE2
-endif
-

 ### 3.7 pext
 ifeq ($(pext),yes)
        CXXFLAGS += -DUSE_PEXT
 ### 3.7 pext
 ifeq ($(pext),yes)
        CXXFLAGS += -DUSE_PEXT


@@ -592,9 +620,10 @@ help:
        @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-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                  > x86 64-bit generic"
-       @echo "x86-32                  > x86 32-bit (also enables MMX and SSE)"
-       @echo "x86-32-old              > x86 32-bit fall back for old hardware"
+       @echo "x86-64                  > x86 64-bit generic (with sse2 support)"
+       @echo "x86-32-sse41-popcnt     > x86 32-bit with sse41 and popcnt support"
+       @echo "x86-32-sse2             > x86 32-bit with sse2 support"
+       @echo "x86-32                  > x86 32-bit generic (with mmx and sse support)"

        @echo "ppc-64                  > PPC 64-bit"
        @echo "ppc-32                  > PPC 32-bit"
        @echo "armv7                   > ARMv7 32-bit"
        @echo "ppc-64                  > PPC 64-bit"
        @echo "ppc-32                  > PPC 32-bit"
        @echo "armv7                   > ARMv7 32-bit"


@@ -624,7 +653,7 @@ help:
        @echo "make -j build ARCH=x86-64-ssse3 COMP=clang"
        @echo ""
 ifneq ($(empty_arch), yes)
        @echo "make -j build ARCH=x86-64-ssse3 COMP=clang"
        @echo ""
 ifneq ($(empty_arch), yes)

-       @echo "-------------------------------\n"
+       @echo "-------------------------------"

        @echo "The selected architecture $(ARCH) will enable the following configuration: "
        @$(MAKE) ARCH=$(ARCH) COMP=$(COMP) config-sanity
 endif
        @echo "The selected architecture $(ARCH) will enable the following configuration: "
        @$(MAKE) ARCH=$(ARCH) COMP=$(COMP) config-sanity
 endif


@@ -719,11 +748,13 @@ config-sanity:
        @echo "os: '$(OS)'"
        @echo "prefetch: '$(prefetch)'"
        @echo "popcnt: '$(popcnt)'"
        @echo "os: '$(OS)'"
        @echo "prefetch: '$(prefetch)'"
        @echo "popcnt: '$(popcnt)'"

+       @echo "pext: '$(pext)'"

        @echo "sse: '$(sse)'"
        @echo "sse: '$(sse)'"

+       @echo "mmx: '$(mmx)'"
+       @echo "sse2: '$(sse2)'"

        @echo "ssse3: '$(ssse3)'"
        @echo "sse41: '$(sse41)'"
        @echo "avx2: '$(avx2)'"
        @echo "ssse3: '$(ssse3)'"
        @echo "sse41: '$(sse41)'"
        @echo "avx2: '$(avx2)'"

-       @echo "pext: '$(pext)'"

        @echo "avx512: '$(avx512)'"
        @echo "vnni: '$(vnni)'"
        @echo "neon: '$(neon)'"
        @echo "avx512: '$(avx512)'"
        @echo "vnni: '$(vnni)'"
        @echo "neon: '$(neon)'"


@@ -744,11 +775,13 @@ config-sanity:
        @test "$(bits)" = "32" || test "$(bits)" = "64"
        @test "$(prefetch)" = "yes" || test "$(prefetch)" = "no"
        @test "$(popcnt)" = "yes" || test "$(popcnt)" = "no"
        @test "$(bits)" = "32" || test "$(bits)" = "64"
        @test "$(prefetch)" = "yes" || test "$(prefetch)" = "no"
        @test "$(popcnt)" = "yes" || test "$(popcnt)" = "no"

+       @test "$(pext)" = "yes" || test "$(pext)" = "no"

        @test "$(sse)" = "yes" || test "$(sse)" = "no"
        @test "$(sse)" = "yes" || test "$(sse)" = "no"

+       @test "$(mmx)" = "yes" || test "$(mmx)" = "no"
+       @test "$(sse2)" = "yes" || test "$(sse2)" = "no"

        @test "$(ssse3)" = "yes" || test "$(ssse3)" = "no"
        @test "$(sse41)" = "yes" || test "$(sse41)" = "no"
        @test "$(avx2)" = "yes" || test "$(avx2)" = "no"
        @test "$(ssse3)" = "yes" || test "$(ssse3)" = "no"
        @test "$(sse41)" = "yes" || test "$(sse41)" = "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 "$(avx512)" = "yes" || test "$(avx512)" = "no"
        @test "$(vnni)" = "yes" || test "$(vnni)" = "no"
        @test "$(neon)" = "yes" || test "$(neon)" = "no"