ARM lsb/msb assembly

Implement lsb/msb using armv7 assembly instructions.
msb is the easiest one, using a gcc intrinsic that generates
code using the ARM's clz instruction. lsb is also using this
clz instruction, but with the help of ARM's 'rbit' (bit
reversing) instruction. This leads to a >2% speed gain.

I also renamed 'arm-32' to the more meaningfull 'armv7' in the Makefile

No functional change.

diff --git a/src/Makefile b/src/Makefile
index c73b94f390a177bd1dd0b65c9d43f39cb5b42e32..17e61773d19615b6289e085881ddb927ac44c9c9 100644 (file)
--- a/src/Makefile
+++ b/src/Makefile
@@ -126,12 +126,12 @@ ifeq ($(ARCH),x86-32-old)
 endif
 
 #arm section
 endif
 
 #arm section

-ifeq ($(ARCH),arm-32)
-       arch = arm-32
+ifeq ($(ARCH),armv7)
+       arch = armv7

        os = any
        bits = 32
        prefetch = yes
        os = any
        bits = 32
        prefetch = yes

-       bsfq = no
+       bsfq = yes

        popcnt = no
 endif
 
        popcnt = no
 endif
 


@@ -276,7 +276,7 @@ ifeq ($(optimize),yes)
                        endif
                endif
 
                        endif
                endif
 

-               ifeq ($(arch),arm-32)
+               ifeq ($(arch),armv7)

                        CXXFLAGS += -fno-gcse
                endif
        endif
                        CXXFLAGS += -fno-gcse
                endif
        endif


@@ -315,7 +315,7 @@ endif
 
 ### 3.7 prefetch
 ifeq ($(prefetch),yes)
 
 ### 3.7 prefetch
 ifeq ($(prefetch),yes)

-       ifneq ($(arch),arm-32)
+       ifneq ($(arch),armv7)

                CXXFLAGS += -msse
                DEPENDFLAGS += -msse
        endif
                CXXFLAGS += -msse
                DEPENDFLAGS += -msse
        endif


@@ -376,7 +376,7 @@ help:
        @echo "osx-ppc-32           > PPC-Mac OS X 32 bit"
        @echo "osx-x86-64           > x86-Mac OS X 64 bit"
        @echo "osx-x86-32           > x86-Mac OS X 32 bit"
        @echo "osx-ppc-32           > PPC-Mac OS X 32 bit"
        @echo "osx-x86-64           > x86-Mac OS X 64 bit"
        @echo "osx-x86-32           > x86-Mac OS X 32 bit"

-       @echo "arm-32               > ARM 32 bit"
+       @echo "armv7                > ARMv7 32 bit"

        @echo "general-64           > unspecified 64-bit"
        @echo "general-32           > unspecified 32-bit"
        @echo ""
        @echo "general-64           > unspecified 64-bit"
        @echo "general-32           > unspecified 32-bit"
        @echo ""


@@ -466,7 +466,7 @@ config-sanity:
        @test "$(debug)" = "yes" || test "$(debug)" = "no"
        @test "$(optimize)" = "yes" || test "$(optimize)" = "no"
        @test "$(arch)" = "any" || test "$(arch)" = "x86_64" || test "$(arch)" = "i386" || \
        @test "$(debug)" = "yes" || test "$(debug)" = "no"
        @test "$(optimize)" = "yes" || test "$(optimize)" = "no"
        @test "$(arch)" = "any" || test "$(arch)" = "x86_64" || test "$(arch)" = "i386" || \

-        test "$(arch)" = "ppc64" || test "$(arch)" = "ppc" || test "$(arch)" = "arm-32"
+        test "$(arch)" = "ppc64" || test "$(arch)" = "ppc" || test "$(arch)" = "armv7"

        @test "$(os)" = "any" || test "$(os)" = "osx"
        @test "$(bits)" = "32" || test "$(bits)" = "64"
        @test "$(prefetch)" = "yes" || test "$(prefetch)" = "no"
        @test "$(os)" = "any" || test "$(os)" = "osx"
        @test "$(bits)" = "32" || test "$(bits)" = "64"
        @test "$(prefetch)" = "yes" || test "$(prefetch)" = "no"

diff --git a/src/bitboard.h b/src/bitboard.h
index 4c7c948a8d28d0dfa9e8bc8c08f88471797b2b9d..6aa35874df7d3a244a8be9a919c81e09cc46105e 100644 (file)
--- a/src/bitboard.h
+++ b/src/bitboard.h
@@ -247,6 +247,21 @@ FORCE_INLINE Square msb(Bitboard b) {
   return (Square) index;
 }
 
   return (Square) index;
 }
 

+#  elif defined(__arm__)
+
+FORCE_INLINE int lsb32(uint32_t v) {
+  __asm__("rbit %0, %1" : "=r"(v) : "r"(v));
+  return __builtin_clz(v);
+}
+
+FORCE_INLINE Square msb(Bitboard b) {
+  return (Square) (63 - __builtin_clzll(b));
+}
+
+FORCE_INLINE Square lsb(Bitboard b) {
+  return (Square) (uint32_t(b) ? lsb32(uint32_t(b)) : 32 + lsb32(uint32_t(b >> 32)));
+}
+

 #  else
 
 FORCE_INLINE Square lsb(Bitboard b) { // Assembly code by Heinz van Saanen
 #  else
 
 FORCE_INLINE Square lsb(Bitboard b) { // Assembly code by Heinz van Saanen