In case people is interested this is the patch to enable hardware POPCNT support with gcc in Stockfish 1.6.2. Many thanks to Laurent Desnogues for testing.
mcostalba wrote:In case people is interested this is the patch to enable hardware POPCNT support with gcc in Stockfish 1.6.2. Many thanks to Laurent Desnogues for testing.
Date: Sat, 30 Jan 2010 19:18:27 +0100
Subject: [PATCH] Add hardware POPCNT support for gcc
With new target 'make gcc-popcnt' it is now
possible to compile with enabled hardware POPCNT
support also with gcc. Until now was possible only
for Intel and MSVC compilers.
When this instruction is supported by CPU, for instance
on Intel i7 or i5 family, produced binary is a bit faster.
---
src/Makefile | 8 ++++++++
src/bitcount.h | 24 ++++++++++++++++++++++++
2 files changed, 32 insertions(+), 0 deletions(-)
diff --git a/src/Makefile b/src/Makefile
index 1388f46..3963e55 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -80,6 +80,7 @@ help:
@echo "Makefile options:"
@echo ""
@echo "make > Default: Compiler = g++"
+ @echo "make gcc-popcnt > Compiler = g++ + popcnt-support"
@echo "make icc > Compiler = icpc"
@echo "make icc-profile > Compiler = icpc + automatic pgo-build"
@echo "make icc-profile-popcnt > Compiler = icpc + automatic pgo-build + popcnt-support"
@@ -108,6 +109,13 @@ gcc:
CXXFLAGS="$(GCCFLAGS)" \
all
+gcc-popcnt:
+ $(MAKE) \
+ CXX='g++' \
+ CXXFLAGS="$(GCCFLAGS) -DUSE_POPCNT" \
I think that if you want to force usage of the popcnt instruction with gcc you will have to add -mpopcnt. OTOH if you're compiling on a machine that has support for this instruction that flag is useless.
ldesnogu wrote:
I think that if you want to force usage of the popcnt instruction with gcc you will have to add -mpopcnt. OTOH if you're compiling on a machine that has support for this instruction that flag is useless.
I thought you had already tested the above patch and reported that it works. Am I missing something ?
Regarding -mpopcnt I think is related to use gcc builtins, in this case I have injected popcnt instruction directly by an __asm__ statement.
Also -DUSE_POPCNT is not used by gcc but is passed as is as a define, so that the #elif in the patch is enabled.
Oh, you're right, I kind of mixed my patch, which uses the builtin, and yours. I guess I should refrain from answering posts too early on a Sunday morning