projects / mesa.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 2a33510) | patch
Add an accelerated version of F_TO_I for x86_64
authorJason Ekstrand <jason.ekstrand@intel.com>
Mon, 21 Jul 2014 23:46:39 +0000 (16:46 -0700)
committerJason Ekstrand <jason.ekstrand@intel.com>
Thu, 24 Jul 2014 19:44:56 +0000 (12:44 -0700)
commit989d2e370993c87d1bbda4950657bfcc5b0a58dd
tree1b9a048a63e291fb68d7c1a2bf00e08d69d7e672tree
parent2a33510f1649f2ef5c5b2d693aa89ef0efc5dcfbcommit | diff
Add an accelerated version of F_TO_I for x86_64

According to a quick micro-benchmark, this new version is 20% faster on my
Haswell laptop.

v2: Removed the XXX note about x86_64 from the comment
v3: Use an intrinsic instead of an __asm__ block.  This should give us MSVC
    support for free.
v4: Enable it for all x86_64 builds, not just with USE_X86_64_ASM

Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
src/mesa/main/imports.h diff | blob | history