diff options
| author | Ling Ma <ling.ml@alibaba-inc.com> | 2014-07-14 00:02:52 -0400 |
|---|---|---|
| committer | H.J. Lu <hjl.tools@gmail.com> | 2014-07-30 08:02:35 -0700 |
| commit | 05f3633da4f9df870d04dd77336e793746e57ed4 (patch) | |
| tree | bd980df4702ffd9ebfa9cb4dac8823a659f0c430 /sysdeps/x86_64/multiarch/memmove_chk.c | |
| parent | a53fbd8e6cd2f69bdfa3431d616a5f332aea6664 (diff) | |
| download | glibc-05f3633da4f9df870d04dd77336e793746e57ed4.tar.gz glibc-05f3633da4f9df870d04dd77336e793746e57ed4.tar.xz glibc-05f3633da4f9df870d04dd77336e793746e57ed4.zip | |
Improve 64bit memcpy performance for Haswell CPU with AVX instruction
In this patch we take advantage of HSW memory bandwidth, manage to reduce miss branch prediction by avoiding using branch instructions and force destination to be aligned with avx instruction. The CPU2006 403.gcc benchmark indicates this patch improves performance from 2% to 10%.
Diffstat (limited to 'sysdeps/x86_64/multiarch/memmove_chk.c')
| -rw-r--r-- | sysdeps/x86_64/multiarch/memmove_chk.c | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/sysdeps/x86_64/multiarch/memmove_chk.c b/sysdeps/x86_64/multiarch/memmove_chk.c index cb1acb6598..44344f2820 100644 --- a/sysdeps/x86_64/multiarch/memmove_chk.c +++ b/sysdeps/x86_64/multiarch/memmove_chk.c @@ -25,11 +25,13 @@ extern __typeof (__memmove_chk) __memmove_chk_sse2 attribute_hidden; extern __typeof (__memmove_chk) __memmove_chk_ssse3 attribute_hidden; extern __typeof (__memmove_chk) __memmove_chk_ssse3_back attribute_hidden; +extern __typeof (__memmove_chk) __memmove_chk_avx_unaligned attribute_hidden; #include "debug/memmove_chk.c" libc_ifunc (__memmove_chk, - HAS_SSSE3 + HAS_AVX ? __memmove_chk_avx_unaligned : + (HAS_SSSE3 ? (HAS_FAST_COPY_BACKWARD ? __memmove_chk_ssse3_back : __memmove_chk_ssse3) - : __memmove_chk_sse2); + : __memmove_chk_sse2)); |