[PATCH] x86_64: Silvermont optimized memcpy, mempcpy, memmove, memcpy_chk, mempcpy_chk, memmove_chk

[Andrew Senkevich <andrew dot n dot senkevich at gmail dot com>]

Hi,

this patch contains Silvermont optimized versions of memcpy, mempcpy,
memmove, memcpy_chk, mempcpy_chk and memmove_chk.

Performance benefit introduced by this patch is about 10% for memcpy
and memmove on big lengths and about 20% for memcpy on short lengths.
Measurements results and patch attached.

Change log:

2014-08-14  Andrew Senkevich  <andrew.n.senkevich@gmail.com>

        * sysdeps/x86_64/multiarch/Makefile: Added new files to build.
        * sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Added testing of new
        functions.
        * sysdeps/x86_64/multiarch/init-arch.h (HAS_SLOW_SSE4_2): Define.
        * sysdeps/x86_64/multiarch/memcpy-sse2-unaligned-slm.S: New file,
        contains Silvermont optimized implementation with sse2
unaligned loads/stores.
        * sysdeps/x86_64/multiarch/memmove-sse2-unaligned-slm.S: Likewise.
        * sysdeps/x86_64/multiarch/mempcpy-sse2-unaligned-slm.S: Likewise.
        * sysdeps/x86_64/multiarch/memmove.c: Added support for new function.
        * sysdeps/x86_64/multiarch/memcpy.S: Likewise.
        * sysdeps/x86_64/multiarch/memcpy_chk.S: Likewise.
        * sysdeps/x86_64/multiarch/mempcpy.S: Likewise.
        * sysdeps/x86_64/multiarch/mempcpy_chk.S: Likewise.


--
Best wishes,
Andrew

Attachment: memcpy_slm_x86_64.tar.bz2
Description: BZip2 compressed data

Attachment: memmove_slm_x86_64.tar.bz2
Description: BZip2 compressed data

Attachment: memcpy_mempcpy_memmove_with_chk_sse2_unaligned_slm_x86_64.patch
Description: Binary data