]> www.infradead.org Git - users/dwmw2/qemu.git/commit
AVX512 support for xbzrle_encode_buffer
authorling xu <ling1.xu@intel.com>
Wed, 16 Nov 2022 15:29:22 +0000 (23:29 +0800)
committerJuan Quintela <quintela@redhat.com>
Thu, 9 Feb 2023 20:14:43 +0000 (21:14 +0100)
commitaf25ac3ad07a9256d359cb24601b13e9109d8a8b
tree1f69bcb15d3f1d18bbd39cc2c01b96708e22aeee
parent2aadffb0bd38730a62bc81649300130994b86aee
AVX512 support for xbzrle_encode_buffer

This commit is the same with [PATCH v6 1/2], and provides avx512 support for xbzrle_encode_buffer
function to accelerate xbzrle encoding speed. Runtime check of avx512
support and benchmark for this feature are added. Compared with C
version of xbzrle_encode_buffer function, avx512 version can achieve
50%-70% performance improvement on benchmarking. In addition, if dirty
data is randomly located in 4K page, the avx512 version can achieve
almost 140% performance gain.

Signed-off-by: ling xu <ling1.xu@intel.com>
Co-authored-by: Zhou Zhao <zhou.zhao@intel.com>
Co-authored-by: Jun Jin <jun.i.jin@intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
meson.build
meson_options.txt
migration/ram.c
migration/xbzrle.c
migration/xbzrle.h
scripts/meson-buildoptions.sh