quantize avx: copy implementation to intrinsics

Multimedia / WebM - Johann [google.com] - 23 August 2017 12:19 EDT

Adds an early exit based on ptest. Slightly slower than ssse3 in the full case because of the extra check, but potentially faster if lots of rows can be skipped.

Very close in speed to the assembly.

Can run in 32 bit, unlike the assembly. Allows reworking the function prototype to use structs.

7c27872 quantize avx: copy implementation to intrinsics
test/vp9_quantize_test.cc | 7 +-
vpx_dsp/vpx_dsp.mk | 1 +
vpx_dsp/vpx_dsp_rtcd_defs.pl | 2 +-
vpx_dsp/x86/quantize_avx.c | 202 ++++++++++++++++++++++++++++++++++++
vpx_dsp/x86/quantize_avx_x86_64.asm | 1 -
5 files changed, 208 insertions(+), 5 deletions(-)

Upstream: git.chromium.org


  • Share