]> granicus.if.org Git - libjpeg-turbo/commit
AVX2: Avoid expensive AVX-SSE transitions
authorDRC <information@libjpeg-turbo.org>
Sat, 9 Jul 2016 01:10:24 +0000 (20:10 -0500)
committerDRC <information@libjpeg-turbo.org>
Sat, 9 Jul 2016 01:10:24 +0000 (20:10 -0500)
commita7c2f979390706e3b834c80943ad490f34bbef16
tree27f051d95972814e52391ace633c9801c6e13b24
parente06ccbe3f703929353a4eb15dc472961850f31de
AVX2: Avoid expensive AVX-SSE transitions

Refer to
https://software.intel.com/sites/default/files/m/d/4/1/d/8/11MC12_Avoiding_2BAVX-SSE_2BTransition_2BPenalties_2Brh_2Bfinal.pdf
for more information.  This eliminates all AVX-SSE transitions detected
with the Intel SDE tool.
12 files changed:
simd/jccolext-avx2-64.asm
simd/jccolext-avx2.asm
simd/jcgryext-avx2-64.asm
simd/jcgryext-avx2.asm
simd/jcsample-avx2-64.asm
simd/jcsample-avx2.asm
simd/jdcolext-avx2-64.asm
simd/jdcolext-avx2.asm
simd/jdmrgext-avx2-64.asm
simd/jdmrgext-avx2.asm
simd/jdsample-avx2-64.asm
simd/jdsample-avx2.asm