Commit e4363d06 authored by Adrien Maglo's avatar Adrien Maglo Committed by Thomas Daede

motion estimation: fix the 128x64 AVX2 SAD version

Use the paddd instruction instead of paddw as the results are on 32 bits.
This allows to pass the tests.
parent 68d8f7aa
......@@ -84,7 +84,7 @@ cglobal sad%1x%2_avg, 5, ARCH_X86_64 + %3, 6, src, src_stride, \
jg .loop
vextracti128 xm1, m0, 1
paddw xm0, xm1
paddd xm0, xm1
movhlps xm1, xm0
paddd xm0, xm1
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment