Descriptiontest for unaligned vs aligned for CopyRow_SSE2
improves performance on older CPUs where movdqa is faster.
TBR=harryjin@google.com
On N270 Atom
Was ARGBToARGB_Opt (8424 ms)
Now ARGBToARGB_Opt (8050 ms)
BUG=libyuv:492
Committed: https://chromium.googlesource.com/libyuv/libyuv/+/0815568a502c509970cd1177ed6f908305adcaa0
Patch Set 1 #
Messages
Total messages: 6 (3 generated)
|