search for: essick

...ated code with -O2 between the two versions. (Literally none for x86_64, and only what seem to be some non-signficiant offset changes for i386). So if this improves some other cases (notably, it should avoid problems with -fsanitize=address) I think this is okay. On Aug 18, 2017, at 5:25 PM, Ray Essick <essick at google.com<mailto:essick at google.com>> wrote: Jonathan, Here's the code difference we see with the recent change -- what amounts to reverting your change from a couple years back. It doesn't look like we're getting superfluous instructions from clang now. the...

We see the MOVQ instruction but this patch deliberately uses it rather than MOVQDA (load 128-bits aligned). We were seeing that with the trace below, the final invocation is not 128-bit aligned but MOVQDA insists on it (the calling function was pitch_sse4_1.c:90, in the 4-way N - i >= 4 loop). 07-31 11:00:13.469 210 2540 <(469)%20210-2540> D opus_sse1: RBE celt_inner_prod_sse4_1: x

Jonathan, Here's the code difference we see with the recent change -- what amounts to reverting your change from a couple years back. It doesn't look like we're getting superfluous instructions from clang now. the bad behavior for us was the alignment exception on the movdqa instructions when the input data wasn't 128-bit aligned. We had to change something because the code as-is