> I can only repeat the point I made before - ASM is not about safety, > its about speed. > If you can re-write it to not be a single cycle slower and not have > this particular behavior, great. > > In the meantime, can we apply this patch and get the crash done with? > > FWIW, all other callers of vector_fmul_scalar don't seem to have a > chance of the value being zero.
+1 _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel