What Pebas suggests I guess makes it so the SIMD code isn't "hit in the wrong way" as to cause a bug. I would try my hand at it, but it's very hard to find a bug in someone's code by limiting it down when you have no experience with the PulseAudio code base or with the x86 variant of assembly. Is two days of constant work really worth it when someone out there probably knows the fix anyway?
What Pebas suggests I guess makes it so the SIMD code isn't "hit in the wrong way" as to cause a bug. I would try my hand at it, but it's very hard to find a bug in someone's code by limiting it down when you have no experience with the PulseAudio code base or with the x86 variant of assembly. Is two days of constant work really worth it when someone out there probably knows the fix anyway?