We've made lots of measurable, impactful perf improvements on Windows and Linux over the last few months. We are now noticeably superior perf-wise to Chrome across most tests as well as on the overall (comprehensive) web audio benchmark for Windows and Linux. On Mobile (Android) we're about even and on Mac, we're about even on MacBook Pros and actually measuring worse than Chrome on old Macs (like the Minis used in the benchmark tests). Before we declare "victory" with our perf work, we want at a minimum to understand why we haven't see the same perf improvements on Mac. Preferably we'll make the changes needed to improve our perf on Mac such that we are better across all desktop platforms. At the same time, it's worth investigating if we can get noticeably better performance on Android as well. We may turn this into a meta bug, depending on what we find.

I noticed this morning that libav is not being used on os x [1]. That might explain the poor performance on the convolution reverb benchmark on mac relative to linux and windows. [1] https://dxr.mozilla.org/mozilla-central/source/configure.in#6196

I expect most Mac users to be running the 64-bit binary, where *:x86_64 should match.

(In reply to Karl Tomlinson (back Dec 21 ni?:karlt) from comment #2) > I expect most Mac users to be running the 64-bit binary, where *:x86_64 > should match. Yes, my mistake. I double checked the old mac mini we're running the mozbench benchmarks on and even it is 64 bit.

Mass change P1->P2 to align with new Mozilla triage process

Moving to p3 because no activity for at least 1 year(s). See https://github.com/mozilla/bug-handling/blob/master/policy/triage-bugzilla.md#how-do-you-triage for more information