I was using a recent nightly (2011-11-16, Win7, x86) but I don't think this is a new problem as it happens from time to time. I attached a profiler (http://lukestackwalker.sourceforge.net/) and it showed the following sampling: ================= Thread 0x1668 [445 samples, 61 funcs, 41.73s CPU time] Function Samples gfxASurce::CheckSurfaceSize 33.9% gfxFontGroup::ForEachFontInternal 25.6% js_RemoveRoot 7.0% NtWaitForSingleObject 2.7% TrailUpVec 2.5% gfxASurface::Release 1.8% NS_CycleCollectorSuspect2_P 1.6% js::TypedArray::obj_setSpecialAttributes 1.3% JSScript::jitDataSize 1.3% mozilla::layers::LayerManager::Mutated 1.1% Everything else is < 1%. I saved the profile session so I can get more info if needed ================== "Heavy" tabs that were open were Gmail, Twitter, IRCCloud and Grooveshark The profiler also provides some call graphs, I'll attach them in a moment. I have HW accel disabled due to my current driver version being blocked.

Ok, don't know how useful this is, but here are the call graphs for the top 6 sampled functions

See also Bug 703455, a similar prob that I can reproduce.

Were you doing anything in particular when this happened? What extensions do you have? What does about:memory look like? Marking as [Snappy:P2] as it looks like there could be a bottleneck here that should be investigated.