this is the right changeset, and it seems to hold stable as I did some retriggers. here is a graph: http://graphs.mozilla.org/graph.html#tests=[[297,131,37]]&sel=1397501445000,1397674245000&displayrange=7&datatype=running on datazilla you can see the individual tests which are failing: https://datazilla.mozilla.org/?start=1397072986&stop=1397677786&product=Firefox&repository=Mozilla-Inbound-Non-PGO&os=win&os_version=5.1.2600&test=tcanvasmark&project=talos Here is view on tbpl: https://tbpl.mozilla.org/?tree=Mozilla-Inbound&fromchange=4121b2865a13&tochange=f1660c847ca2&jobname=Windows%20XP%2032-bit%20mozilla-inbound%20talos%20chromez here is a bit more about canvasmark: https://wiki.mozilla.org/Buildbot/Talos/Tests#CanvasMark This seems like a large regression.

Matt, could you comment on why this might be the case? It looks like :jwatt will be afk for the next week- would like to understand this prior to the uplift in 1.5 weeks.

We filed bug 996929 for this already, hopefully it'll be fixed today.

thanks for being on top of this before I was, I will resolve this as a dup of that.