There don't appear to be relevant GPU related changes at or near the failing build, but there was a V8 update which could impact the performance of this test: https://codereview.chromium.org/2756513005

It flakes on GPU waterfall starting from
https://build.chromium.org/p/chromium.gpu/builders/Win7%20Debug%20%28NVIDIA%29/builds/50250

I fails consistently on GPU.FYI waterfall starting from
https://build.chromium.org/p/chromium.gpu.fyi/builders/Win10%20Debug%20%28NVIDIA%29/builds/702
https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28AMD%29/builds/1406
https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28NVIDIA%29/builds/20235

They all have https://codereview.chromium.org/2756513005 in them.

Considering that these are all Debug bots, I'm not sure if a minor performance regressions is very relevant. Ken, should we just increase the timeouts?

I think this has actually been fixed by the next v8-autoroll, https://codereview.chromium.org/2754273002.

Looks like https://codereview.chromium.org/2758503002 was the culprit.

That saying, the timeout is currently 180 sec, and I see the test passing in 160 sec, so we are pretty close.
Good thing about it is that we can easily catch v8 regressions.

If we're on the edge here then maybe we should increase the timeouts. Patches welcome.

For the time being let's close this as WontFix though.