There's a bunch of unexpected passes (probably "less fuzzy than the minimum bound") and a bunch of fails which can very reasonably just be fuzzed away.
Created attachment 8941575 [details] fuzzier.txt Here's the ones I indentified as clearly fine to bump the fuzz on:
Created attachment 8941577 [details] passes.txt Here's the ones I identified as unexpectedly and reasonably passing. Some are still a bit fuzzy and just need the lower-bound dropped.
Fuzziness values can be extracted from https://treeherder.mozilla.org/#/jobs?repo=try&revision=1793dfceeded3fba0239db7cfee7f4419911360c&selectedJob=155320960 for whoever does this (probably me)
Oh! One thing to keep in mind is that my windows try pushes (such as the one above) include a tweak to the reftest harness that effectively does fuzzy-if(webrender,2,infinity) on all the tests. See patch at . I did this because otherwise there a lot more failures. However for actually getting the reftests greened up we will want to not do this hack. This hack is also responsible for some of the unexpected-pass results that show up.  https://hg.mozilla.org/try/rev/28b2e036e822ed0f3e30e96f7c8854aae6e30a5c
Assignee: nobody → a.beingessner
Priority: -- → P1
You need to log in before you can comment on or make changes to this bug.