📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/179efdd8140000

See https://crbug.com/911052#c8. The regression is clear visually, my guess is high variance makes pinpoint think the regression is not conclusive. Perhaps we could tweak parameters on-demand in such cases? Failed bisects leave sheriffs with almost no options except guesswork or manual local bisects so it'd be great to avoid them as much as possible.