We identified an issue with Chrome 55 that has introduced variability in some SpeedCurve results starting Dec 9 2016. 

Chrome 55 has a process that consumes high amounts of CPU for 1-2 seconds during startup. When this happens in the wild, it usually only happens once per browser session so isn't very noticeable. However, with WebPageTest (the technology underlying SpeedCurve), because the browser is restarted for every test, the problem occurs more frequently. 

We've verified that the problem is resolved in Chrome 56 but that won't be stable for a few weeks, so we've rolled back to Chrome 54 and confirmed that the problem is resolved. Please note that this issue affects Chrome and emulated devices but does not occur in the other SpeedCurve browsers including Internet Explorer and Firefox.

To track SpeedCurve browser updates please refer to our Product Updates page. 

Did this answer your question?