It looks like Pinpoint is identifying multiple issues for performance regressions when some are minimally involved or not involved.

For example,  issue 804316  identifies a memory regression and implicates multiple CLs including my CL https://chromium-review.googlesource.com/871497, but I'm pretty sure this CL is not affecting memory.  

Maybe there's a bug in Pinpoint with multiple CLs?  At the very least, the handling of performance issues with multiple causes seems like it could be improved.  It looks like the performance issue ends up being assigned to the owner of the last cause-CL, but it would be better to assign it to the owner of the cause-CL with highest impact.  Also wouldn't it be better to issue multiple performance bugs, one for each cause?  Until till then, maybe add instructions indicating that all cause-CLs need to be addressed (assuming they do all cause a negative impact)?

 Issue 804319  is another example of how this issue can cause confusion and difficulty addressing the underlying cause(s).