New issue
Advanced search Search tips

Issue 903982 link

Starred by 1 user
Status: Untriaged
Owner: ----
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug
Build-Toolchain



Sign in to add a comment

Different benchmark run status found in a crosperf run

Project Member Reported by tcwang@chromium.org, Nov 9 
Recently, I found a crosperf report that has disparity of a benchmark run status for benchmark page_cycler_v2.typical_25. In the performance section, it shows the benchmark failed on all the iterations (3 times). So there's no performance numbers. But in the "Benchmark Run Status" section, it shows the benchmark succeeded. The link to the report is attached here. The report is generated on veyron_minnie for llvm_next.

https://groups.google.com/a/google.com/d/msg/c-compiler-chrome/yVMNmr6MztM/aOuiOH0qAgAJ

Comment 1 by cmt...@chromium.org, Nov 9 

My experience with this in the past has been that this happens when the benchmarks are taking too long to run, and the page_cycler tests time out.  Often a cause for slow benchmarks are either zombie processes or  disks getting full.

Ting-yuan wrote a script on chrotomation2 (in mobiletc-prebuild) to help find old processes that need to be killed: 

~/laszio/ps-old
►

Sign in to add a comment