New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 774195 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Last visit > 30 days ago
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 1
Type: Bug



Sign in to add a comment

large performance degradation observed by crosbolt in R62 for speedometer.

Project Member Reported by llozano@chromium.org, Oct 12 2017

Issue description

speedometer benchmark has changed as explained in here: https://chromeperf.appspot.com/report?sid=85d1d2a4e09aa185c94675159d5dfb0c12918571003846e4b5083975161e6fd8

I don't understand what version of speedometer is Chrome OS using. So, one of the owners of crosbolt needs to decide which version we need to track. This is most probably related to the degradation that crosbolt sees in 62:

https://cros-goldeneye.corp.google.com/chromeos/console/listCrosbolt?graphSKU=samus_intel_broadwell_i5_2.2g_4Gb&graphTest=speedometer%2FTotal.Total

I am not sure if there is anything to fix here but the graph shown by crosbolt for R62 is worry some because it shows a very large degradation. 

Also, Which version of speedometer are we going to track from now on?

I will mark this as a P1 (instead of as a P0) as I believe there is no real degradation and it is just an issue with the benchmark. 

(There is a real degradation being tracked in  issue 773561 ).
 
Cc: laszio@chromium.org llozano@chromium.org cros-perf-detectives@google.com
Hi Luis,

Is it true that speedometer benchmark has changed? According to this graph, the performance is back to normal in R63, so it might be that some regression sneaked into R62 but was fixed in R63:

https://chromeperf.appspot.com/report?sid=145ee81c2a5c1cfd00510c45742f6c0f10e41b6997d4a74d3cdf4d5e40bc0898

We are tracking it in crbug/773938


ben, I really don't know. 
I suspect such huge performance variation is due to some change in the benchmark.

I posted the wrong link before, The possible explanation I found for this is: 
(actually, someone from Intel found this)
https://bugs.chromium.org/p/chromium/issues/detail?id=756990

I would start there and try to match the timeline.

And we need to decide which version of speedometer we are going to track.

feel free to close this issue as duplicate if appropriate.

Sign in to add a comment