Closed Bug 1438495 Opened 6 years ago Closed 6 years ago

4.48% Strings PerfCompressWhitespace (osx-10-10) regression on push 5f7ca17beb3d (Thu Feb 8 2018)

Categories

(Firefox :: Untriaged, defect)

Unspecified
macOS
defect
Not set
normal

Tracking

()

RESOLVED INVALID

People

(Reporter: igoldan, Unassigned)

References

Details

(Keywords: perf, regression)

We have detected a platform microbenchmarks regression from push:

https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=583dac82f2b5c76394515da66173cb76f285bf66&tochange=5f7ca17beb3dcaaa48c4d6339c6e7058e6ffc07d

As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

  4%  Strings PerfCompressWhitespace osx-10-10 opt      249,533.08 -> 260,716.50


You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=11494

On the page above you can see an alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the jobs in a pushlog format.

To learn more about the regressing test(s), please see: https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Automated_Performance_Testing_and_Sheriffing/Platform_Microbenchmarks
::mak I am not too sure bug 1433305 caused this. The tests are very noisy, but clearly show that a regression got in a week ago, according to the minimum values. Can you confirm the connection between the perf regression and your bug?
Flags: needinfo?(mak77)
I'm sorry, but it's not possible for that patch to cause any perf regression, it's removing one API that is only used by xpcshell tests.
And it's even less likely for PerfCompressWhitespace.
Flags: needinfo?(mak77)
No longer blocks: 1433305
As a rule, these functions are so low level, that only the code in xpcom/string files could possibly make a difference.  Was there anything in the past week or so that touched any of those files?
(In reply to Milan Sreckovic [:milan] from comment #3)
> As a rule, these functions are so low level, that only the code in
> xpcom/string files could possibly make a difference.  Was there anything in
> the past week or so that touched any of those files?

Yes, there have been multiple changes. You can see them by typing |hg log xpcom/string --date 2018|.
The closest one would be bug 1435924, but it landed after the regression appeared.

The bugs I noticed are these ones (date reported by Pulsebot):
Feb 06 - bug 1435924
Feb 02 - bug 1434689
Jan 18 - bug 1431449
Jan 18 - bug 1431261
> The bugs I noticed are these ones (date reported by Pulsebot):
> Feb 06 - bug 1435924
> Feb 02 - bug 1434689
> Jan 18 - bug 1431449
> Jan 18 - bug 1431261

I don't think any of these is the cause.
:baku Do you think bug 1436692 has anything to do with this performance regression?
Flags: needinfo?(amarchesini)
(In reply to Ionuț Goldan [:igoldan], Performance Sheriffing from comment #6)
> :baku Do you think bug 1436692 has anything to do with this performance
> regression?

Bug 1436692 exposes an internal value in PerformanceObserver. It doesn't touch any string implementation.
I don't think this is the cause.
Flags: needinfo?(amarchesini)
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.