Closed Bug 1417802 Opened 7 years ago Closed 6 years ago

10.72 - 17.01% Strings PerfIsASCIIHundred / Strings PerfIsUTF8Fifteen (osx-10-10) regression on push db45eedf71b32bfb672d08d6cb087d6be1d4c466 (Tue Nov 7 2017)

Categories

(Core :: CSS Parsing and Computation, defect)

Unspecified
macOS
defect
Not set
normal

Tracking

()

RESOLVED DUPLICATE of bug 1415514
Tracking Status
firefox59 --- wontfix

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?changeset=db45eedf71b32bfb672d08d6cb087d6be1d4c466

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

Regressions:

 17%  Strings PerfIsUTF8Fifteen osx-10-10 opt      1,728.04 -> 2,022.00
 11%  Strings PerfIsASCIIHundred osx-10-10 opt     1,924.29 -> 2,130.50

Improvements:

  7%  Strings PerfIsUTF8Example3 osx-10-10 opt      8,183.50 -> 7,618.75


You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=10386

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
Component: Untriaged → CSS Parsing and Computation
Product: Firefox → Core
:xidorn This is a strange regression. Initially I ignored it, but it showed up on Beta also, so it's something real about it.
Initial push of bug 1417802 didn't cause any performance change. The backout however, caused the regressions you see above: on landing branch, then on inbound after merge, then on beta after merge.
The second push you then made didn't fix this regression.

Do you know why this happened? Can we resolve this easily?
Flags: needinfo?(xidorn+moz)
I don't quite believe that change can regress this test. That is purely a Servo side change, and this microbenchmark seems to be testing some Gecko underlying stuff.

I have no clue on what happens there. Maybe it changes how the test is optimized in link time, and bug 1396208 makes such optimization no longer applied, and consequently the re-landing of bug 1374178 (after bug 1396208) doesn't change anything?
Flags: needinfo?(xidorn+moz)
This smells like a dupe of bug 1415514. Let's ask Henri when he gets back.
Flags: needinfo?(hsivonen)
(In reply to Jet Villegas (:jet) from comment #3)
> This smells like a dupe of bug 1415514. Let's ask Henri when he gets back.

Looks a lot like a duplicate of that one with the changeset in comment 0 being misattributed.
Flags: needinfo?(hsivonen)
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → DUPLICATE
Thanks for your prompt feedback!
You need to log in before you can comment on or make changes to this bug.