Closed Bug 1137104 Opened 9 years ago Closed 8 years ago

16% regression on shumway-splay on February 26th, 2015

Categories

(Firefox Graveyard :: Shumway, defect)

Product:
Firefox Graveyard
Component:
Shumway
Platform:
x86_64
Linux
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
RESOLVED INCOMPLETE

People

(Reporter: h4writer, Assigned: till)

References

Details

(Keywords: perf) 

AreWeFastYet is showing a regression on Shumway splay:
http://arewefastyet.com/#machine=28&view=single&suite=shumway&subtest=splay&start=1424184636&end=1424928007

Mozilla: 16.4% regression
Chrome: 30% regression

Local testing showed this is a shumway issue. Bisecting points to (as first bad revision):
https://github.com/mozilla/shumway/commit/02e1c500298fe375ea18bdcbd0640bd5a5ddfe27
Flags: needinfo?(till)
Some more bisecting shows this to be caused by https://github.com/mozilla/shumway/commit/18e7617bf4e55c96b496116fc5ed41ae363ef7eb. Which is annoying, because that'll be hard to debug and probably even harder to fix :(
Flags: needinfo?(till)
This is caused by a change I made to how our builtins are initialized. Will see how to fix.
Assignee: nobody → till
Till says this not really actionable. He wants to test with new compiler.
Blocks: shumway-1.0
Blocks: shumway-m5
No longer blocks: shumway-1.0
Keywords: perf
Product: Firefox → Firefox Graveyard
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.