JS demo is 2.6x slower in Nightly compared to Chrome
Categories
(Core :: JavaScript Engine, task, P3)
Tracking
()
People
(Reporter: mayankleoboy1, Unassigned)
References
(Blocks 1 open bug, )
Details
Go to https://dgtized.github.io/shimmers/#/sketches/wave-function-collapse?seed=1151015197
Click on the "Cycle Mode" button once to change it to "cells"
Click on "Solve" button'
Nightly: https://share.firefox.dev/3Sk8Tpw (65s)
Chrome: https://share.firefox.dev/47wtA69 (28s)
|
||
Updated•2 years ago
|
|
||
Comment 1•2 years ago
|
||
Interesting that we spend 2.3% of time in the 'isForwarded' call. Aside from that, profile is quite flat -- haven't looked deeply though during triage.
|
||
Comment 2•2 years ago
|
||
We've got a high nursery promotion rate here. I wonder if there we are missing a pretenuring opportunity.
|
Reporter | |
Comment 3•2 years ago
•
|
||
(In reply to Jon Coppeard (:jonco) (PTO until 4th March) from comment #2)
We've got a high nursery promotion rate here. I wonder if there we are missing a pretenuring opportunity.
Is that the "Nursery allocations since last minor GC" metric that appears when you hover mouse over the nursery markers? What is considered as "high nursery promotion rate" ?
Are such cases worthwhile to file bugs (where the rate is high and optionally we are slower than the competition)?
|
|
||
Comment 4•2 years ago
|
||
(In reply to Mayank Bansal from comment #3)
Is that the "Nursery allocations since last minor GC" metric
It's the percentage that appears at the end of "Bytes tenured". We aim to keep this around 2% but it depends on workload and other factors (like how large we can make the nursery).
|
|
Reporter | |
Comment 5•2 years ago
|
||
Nightly: https://share.firefox.dev/3z8hKEd (56s)
Chrome: https://share.firefox.dev/47tEdrG (23s)
|
|
Reporter | |
Comment 6•1 year ago
|
||
Nightly: https://share.firefox.dev/3SEuR68 (65s)
|
||
Comment 7•1 year ago
|
||
Nothing in particular stands out to me from the profile. The tenuring rate hasn't improved much, but it's probably worth pointing out that we're 2.5x slower and only spending 17% of our time in GC, so some of the gap must be outside of GC.
Description
•