70.95 - 61.23% glterrain / glterrain + 1 more (Linux) regression on Tue March 1 2022
Categories
(Core :: Graphics: CanvasWebGL, defect)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox-esr91 | --- | unaffected |
| firefox97 | --- | unaffected |
| firefox98 | --- | unaffected |
| firefox99 | --- | wontfix |
| firefox100 | --- | wontfix |
People
(Reporter: aesanu, Unassigned)
References
(Regression)
Details
(4 keywords)
Perfherder has detected a talos performance regression from push 1edd1e4056e62592fe5f457926e5012ad30003d6. As author of one of the patches included in that push, we need your help to address this regression.
Regressions:
| Ratio | Test | Platform | Options | Absolute values (old vs new) |
|---|---|---|---|---|
| 71% | glterrain | linux1804-64-shippable-qr | e10s fission stylo webrender | 5.74 -> 9.82 |
| 71% | glterrain | linux1804-64-shippable-qr | e10s fission stylo webrender | 5.70 -> 9.74 |
| 61% | glterrain | linux1804-64-shippable-qr | e10s fission stylo webrender-sw | 5.38 -> 8.68 |
Details of the alert can be found in the alert summary, including links to graphs and comparisons for each of the affected tests. Please follow our guide to handling regression bugs and let us know your plans within 3 business days, or the offending patch(es) will be backed out in accordance with our regression policy.
For more information on performance sheriffing please see our FAQ.
|
||
Comment 1•2 years ago
|
||
Set release status flags based on info from the regressing bug 1129492
|
||
Comment 2•2 years ago
|
||
Kelsey, does this look like an expected regression from remote WebGL?
|
||
Comment 3•2 years ago
|
||
== Change summary for alert #33488 (as of Mon, 07 Mar 2022 13:30:05 GMT) ==
Improvements:
| Ratio | Test | Platform | Options | Absolute values (old vs new) |
|---|---|---|---|---|
| 12% | espn fnbpaint | linux1804-64-shippable-qr | cold fission webrender | 507.23 -> 444.96 |
| 8% | espn ContentfulSpeedIndex | linux1804-64-shippable-qr | cold fission webrender | 753.25 -> 691.67 |
| 8% | espn dcf | linux1804-64-shippable-qr | cold fission webrender | 717.06 -> 659.88 |
| 8% | espn fnbpaint | linux1804-64-shippable-qr | fission warm webrender | 159.21 -> 147.04 |
| 7% | espn fnbpaint | linux1804-64-shippable-qr | fission warm webrender | 158.54 -> 147.12 |
| ... | ... | ... | ... | ... |
| 4% | espn dcf | linux1804-64-shippable-qr | fission warm webrender | 259.81 -> 250.29 |
For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=33488
|
|
||
Updated•2 years ago
|
|
|
||
Comment 4•2 years ago
|
||
Set release status flags based on info from the regressing bug 1129492
|
||
Comment 5•2 years ago
|
||
The reports in comment #3 are confusing: in a lot of cases the same test is reported as both a regression and an improvement, and looking at the graphs there doesn't seem to be a significant change. The only exception is the warm tests, which do seem to show actual improvements. I asked about this on Matrix #perftest, but for now I'm going to assume that the actual effect is an improvement or at least neutral overall.
As for glterrain, my understanding is that it's not very representative of real-world WebGL performance, but Kelsey is the authority there.
|
||
Comment 6•2 years ago
|
||
ESPN must use webgl, which suffers on cold-load when we move webgl out-of-process.
That makes sense, and I think that's acceptable.
glterrain is pretty synthetic, and probably just suffers from moving out-of-process. I consider that regression acceptable.
Thanks!
|
||
Comment 7•2 years ago
|
||
(In reply to Jed Davis [:jld] ⟨⏰|UTC-6⟩ ⟦he/him⟧ (away until 04-12) from comment #5)
The reports in comment #3 are confusing: in a lot of cases the same test is reported as both a regression and an improvement, and looking at the graphs there doesn't seem to be a significant change.
:alexandrui can you double check this alert? We shouldn't have the same test both regress and improve for the same push.
|
||
Updated•2 years ago
|
Description
•