380.08 - 6.02% pdfpaint issue15559.pdf / pdfpaint bug931481.pdf + 271 more (Linux, OSX, Windows) regression on Wed March 5 2025
Categories
(Firefox :: PDF Viewer, defect)
Tracking
()
Tracking | Status | |
---|---|---|
firefox-esr128 | --- | unaffected |
firefox136 | --- | unaffected |
firefox137 | --- | unaffected |
firefox138 | --- | fixed |
People
(Reporter: intermittent-bug-filer, Unassigned)
References
(Regression)
Details
(4 keywords)
Perfherder has detected a talos performance regression from push 22c9d3c2280fcb14d95687bed93333f355d0b1ce. 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) |
---|---|---|---|---|
380% | pdfpaint issue15559.pdf | windows11-64-24h2-shippable | e10s fission stylo webrender | 468.93 -> 2,251.24 |
360% | pdfpaint issue12306.pdf | macosx1470-64-shippable | e10s fission stylo webrender | 284.29 -> 1,308.74 |
360% | pdfpaint issue15559.pdf | windows11-64-shippable-qr | e10s fission stylo webrender | 466.50 -> 2,146.37 |
356% | pdfpaint issue15559.pdf | macosx1470-64-shippable | e10s fission stylo webrender-sw | 995.55 -> 4,537.32 |
346% | pdfpaint issue15559.pdf | macosx1470-64-shippable | e10s fission stylo webrender | 1,000.74 -> 4,467.44 |
334% | pdfpaint issue12306.pdf | macosx1470-64-shippable | e10s fission stylo webrender-sw | 291.29 -> 1,264.02 |
300% | pdfpaint issue12306.pdf | windows11-64-shippable-qr | e10s fission stylo webrender | 172.12 -> 688.42 |
299% | pdfpaint issue12306.pdf | linux1804-64-shippable-qr | e10s fission stylo webrender-sw | 329.46 -> 1,314.29 |
296% | pdfpaint issue12306.pdf | linux1804-64-shippable-qr | e10s fission stylo webrender | 337.71 -> 1,337.96 |
293% | pdfpaint issue15559.pdf | windows11-64-shippable-qr | e10s fission stylo webrender-sw | 539.84 -> 2,123.42 |
... | ... | ... | ... | ... |
6% | pdfpaint bug1123803.pdf | windows11-64-24h2-shippable | e10s fission stylo webrender-sw | 233.08 -> 247.31 |
6% | pdfpaint issue15629.pdf | windows11-64-24h2-shippable | e10s fission stylo webrender-sw | 149.14 -> 158.26 |
6% | pdfpaint issue11330.pdf | macosx1470-64-shippable | e10s fission stylo webrender | 291.81 -> 309.58 |
6% | pdfpaint ocs.pdf | windows11-64-24h2-shippable | e10s fission stylo webrender | 158.18 -> 167.75 |
6% | pdfpaint bug931481.pdf | macosx1470-64-shippable | e10s fission stylo webrender-sw | 213.55 -> 226.40 |
Improvements:
Ratio | Test | Platform | Options | Absolute values (old vs new) |
---|---|---|---|---|
7% | pdfpaint issue11306.pdf | macosx1470-64-shippable | e10s fission stylo webrender | 645.72 -> 598.66 |
6% | pdfpaint issue12716.pdf | linux1804-64-shippable-qr | e10s fission stylo webrender | 393.43 -> 369.01 |
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 patch(es) may be backed out in accordance with our regression policy.
If you need the profiling jobs you can trigger them yourself from treeherder job view or ask a sheriff to do that for you.
You can run all of these tests on try with ./mach try perf --alert 44232
The following documentation link provides more information about this command.
For more information on performance sheriffing please see our FAQ.
If you have any questions, please do not hesitate to reach out to fbilt@mozilla.com.
Comment 1•4 days ago
|
||
Set release status flags based on info from the regressing bug 1951955
Comment 2•4 days ago
|
||
Impressive regression... :(
This due to the patch for https://bugzilla.mozilla.org/show_bug.cgi?id=860023.
We added a wasm binding of qcms in order to be able to apply an ICC profile embedded in the pdf to an image.
For example, in issue15559.pdf there is an image 4961×7016 and an ICC profile and the fallback colorspace is RGB which means that before this patch we just did nothing on the image pixels. So we were expecting this perf regression.
In looking in the profiler, we spend around 1.3s in qcms_convert_array
which is a way to much.
But it'll be improved in the next days thanks to an update of the code using qcms: calling precache_output_transform makes qcms_convert_array
almost 3x faster.
Comment 3•4 days ago
|
||
Updated•3 hours ago
|
Description
•