1969312 - 126.9 - 2.33% espn largestContentfulPaint / netflix largestContentfulPaint + 17 more (Linux) regression on Tue May 27 2025

Status: RESOLVED WONTFIX

(Core :: DOM: Performance APIs, defect)

Description

[Treeherder Bug Filer]

Perfherder has detected a browsertime performance regression from push c3ac0c48fcab094b5b9324b905a29629011851ef. As author of one of the patches included in that push, we need your help to address this regression.

Please acknowledge, and begin investigating this alert within 3 business days, or the patch(es) may be backed out in accordance with our regression policy. Our guide to handling regression bugs has information about how you can proceed with this investigation.

If you have any questions or need any help with the investigation, please reach out to [e-mail]. Alternatively, you can find help on Slack by joining #perf-help, and on Matrix you can find help by joining #perftest.

Regressions:

Ratio	Test	Platform	Options	Absolute values (old vs new)	Performance Profiles
127%	espn largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	123.88 -> 281.08	Before/After
103%	yahoo-mail largestContentfulPaint	linux1804-64-shippable-qr	bytecode-cached fission warm webrender	50.29 -> 102.19	Before/After
80%	yahoo-mail largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	50.36 -> 90.81	Before/After
46%	expedia largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	121.58 -> 177.38	Before/After
44%	espn largestContentfulPaint	linux1804-64-shippable-qr	cold fission webrender	501.47 -> 719.91	Before/After
20%	expedia largestContentfulPaint	linux1804-64-shippable-qr	cold fission webrender	1,662.90 -> 1,990.63	Before/After
17%	microsoft largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	123.28 -> 143.71
14%	reddit largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	126.82 -> 144.02
11%	cnn largestContentfulPaint	linux1804-64-shippable-qr	bytecode-cached fission warm webrender	485.61 -> 537.40	Before/After
10%	youtube largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	1,020.75 -> 1,121.77	Before/After
...	...	...	...	...	...
5%	google-slides largestContentfulPaint	linux1804-64-shippable-qr	bytecode-cached fission warm webrender	1,098.73 -> 1,155.23	Before/After
5%	imdb largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	869.12 -> 911.68
5%	twitter largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	599.38 -> 627.51
4%	google-slides largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	1,679.17 -> 1,738.70	Before/After
2%	netflix largestContentfulPaint	linux1804-64-shippable-qr	fission warm webrender	638.92 -> 653.82

Details of the alert can be found in the alert summary, including links to graphs and comparisons for each of the affected tests.

If you need the profiling jobs you can trigger them yourself from treeherder job view or ask [e-mail] to do that for you.

You can run all of these tests on try with ./mach try perf --alert 45323

The following documentation link provides more information about this command.

Comment 1

[BugBot [:suhaib / :marco]]

Set release status flags based on info from the regressing bug 1963917

Comment 2

[Sean Feng [:sefeng211]]

I think this is a base line change, I am going to run some tests to confirm that.

Comment 3

[Sean Feng [:sefeng211]]

Yeah, this is a base line change.

So we use renderTime as the value for LargestContentfulPaint, and this is how browsertime gathers that information. It takes a max of renderTime and loadTime.

I manually loaded ESPN, so this is how the LCP entry looks like before bug 1963917.

{
  "name": "",
  "entryType": "largest-contentful-paint",
  "startTime": 251,
  "duration": 0,
  "renderTime": 0,
  "loadTime": 251,
  "size": 174240,
  "id": "",
  "url": "https://a.espncdn.com/combiner/i?img=%2Fphoto%2F2025%2F0528%2Fr1499487_1296x518_5%2D2.jpg&w=1320&h=528&scale=crop&cquality=40&location=center&format=jpg",
  "paintTime": 0,
  "presentationTime": null
}

Max(renderTime, loadTime) is 251.

After that patch, it becomes

{
  "name": "",
  "entryType": "largest-contentful-paint",
  "startTime": 648,
  "duration": 0,
  "renderTime": 648,
  "loadTime": 262,
  "size": 174240,
  "id": "",
  "url": "https://a.espncdn.com/combiner/i?img=%2Fphoto%2F2025%2F0528%2Fr1499487_1296x518_5%2D2.jpg&w=1320&h=528&scale=crop&cquality=40&location=center&format=jpg",
  "paintTime": 648,
  "presentationTime": null
}

notice that now renderTime is exposed, so Max(renderTime, loadTime) becomes 648 instead of 262 for this case.

Does it sound correct you sparky?

Comment 4

[Greg Mierzwinski [:sparky]]

Yes, that looks correct to me. I believe that this can be closed as a WONTFIX.

Just a note that these results now come from the following support script instead of the results.py file: https://searchfox.org/mozilla-central/source/testing/raptor/browsertime/support-scripts/browsertime_pageload.py#22

Comment 5

[Sean Feng [:sefeng211]]

Thanks!