Closed Bug 900165 Opened 11 years ago Closed 2 years ago

Reflow timer test is significantly slower than webkit

Categories

(Core :: Layout, defect, P5)

ARM
Android
defect

Tracking

()

RESOLVED WORKSFORME
Tracking Status
fennec + ---

People

(Reporter: jwir3, Unassigned)

References

()

Details

(Note: This is split off from bug 898416). Once bug 898416 is fixed, we will likely be back to behavior of Firefox for Android that we had prior to the landing of bug 725981. Unfortunately, this means that our performance on the test (~8 minutes to complete) is still significantly longer than webkit/chrome's performance (~10s to complete). Currently, because bug 898416 has not been fixed, you will need to push rev d10b1ac51ece on top of 2da17db2a304 in order to test this behavior.
tracking-fennec: --- → +
Note that we're anticipating 2 minutes by the time bug 898416 is fixed, but, yes, that is still not 10 seconds.
filter on [mass-p5]
Priority: -- → P5
I ran this on Nightly 52 on a Nexus 4 (same device that mfinkle used in bug 898416 comment 2). The results I got: testDisplay: 1686 testVisibility: 403 testFourClassReflows: 1640 testFourScriptReflows: 1579 testFontSizeEm: 1583 Wall-clock runtime was 1 minute and 55 seconds. For Chrome on the same device I got these times: testDisplay: 993 testVisibility: 379 testFourClassReflows: 1248 testFourScriptReflows: 1184 testFontSizeEm: 1229 Wall-clock runtime was around 43 seconds. Sounds like we still have some work to do here, but this bug is sitting in the wrong component. According to bug 898416 this is more Layout or Graphics than Fennec.
Component: Graphics, Panning and Zooming → Layout
OS: Linux → Android
Product: Firefox for Android → Core
Hardware: x86_64 → ARM
Severity: normal → S3

Firefox Nightly 112.0a1 (2023-02-24):
{ "testDisplay": "35", "testVisibility": "11", "testFourClassReflows": "43", "testFourScriptReflows": "42", "testFontSizeEm": "42" }

Chromium Canary 113.0.5616.0 (2023-02-25)
{ "testDisplay": "72", "testVisibility": "7", "testFourClassReflows": "86", "testFourScriptReflows": "88", "testFontSizeEm": "85" }

Daniel, do you think its safe to close this bug now?

Yeah. Not sure if you were testing on mobile, but I think that's where the focus was here -- so I tested on Android (my Pixel 6a phone) and got:

Firefox Nightly 112:
{ "testDisplay": "145", "testVisibility": "50", "testFourClassReflows": "250", "testFourScriptReflows": "122", "testFontSizeEm": "476" }

Chrome 110:
{ "testDisplay": "205", "testVisibility": "39", "testFourClassReflows": "230", "testFourScriptReflows": "220", "testFontSizeEm": "217" }

Looks like we're in the same neighborhood as Blink now, and winning sometimes / in some categories.

Status: NEW → RESOLVED
Closed: 2 years ago
Flags: needinfo?(dholbert)
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.