Open Bug 1907003 Opened 8 months ago Updated 6 months ago

Canvas2dVideo on Linux has become bimodal/trimodal since ~Jun2024

Categories

(Testing :: Talos, defect, P3)

Product:
Testing
Component:
Talos
Type:
defect

Tracking

(Not tracked)

People

(Reporter: mayankleoboy1, Assigned: sotaro, NeedInfo)

References

Details

(Whiteboard: [fxp])

Canvas2dVideo on Linux has become bimodal/trimodal since ~Jun24 after being very stable since ~Mar24.
This may have started with bug 1901192 and/or bug 1901475 .
Maybe we are falling-off the fast path half of the time?

For comparison, same test in Windows is stable

Summary: Canvas2dVideo on Linux has become bimodal since ~Jun2024 → Canvas2dVideo on Linux has become bimodal/trimodal since ~Jun2024
Whiteboard: [fxp]

@sotoro as the author of the patches from the bugs mentioned in comment 0, do you have any thoughts on whether these may have introduced the multimodal pattern?

Flags: needinfo?(sotaro.ikeda.g)

deleted

Severity: -- → S3
Priority: -- → P3

I take the bug.

Canvas2dVideo was improved by Bug 1887818 on last March. But the score became wrong. It seemed to be broken as in Bug 1901192 Comment 4. Bug 1887818 caused the CPU usage of the CanvasRenderer thread to be almost 100% while Canvas2dVideo was running. This seemed to be breaking Canvas2dVideo scores.

Bug 1901192 and Bug 1901475 just improved CanvasRenderer thread's performance. I wonder if there is a case that Canvas2dVideo score became wrong like Bug 1901192 Comment 4.

Flags: needinfo?(sotaro.ikeda.g)
Assignee: nobody → sotaro.ikeda.g

hi :sotaro this came up in our grooming, we were wondering what the current state of this is? thanks

Flags: needinfo?(sotaro.ikeda.g)
You need to log in before you can comment on or make changes to this bug.