Closed Bug 1387841 Opened 8 years ago Closed 3 years ago

Slow SVG performance and hang in Chalkboard HTML5 Benchmark

Categories

(Core :: SVG, defect, P3)

Product:
Core
Component:
SVG
Version:
54 Branch
Type:
defect

Tracking

()

Status:
RESOLVED WORKSFORME
Tracking Flags:
Tracking Status
firefox57 --- wontfix
firefox58 --- wontfix
firefox59 --- ?

People

(Reporter: mockingbird, Unassigned)

References

(Blocks 1 open bug,
URL
)

Details

(Keywords: html5, perf)

User Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/62.0.3175.0 Safari/537.36 Steps to reproduce: Ran the Chalkboard performance benchmark: https://testdrive-archive.azurewebsites.net/performance/chalkboard/ Actual results: The test crashes. about:support reports this error: CP+[GFX1-]: Failed to allocate a surface due to invalid size (CDT) Size(14321,9828) Expected results: The test should complete within 5-15 seconds on my platform with hardware rasterization.
Keywords: html5
See Also: → 1289866
No crash here, but I'm also seeing many errors in about:support under Graphics Failure Log: >(#0) CP+[GFX1-]: Failed to allocate a surface due to invalid size (CDT) Size(14321,9828) >(#118) CP+[GFX1-]: Failed to allocate a surface due to invalid size (CDT) Size(14724,10104)
Status: UNCONFIRMED → NEW
Ever confirmed: true
(In reply to Virtual_ManPL [:Virtual] - (please needinfo? me - so I will see your comment/reply/question/etc.) from comment #1) > No crash here, but I'm also seeing many errors in about:support under > Graphics Failure Log: > >(#0) CP+[GFX1-]: Failed to allocate a surface due to invalid size (CDT) Size(14321,9828) > >(#118) CP+[GFX1-]: Failed to allocate a surface due to invalid size (CDT) Size(14724,10104) The test crashes, firefox doesn't itself crash, but the test will not complete and after the test fails, those errors show up in about support. Can you please confirm? Close Firefox, close the process, start firefox, look at abbout:support, it is clean. Run the benchmark, the test will crash in the middle. Look in about:support, the errors are there. Clearly something is wrong.
Flags: needinfo?(Virtual)
I'm passing test, no any crash or hang. Just same errors.
Blocks: paint-fast
Flags: needinfo?(Virtual)
(In reply to Virtual_ManPL [:Virtual] - (please needinfo? me - so I will see your comment/reply/question/etc.) from comment #3) > I'm passing test, no any crash or hang. Just same errors. The test does not crash but the animation disappears several seconds in. Please confirm.
Flags: needinfo?(Virtual)
I can't confirm dispersing animation, hangs or crashes, just like I wrote, I'm passing test without any visible issues, except these errors in about:support under Graphics Failure Log.
Flags: needinfo?(Virtual)
That's because you have hardware acceleration disabled in options. Please untick "use recommended performance settings" tick "use hardware acceleration when available" and set content processes to 7 Thank you.
Flags: needinfo?(Virtual)
(In reply to mockingbird from comment #6) > That's because you have hardware acceleration disabled in options. > > Please untick "use recommended performance settings" > > tick "use hardware acceleration when available" > > and set content processes to 7 > > Thank you. No, I have also enabled hardware acceleration, same as e10s (Electrolysis) which is also enabled. Looks like the test hang is specific to you only or you're using some extension which interfere with test. But I can reproduce errors.
Flags: needinfo?(Virtual)
URL: https://testdrive-archive.azurewebsit...
Has Regression Range: --- → no
Has STR: --- → yes
Summary: SVG slow performance and crashing → Slow SVG performance and hang in Chalkboard HTML5 Benchmark
Thanks, you are correct. Disabling AdBlock Plus allowed the test to complete. It scored at 156.68 though, which is atrocious. I also noticed that there are other reports of this: 786064 1135411 1135548 1387841 Can you please report your score?? Thanks.
Flags: needinfo?(Virtual)
I got these results on my PC (E6550 @3,5GHz + 4GB RAM + GTX 750 Ti): Mozilla Firefox 57.0a1 (2017-08-20) (64-bit) - 29.93 Seconds Mozilla Firefox 57.0a1 (2017-08-20) (32-bit) - 32.73 Seconds Mozilla Firefox 55.0.1 (64-bit) - 29.05 Seconds Mozilla Firefox 55.0.1 (32-bit) - 30.79 Seconds Google Chrome (Dev) 62.0.3188.4 (32-bit) - 6.73 Seconds Google Chrome 60.0.3112.101 (64-bit) - 6.27 Seconds Google Chrome 60.0.3112.101 (32-bit) - 6.53 Seconds
Flags: needinfo?(Virtual)
Keywords: perf
See Also: → 1135411, 786064, 1135548
Windows 7 64-bit on AMD FX-6300, 16GB RAM, AMD 380X: Mozilla Firefox ESR 52.3.0 32-bit - 73.23 Seconds & very choppy, 50% CPU load Mozilla Firefox Nightly 32-bit "DogeFox" 57.0a1 (2017-08-20) - 27.56 Seconds & choppy, 50% CPU load Pale Moon 27.4.1 32-bit - 19.15 Seconds & smooth, 15-20% CPU load Google Chrome 60.0.3112.101 32-bit - 5.91 Seconds & smooth, 30% CPU load Microsoft Internet Explorer 11.0.35 32-bit - 7.19 Seconds & smooth but choppy sound, 25% CPU load Nightly definitely has improvements over previous Firefox versions, but still lagging behind the others. Especially the high CPU and choppy display is pretty bad.
Priority: -- → P3

On latest nightly : 6.7s
Chrome: 11s

I am calling it fixed.

Status: NEW → RESOLVED
Closed: 3 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.