About Video performance on HTML by using WebView on AndroidKK4.4.2

[yuuki]

Dear all

Our customer plays a video on HTML by using WebView of AndroidKK4.4.2.

<Environment>
Baord: MX6Q-SABRE-SDB
Display: HDMI output

OS: AndroidKK4.4.2 FSL BSP
Package: AndroidKK4.4.2_Extended Feature Packages(android_kk4.4.2_1.0.0-ga_omxplayer_source)
Patchs: 0001-Fix-the-video-only-stream-can-t-play-issue.patch
             html5_video_issue.patch

<Test>
Issue_01_video_performance.pdf: outline
Issue_01_index9.mp4/Issue_01_index10.mp4: captured videos
Issue_01_sample: test samples

They test Video Playback in this environment.

When two plane is composed, frame rates decrease.
- WebView is overlaid on the picture.
- The frame rates decrease from 25fps to 12fps

When it is not overlaid, this problem does not happen.

When Chromium is used, this problem does not happen.
(In this case, even if it is overlaid, the problem does not occur.)

It seems that 2D acceleration doesn't work well when WebView is overlaid on the picture.

They investigate this problem.
Which cord should they refer to?

May I have advice?

Best Regards
Yuuki

Original Attachment has been moved to: Issue_01_sample.zip

[yuuki]

Daer all,

Can anyone answer this question?

Please help me.

Best Regards,
Yuuki

[yuuki]

Daer all,

Our customer has to settle a problem by the end of February.
Does anyone have any information?

Please help me.

Best Regards,
Yuuki

[SergioSolis]

Hello Yuuki,

Our team of experts is reviewing your case and I will be updating you as soon as I have a response from them.

[shigenobukatagi]

Hi Sergio,

I'm working with Yuuki-san for this case, let me share the issue reproducing status.

We have confirmed the issue can be seen on our SabreSD board.

You may know but let me share how to check customer's issue just in case.

$ setprop debug.sf.showfps 1

$ stop

$ start

Here we can see framerate of SurfaceFlinger at logcat.

Attached 2 shows how fps varies.

Also I checked SurfaceFlinger profile using following command.

$ dumpsys SurfaceFlinger

log files are also attached. so far, I didn't see any big difference in them, but I hope these help you.

Let me know if you have any problem to reproduce it.

Best Regards,

Shigenobu

[SergioSolis]

Yuuki, Shigenobu, where can I get the patches you mention about in this thread?

[yuuki]

Dear Sergio-san,

Thank you for your support.

I attach patches.

I wish that you reproduce this problem.

Best Regards,
Yuuki Murasato

[yuuki]

Dear Sergio-san,

Sorry to chase you, how about the situation afterward?
Please advise of your findings.

Best Regards,
Yuuki

[SergioSolis]

Hello Yuuki, here's the response from our R&D team:

"By top CPU usage, the Browser hold CPU usage is different in these two cases.

index10.html is 10%-15%, index9.html 20%-25%.

The fps performance drop is caused by Browser.

Browser like SurfaceFliner, it will combine all browser's layer to one surfaceflinger's layer.

The two cases's difference is index9.html has one more browser's layer.

So, Browser will cost more to do combination, the performance will drop.

This is normal case.

Then Chromium doesn't have this issue that Chromium should do optimization.

It is recommended to use Chromium rather than default Browser."

Please let me know of any question you may have.

Have a great day,
Sergio

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[yuuki]

Dear SergioSolis-san,

Thank you very much for your support.

About this problem, local FSL FAE has been contacted to R&D team internally.
Therefore I close this thread.

Best Regards,
Yuuki

[SergioSolis]

Very well, I will set this thread as "assume answered" then.