Thorium

[Dixel]

3 hours ago, 66cats said:

[long & OT]

Elsewhere in the intertubes, found https://test-videos.co.uk/vids/sintel/mp4/av1/1080/Sintel_1080_10s_1MB.mp4. And, of course, it freezes.

Inability of this browser to properly play such common format as AV1, which is literally everywhere now, is indeed on-topic, so don't be shy to post your issues.

I just checked the link, it also froze on me, so we both have the same issue on the  same series of 700 nVidia cards, despite being on different OS.

[Dixel]

3 hours ago, 66cats said:

I think he meant the opposite, he only enabled AV1 (and the browser showed nothing, there being nothing to show). That's how i read it.

"enhanced-h264ify" needs to be fully uninstalled for the browser to be able to properly load AV1, HEVC and VP9.

[Dixel]

5 hours ago, 66cats said:

Also happens with Supermium on this box (980x/gtx770), just checked.

I don't have GTX 980 or anything from 900 series, but I've just tried on an El-cheapo Acer notebook with nVidia 820M,

Driver version 348.12 (Acer exclusive, came on the disk). It has the same issue with AV1 freezing!

S,o we have a third, fully confirmed case!

[66cats]

12 minutes ago, Dixel said:

we have a third, fully confirmed case!

I might have been too quick on the trigger, maybe that file itself is corrupt. Try https://bitmovin.com/demos/av1, see if it freezes for you with IDA-RE-things' progwrp.dll (it does for me), and plays normally with win32's (it does, for me).

[Dixel]

3 hours ago, 66cats said:

see if it freezes for you with IDA-RE-things' progwrp.dll (it does for me), and plays normally with win32's (it does, for me).

IDA-RE-things says the dll has nothing to do with the acceleration. Probably nothing more than placebo effect and/or pure coincidence?

"We have no any Video decode acceleration handling in progwrp.dll. "

https://msfn.org/board/topic/186133-thorium/?do=findComment&comment=1267631

 

[IDA-RE-things]

Yes, I have wrote that I have installed "enhanced-h264ify" to enable requered video formats and disable not requered ones. When I have leaved only AV1 as "allowed", The youtube player says me on that page: "The video cannot be played in this browser."

Its because youtube gives such formats not for all possible videos on it. And its because I have requested for test link. But have found by myself.

All played without freezes and problems with my gfx card and CPU. Can't reproduce on my side.

Edited June 11, 2024 by IDA-RE-things

[66cats]

2 minutes ago, IDA-RE-things said:

. Can't reproduce on my side.

Can only show you screen recordings:

win32's (version 1.1.something)

yours (version 1.2.something)

Again, if you know of a better way to show you what's happening, i'll give it a shot.

[IDA-RE-things]

@66cats, thanks, I have compared 2 your recordings.

And have found how to reproduce such freezes. Its occured only on weak CPU's (I have checked on one). When CPU usage is very high.

It because in my Adapter DLL, the GPU process priority was fine-tuned for very weak CPU's to prevent whole system freezing, when videos played. (Even on 1-core weak CPU w/o hyperthreading). And to prevent audio glitches (in conjunction with Audio process priority). Even by cost of "freezing" of some video frames. Its here by design of my workaround for it.

This problem not happens on fast CPU's, not loading the system very high. (Should not happen at least)

I will experiment with process priorities more, and may be provide some switch to configure it, Or set another process priorities.
 

 

Edited June 11, 2024 by IDA-RE-things

[IDA-RE-things]

Hello, I have created and uploaded small workaround for that issue. the build "5061.2". Now here 2 versions of adapter inside archive. One (without suffix) -- dont changes GPU process priority at all, (for those of you, who having such freezes), and another with suffix -- for wery weak CPU's w/o hypertreading (also works fine on very fast CPU's). Try it and write if it fixes the problem.

Edited June 12, 2024 by IDA-RE-things

[66cats]

3 hours ago, IDA-RE-things said:

2 versions of adapter

Hi, thanks, check your PM.

[D.Draker]

18 hours ago, 66cats said:

Hi, thanks, check your PM.

Why private? You aren't hiding anything from us, by any chance, are you?

[NotHereToPlayGames]

8 hours ago, D.Draker said:

Why private? You aren't hiding anything from us, by any chance, are you?

When I first read that, my thought was "uh oh, another false positive", lol.

[66cats]

30 minutes ago, NotHereToPlayGames said:

hiding anything from us

Two pages of back & forth, debugging my HW issue (still unresolved). Didn't want to derail this thread.

Edited June 13, 2024 by 66cats

[IDA-RE-things]

At least no one bothered us with their unnecessary comments, and this also helped. It is not necessary to show off the entire process of problem solving.

For those of you who have been analogous multicore-handling problem, can test intermediate version of adapter (not fogot to rename to "chrome-xpapi-adapter.dll") :

The problem was in detection of *some* CPU's (not all) as single-core instead of multi-core. And now should be fixed.

https://github.com/IDA-RE-things/Chrome-xp-api-adapter/releases/download/v1.2/chrome-xpapi-adapter.5062.2.for.testing.zip

And try onother one test, Which should show how many cores loaded. Should be loaded all possible cores.

https://webglsamples.org/aquarium/aquarium.html

Should be runned w/o hardware acceleration (i.e without switches like --use-angle=d3d9) to be handled only by software.

Edited June 13, 2024 by IDA-RE-things

[ED_Sln]

2 hours ago, IDA-RE-things said:

And try onother one test, Which should show how many cores loaded. Should be loaded all possible cores.

I checked it in a virtual machine with 6, 12 and 16 cores. With the increase of cores the load only decreases. But the fps even slightly decreased, on 6 cores it is stable 4, and with more cores - 3.