Thorium

[IDA-RE-things]

1 hour ago, ED_Sln said:

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.

Ok, is the result nearest same as it is with original progwrp.dll ? If so, then all ok. If no, I will do more investigation of differences in code of functions.

Because I have only 4-core max CPU on my side. And all works fine for me.

 

And another one: Can you compare also AV1-video decoding with it (about loading of Cores, and entire process of playing it, it should be without freezes ets. (compared to original dll lib)).

We have such test links:

https://bitmovin.com/demos/av1

https://test-videos.co.uk/

Edited June 13, 2024 by IDA-RE-things

[ED_Sln]

46 minutes ago, IDA-RE-things said:

Ok, it the result nearest same as it is with original progwrp.dll ?

WebGL test with the original progwrp.dll showed exactly the same results, both in CPU load with different number of cores and in fps, as well as on 6 was 4 fps, and on 12-16 - 3 fps. The av1 video test also has the same CPU load with both dlls.

[IDA-RE-things]

33 minutes ago, ED_Sln said:

WebGL test with the original progwrp.dll showed exactly the same results, both in CPU load with different number of cores and in fps, as well as on 6 was 4 fps, and on 12-16 - 3 fps. The av1 video test also has the same CPU load with both dlls.

Ok, its fine. I hope you are also done that testing with my build, created for testing (it has some changes in CPU detection code). But because the problem was only with *some* of CPU types, it can be no differences with my previous adapter builds, which also handles multicore CPU's properly.

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

So we waiting for another user, which will have such differences, if there are. (except 66cats).

[ED_Sln]

56 minutes ago, IDA-RE-things said:

Ok, its fine. I hope you are also done that testing with my build, created for testing

Yes, I tested with chrome-xpapi-adapter.5062.2.for.testing. But again, I was testing in a virtual machine. I can also test on a physical computer with i7 3770 with 8 threads.

Upd: Tested it on an i7 3770. The result is the same for the original progwrp.dll and chrome-xpapi-adapter.5062.2.for.testing. On tests: Aquarium 500 fish, 5 fps, CPU load on average 70%. AV1 1080@3000 - 17% CPU load on average.

Edited June 13, 2024 by ED_Sln

[IDA-RE-things]

Ok, thanks, @ED_Sln.

Now we will wait for any other user with such problem.

[dmiranda]

Thank you very much for the diligent support, @IDA-RE-things, I have not properly tried your build. I will do so (hopefully shortly) and report back. Cheers!

[IDA-RE-things]

UPD: finally I have fixed that problem, and 66cats confirmed it in our private messaging. AV1 decoding works fine for him now.
additional problem was in thread synchronization handling on his multicore type of CPU.
So anyone can test fixed version: 

https://github.com/IDA-RE-things/Chrome-xp-api-adapter/releases/download/v1.2/chrome-xpapi-adapter.5062.3.for.testing.zip
and I will create final optimized build later.

[AstroSkipper]

I tested the most recent Chrome XP API Adapter on my system with Pentium 4 single-core and NVIDIA 6200. Generally, this adapter is great as it offers the possibility to enable hardware acceleration. Although the hardware acceleration can be switched on, there are negative side-effects. All popups now have a black background, which then appears as a black frame. Apart from that, the GPU rasterisation does not really work and leads to permanent delays. Furthermore, some of my toolbar icons that belong to extensions disappear.  Thus, I had to switch off GPU rasterisation. Contrary to these problems in Thorium, the hardware acceleration in Mypal 68 and New Moon 28 works great without any problems.  That means hardware acceleration is generally possible on my system and can be used if this feature is supported by a browser, especially with hardware as old as mine. 

Edited June 14, 2024 by AstroSkipper
Update of content

[66cats]

3 hours ago, AstroSkipper said:

hardware acceleration in Mypal 68 and New Moon 28 works great

Edit: My mistake, the "use HW acceleration when available" box wasn't checked. Testing again now.

Another edit: Checking the HW acceleration box makes it slower.

Edited June 14, 2024 by 66cats

[AstroSkipper]

TBH, all this benchmark stuff doesn't help me at all. I play a certain YouTube video on the browsers and know immediately whether hardware acceleration is being used or not. The problem with Thorium, however, is that part of the hardware acceleration (GPU rasterisation) is not running properly on my hardware. Your benchmark images don't help me at all. And comparing different browsers with each other makes no sense either. You have probably not configured Mypal 68 correctly via about:config. Switching on hardware acceleration in the options page of Mypal 68 is not enough, at least in my installation.

[Klemper]

@VistaLover, the issue is resolved in v.126. Not sure if Supermium (Thorium) upstream will follow soon though.

https://github.com/ungoogled-software/ungoogled-chromium/pull/2902

[Klemper]

13 hours ago, IDA-RE-things said:

AV1 decoding works fine for him now.

Is it soft or hard decoding? If my GPU supports it fully, will it be used?

[IDA-RE-things]

5 hours ago, Klemper said:

Is it soft or hard decoding? If my GPU supports it fully, will it be used?

It was software only decoding and rendering. If you will enable GPU acceletation, the CPU load will be less, depending on what parts of acceleration used on you GPU by Chromium code.

[IDA-RE-things]

12 hours ago, AstroSkipper said:

I tested the most recent Chrome XP API Adapter on my system with Pentium 4 single-core and NVIDIA 6200. Generally, this adapter is great as it offers the possibility to enable hardware acceleration. Although the hardware acceleration can be switched on, there are negative side-effects. All popups now have a black background, which then appears as a black frame. Apart from that, the GPU rasterisation does not really work and leads to permanent delays. Furthermore, some of my toolbar icons that belong to extensions disappear.  Thus, I had to switch off GPU rasterisation.

Is shuch HW acceleration works on you Gfx card w/o such side effects, with any of old versions of Chrome, which was officially supported by XP ? (I dont remember version), say some versions released beetween 2008-2013 years. It's need to check the code of libLGESv2, how it handlend on such old Gfx card while translating OpenGL calls to D3D9 calls.

[IDA-RE-things]

@Dixel, @Karla Sleutel, are you not using my DLL at all ? I will know at least. I'm working for people, who need this.