Quote:

Can't you use the gfx card blitter to blit overlay-like YUV (or whatever) pixelformat buffers to RGB bitmaps?

Quote:
Probably they had in the past but BE PPC support has been broken for some time. To make it build at all 'generic cpu' has to be selected in configure options. Which makes it build but still the test-suite fails in many cases (klick).

av1 on the other hand (at least the dav1d guys) is open for contributions from 'alternative' arches and test suite passes on ppc. One point in the roadmap (klick) even is "Accelerate for less common architectures, like PPC, SSE2 or AVX-512." But don't know whether current av1 is faster/slower on ppc compared to vp9.

Yes its off-topic here.

@Andreas_Wolf oh i got decoding and encoding mixed up.

[ Editiert durch ThePlayer 04.03.2021 - 18:49 ]

I am not sure where you guys are going with this. This is certainly off-topic here. I'm not in the business of writing drivers for hardware video decoders and any changes in CGX are totally out of scope. Video playback in Wayfarer will, at least for now, be limited to cards with overlay support.

[ Edited by jacadcaps 04.03.2021 - 18:22 ]

> We need Drivers for newer Cards with at least
> UVD3 support better UVD4 best would be UVD5.

UVD3 because of MPEG-4 decoding, right? And UVD5 because of 4K resolution, right? But why UVD4? Because of Fluid Motion? H.265 decoding starts with UVD6, btw. (and 10-bit colour depth support with UVD6.2).

https://en.wikipedia.org/wiki/Unified_Video_Decoder#Format_support

> So everyone who needs Video encoding could get
> himself an PM 11.2.

I don't think video encoding (VCE/VCN for H.264, UVD6.3+/VCN for H.265) should matter before video decoding is supported.

Quote:

We need Drivers for newer Cards with at least UVD3 support better UVD4 best would be UVD5.
So everyone who needs Video encoding could get himself an PM 11.2.

Quote:

They seem to have no support for BE PPC at all in libvpx.

YT falls back to good-ol h264 if vp9 and av1 aren't available.

Quote:
I thought vp9 would be youtubes' standard video format? What's the (faster) alternative?

As far as I got it av1 will be the standard some time in the future but currently it's not widely deployed. It also got lots of SIMD-code for x86_64 and arm to boost its' performance but I doubt it has equally good Altivec/VSX-code for PPC.

No. But remember that the UI pipeline is a separate process in mainstream browsers. There's no main thread bottleneck there. You're also compositing several textures to build the HTML view, unlike with Wayfarer where everything is rendered into one cairo surface.

Quote:

So - basicaly speaking - all the mainstream browsers use overlay when playing YT videos?

Quote:

Smooth playback without overlay (or rendering to a custom window outside of the HTML view) is not technically possible. A faster CPU or hw video decoding wouldn't help. As a matter of fact, hardware video decoding would likely make things slower if rendering inside the HTML, due to the fact that video frames would have to be read from graphics card memory by the CPU.

> if [...] 240/320p worked smoothly I would be happy,
> specially if it used overlay.

I would prefer 240p/360p to work smoothly without using overlay, because then overlay could probably be used to smoothen 480p ;-)

> Does the tcp/ip stack and the Sana2 driver architecture
> allow to offload cpu as much as other OSes?

The SANA-II driver interface is definitely not fit for current Internet transfer speeds. As Olaf 'olsen' Barthel has been telling during at least the last decade, there's no making use of many capabilities of modern NICs (and doing calculations in software instead) and too much copying around of data. I doubt this is significantly different for MorphOS compared to AmigaOS.

Quote:

That would rock. For G4 machines if lower quality streams like 240/320p worked smoothly I would be happy, specially if it used overlay.

For G5 720p will be very decent and if hardware accelerated video playing worked smoothly it would be a good reason to upgrade.


Does the tcp/ip stack and the Sana2 driver architecture allow to offload cpu as much as other OSes?

Quote:

thanks!

Make sure your system clock is set up correctly.

I keep getting certifcate errors on a fresh install of morphos. Bit of a noob with this can you guys help?

[ Edited by Rats 02.03.2021 - 16:15 ]

Quote:

There's room for improvement vs OWB, for sure. I'm hoping I'd be able to use overlay in windowed mode.

You are the man jacadcaps! Thanks so much for this browser.

Ok, is there still room for improvement for Radeon 9x00 class graphics cards? I think these one are the most common cards used in MorphOS systems. Maybe for 9600/9800, as used in G5s or Powerbooks?

Quote:

Firstly great work again, you don't know how pleased I am to see this progress.

I watch 720p YT on my G5 Quad using OWB at the moment and as long as I let it buffer for a bit at the beginning it handles it fine using the full screen script.

Is what you are trying to achieve with Wayfarer radically different to this?

As for hardware acceleration, if its a option for us with PCIe machines I'd welcome it, even if I have to use a different version on my Powerbook G4 and MacMini.

> So, logicboard based GFX architecture is going to cripple
> most Apple systems, like MacMinis, PowerBooks from having
> working, fluid video within Wayfarer, if I understand you correctly

Without UVD support, there is (apart from potential differences in CPU performance) no less crippling with video on the PowerMacs even with more modern graphics cards put into them.

> I still can't solder one onto my PowerBook, so...

But you could put one into your "2.3GHZ G5", no? :-)

So, logicboard based GFX architecture is going to cripple most Apple systems, like MacMinis, PowerBooks from having working, fluid video within Wayfarer, if I understand you correctly, but will there be an option to save the video file as an .mp4 or such like on the majority of systems?

I'm not so hopeful about that, but that's what I used to do mainly with my MacMini for YouTube videos, but from what you've said - as the video and audio are split into "snippets" and re-assembled "on-the-fly" through the browser's internal capabilities I'm thinking that any script is going to be a non-starter, but I could be wrong . . . go on . . . tell me if it's possible, I can take it!

Would there be a way to implement a lower resolution, less CPU intensive way of handling video from most websites that would still allow MorphOS users to view such content, with just a MacMini or PowerBook, with the GFX limitations they are constrained by?

Quote:

You also cannot solder a 2,7 GHz G5 into it, so anyway it's kinda useless when it comes to YouTube video playback, right?