Catalyst 10.6 - Video Quality Improvements

There are no problems with MPC HC, PowerDVD (7, 8, 9, 10), ArcSoft TMT and ffdshow's DXVA decoder, but VLC's problems are caused by ATi's drivers? How is this possible?

 

Because as stated a few posts back the way VLC are doing it is very slow using ATI drivers.

 

It's bad for them.
It seems the only market ATI is aiming at is the Gamers.

Their 2D performances are bad, now the Video acceleration...

I hope they fix it.
Next time I buy a graphic card it won't be ATI's probably.

 

Hey don't blame ATI for this after all VLC has a crappy DXVA implementation. Hell Media Player Classic HC has DXVA support for 3 years now and they didn't have the advantage of using the FFMPEG library for hardware decoding.

 

Is there any validity that VLC's method is better than all the other products aforementioned?

 

In that thread I linked, caveman-jim reports that AMD is now looking into it.

http://www.rage3d.com/board/showpost.php?p=1336291947&postcount=6

Edit:

http://www.rage3d.com/board/showthread.php?threadid=33965039


​

 

Yeah, so...um. They've had DXVA acceleration in a variety of applications (WMP, MPC, GOM player to name a few) for quite a few years now and just because one lone application can't manage it, suddenly their video accerlation sucks.

My oh my.

Regards,
SB

 

VLC is using DXVA2 - perhaps it's a generic problem with that API.

 

I don't know enough to judge whether it the fault of the Drivers or VLC team.
I just have a feeling it wouldn't happen to nVidia.
nVidia would make sure they can easily use it.

ATI seems to care only about gamers.
It's the attitude we see in the GPGPU arena.

If was the one who makes the call I would take some people to make X.264 GPU Accelerated by ATI.
That's something which would make me buy their stuff.

 

I wonder if you were also this concerned when Nvidia couldn't get pure video to work at all when it was first introduced and heavily marketed?

And considering ATI has already stated that a solution is probably already been conceived... Yes, I'm sure you are right and they just don't care about video acceleration. And it wasn't REALLY ATI that worked with them to figure out what was going wrong.

Regards,
SB

 

Am I the only one to have a bug in the FETCH4 with DF24 on DX9 ?
It seems the sampling pattern is wrong, my shadows are all borked (same with ATi FETCH4 demo).
(Bug was also there in 10.5 and 10.4 AFAIR)

 

I was being sarcastic. I should have put a sarcasm smiley on that.

Regards,
SB

 

Is it something that wasn't in the DXVA2 standard?

 

who the bloody hell is terry

 

Terry Makedon, aka CatalystMaker. Catalyst Product Manager @ AMD.

 

It's the attitude of ATI towards anything which isn't Game accelaration. That's my problem.
Give or Take a prbolem they all do have some issues.
The question is how they deal with it.

I'll quote what's written in "Anandtech" about it:

Taken form the 2nd page, first paragraph.

That's my point.

 

And withing MPC you see different codepaths for ATI and <insert everyone else including unknown GPUs>. What does this tell you about the implementation of a standard?

just because MPC managed to get it working with some crutches doesnt mean there arent problems, otherwise you could call IE6 "standards-compliant" because all major sites adopted special workarounds for this popular disaster.

 

Haters gonna hate...