Tear-Free Acceleration For ATI EXA, Xv

[oblidor]

When you say "fixed", do you mean "fixed while running the experimental patch mentioned in the article" ? The regular drivers will still exhibit tearing.

Ah. I only read the thread not the article. I thought it was about that the driver was fixed. I'm running git version of the ati driver...

[duby229]

Yes you are ,

-ati was not in its death throes ever.. at the time AMD were working on opening specs, Alex, before joining AMD (using some work I'd started) had just redesigned -ati to use randr 1.2, this is why the release gap between 6.6.3 and 6.8.0 was so long.

-ati can't die, what would support all the other cards. -ati isn't all reverse engineered, large parts of it were from ATI before the r300, and a lot of the fixes for r300 came out of helpers in ATI. Only the PCI Express and 3D stuff was ever really reverse engineered.

In any case the future from Red Hat's POV is kernel modesetting, and really for us that will obsolete -radeonhd vs -ati. Currently radeonhd is porting accel code from radeon, with kernel modesetting removing everything but the accel code I don't see the reason to offer the same accel code in two packages.

Dave.

So the question in that case is why was any time ever wasted on radeonhd? I mean it's clear that all the time they used developing modesetting code was a total waste and they are just recycling the acceleration code from radeon anyway, why didnt they just get together with the radeon devs in the first place? The simple fact of the matter is that radeonhd has no innovations. None. The modesetting code is worthless, and they havent contributed jack shit to the acceleration code.

Honestly I've never seen more resources wasted then I've seen with the development of drivers for ATi's hardware.

You've got a complete waste of a closed source driver, several completely wasted open source drivers... At this point the only driver that is worth anything is radeon, and it's so incomplete as to be almost worthless. but the kicker of it all is that if everybody had been working together from the beginning it could have been stable by now.

I blame ATi for it personally. I know damn well it was ATi that pushed Novell into the radeonhd project. I know damn well it was ATi that killed the avivo project. I know damn well it was ATi that killed the original closed driver. And I know damn well that it was ATi that started on the new closed driver.

Personally I think it is well past time for some consolidation. The kernel devs need to tell ATi to shove the closed driver where the sun dont shine. And ATI needs to tell novell to suck on an orange. Or better yet to find a pole, sit on it, and spin around endlessly.....

I've seen at least a dozen highly skilled hackers bitch and moan that there isnt enough qualified developers working on linux graphics drivers to make a big difference, but at the same time, they are all wasting there efforts working against each other.... It's a shame.. A damn dirty shame...

[ivanovic]

Uhm, there is a very good reason for radeonhd:

It works better than radeon. That is at least on my system with a hd3850 it is working better. Do not ask me why, that is just the case. Somehow I think that the default for everything till r5xx will be xf86-video-ati where for everything later on it will likely be xf86-video-radeonhd.

No idea how long it will really take till kernel mode setting will be available. This might as well take many more months. Until then for r6xx+ hardware, radeonhd is IMO preferable.

[bugmenot]

So the question in that case is why was any time ever wasted on radeonhd?

There are reasons why there is radeonhd, search the forums. E.g. http://www.phoronix.com/forums/showp...4&postcount=13.

[some-guy]

So the question in that case is why was any time ever wasted on radeonhd?

At that point (when radeonhd was started, and radeon didn't support r500+), radeon was meant to support r100-r400 cards, radeonhd was meant to support r500+, the problem came with atombios. radeonhd said that using the registers diretly is the better (and free-er) solution, but amd wanted radeonhd to use atombios first, then after it was implemented properly, use registers. But radeonhd still used the registers, so radeon started implementing r500+ support with atombios.

Then radeon went ahead in support, and started implementing things faster that radeonhd.

[duby229]

Uhm, there is a very good reason for radeonhd:

It works better than radeon. That is at least on my system with a hd3850 it is working better. Do not ask me why, that is just the case. Somehow I think that the default for everything till r5xx will be xf86-video-ati where for everything later on it will likely be xf86-video-radeonhd.

No idea how long it will really take till kernel mode setting will be available. This might as well take many more months. Until then for r6xx+ hardware, radeonhd is IMO preferable.

I understand that some people believe that radeonhd has some unexplainable, almost religious value.. It's a shame, but there is nothing I can do about it. So let me ask you what makes radeonhd preferable? They switched over to using atombios, and are recycling the acceleration code from radeon anyway... Radeon is more mature, and offers more functionality. So what -exactly- makes radeonhd preferable?

I'm sorry, I've tried, but I just cant see anything more then a waste of valuable time and effort.

[duby229]

At that point (when radeonhd was started, and radeon didn't support r500+), radeon was meant to support r100-r400 cards, radeonhd was meant to support r500+, the problem came with atombios. radeonhd said that using the registers diretly is the better (and free-er) solution, but amd wanted radeonhd to use atombios first, then after it was implemented properly, use registers. But radeonhd still used the registers, so radeon started implementing r500+ support with atombios.

Then radeon went ahead in support, and started implementing things faster that radeonhd.

So things didnt work out the way things had been envisioned. I can understand that, But in that case why is it still being funded and developed? Once it was discovered that radeon was capable of doing what needed done faster and cheaper than radeonhd, why wasnt it killed right then and there?

The radeonhd devs are fully capable and highly qualified, so why arent they working on radeon right now?

[some-guy]

The radeonhd devs are fully capable and highly qualified, so why arent they working on radeon right now?

FWIK, mostly disagreements on how things should be done.

I'm sure one will be dropped once kms is finished, because at that time the main differences will be 2D and video acceleration, which imho isn't enough reason to have 2 (2D) camps.

BTW, both projects use the same drm and 3D code...

[Extreme Coder]

Or even better, they could merge :P One can dream..
What matters most to me right now is the 3D, I can't wait till they move to Gallium, but I think that won't be till next year..

[duby229]

Or even better, they could merge :P One can dream..
What matters most to me right now is the 3D, I can't wait till they move to Gallium, but I think that won't be till next year..

Nope definitely not till next year.. My biggest complaint is that I cant get the damn xserver to build from git for some damn reason, and nobody seems to have any explanation why. I've asked on this forum several times. I've asked on gentoo's forums several times. I've asked on several mailing lists several times, and so far I've got nothing from nobody. I mean do I actually have to wait until next year?

ATi, do you actually expect people to wait that long? DRI as it is now may not be all that great, but at least the infrastructure is there. Why not get everybody together and stabilize the drivers for DRI right now, and then start working on DRI2? That way we could have a stable and reliable driver now --AND-- later...

Makes sense to me...... But of course lets all vote for change...