This realy pis*ed me off !!!!!

[lucx]

Hy all,

This news "NVIDIA Driver Brings PureVideo Features To Linux" realy pis ses me off.
I give away for free my Nvidia GF 8500 and bought tow ati cards, an radeon HD 4850 and an ASUS M3A78-EMH HDMI aka radeon HD 3200, and now I'am waiting for a "miracle" maybe, form AMD-ATI, because I'am sic off waiting for that documentation wich they say will be released "sooon", and don't mention flgrx driver on wich is worst , i have tearing even in my browser . this is realy bad for us custommers.

Release that r600 documentation!!

[Kano]

At least it does not work for all h264 streams yet, so you don't have to be too annoyed right now But I hope it impoves in the future.

[elanthis]

Hy all,

This news "NVIDIA Driver Brings PureVideo Features To Linux" realy pis ses me off.
I give away for free my Nvidia GF 8500 and bought tow ati cards, an radeon HD 4850 and an ASUS M3A78-EMH HDMI aka radeon HD 3200, and now I'am waiting for a "miracle" maybe, form AMD-ATI, because I'am sic off waiting for that documentation wich they say will be released "sooon", and don't mention flgrx driver on wich is worst , i have tearing even in my browser . this is realy bad for us custommers.

Release that r600 documentation!!

You gave away working hardware to buy hardware which does not work for your needs, on the assumption that it will at some unspecified future point in time start working.

If you're going to be pissed, be pissed at yourself for doing something so foolish. Only buy stuff that actually does work for you right now. ATI has never said that the cards will be working with all features in any specific time frame. Why would you assume that they would be when nobody ever promised that?

Even if the docs were released today, it seems like it's not expected to have complete support for all features with top 3D performance for at least another 6 to 12 months, if not longer.

I too gave away my NVIDIA card and bought a motherboard with integrated ATI graphics, purely because I wanted to get an Open Source graphics stack. I knew I'd likely have to deal with severely lacking drivers for a good time to come. If I ever get to the point where I can't wait any longer for full-featured RS780 support, I'll just buy a $20 X1550 or GeForce 6600 from newegg to tide me over, and I'll do it without whining because I knew what I was getting into.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814127294

http://www.newegg.com/Product/Product.aspx?Item=N82E16814500048

[TechMage89]

Actually, by all accounts, the initial 2d and 3d acceleration for r600 and r700 is already mostly working, but cannot be released until AMD can work out whatever snarly legal issue is holding back everything behind an NDA.

Bridgman says "real soon now" but he still can't give a date, nor any details on what the legal problem is.

[dkasak]

Bridgman says "real soon now" but he still can't give a date, nor any details on what the legal problem is.

Yeah I've heard that one before. The Enlightenment dev team have been promising E17 for almost a decade now, and they don't even blame legal issues.

[bridgman]

I can only give a date by picking something sufficiently conservative that we have a good chance of meeting it. I would rather keep you informed re: the sequence of events and whwere we stand with those events.

In principle the sequence is pretty simple :

1. Figure out what information we need to provide

2. Create documentation or code which conveys that information.

3. In parallel with #2, go through the IP review work required to make a final release decision (essentially a risk/benefit analysis where the risks are generally catastrophic, ie losing the ability to continue selling into other larger OS markets).

With the 6xx family the first step turned out to be extremely difficult, taking perhapsmuch longer than expected. We ended up having to take a different approach, combining #1 and #2 above and having to actually write working driver code before we could be at all sure what information was required.

We are now following a different plan from the one above :

1. Create working code sequences for the functions required to write 2d and 3d drivers in a set of demo programs.

2. Implement those functions in actual driver code -- I'm sure we'll run across some missing stuff in the process -- so we can get working drivers into users' hands.

3. in parallel with #2, conducting an IP review based on the actual code rather than docs. Once we have the driver code working, we will go back and adjust the documentation to reflect what actually works.

#1 is done, #2 and #3 are underway. Step #1 was the really hard part and the most difficult to schedule.

[TechMage89]

Thanks for the update, Bridgman, so you think the process should be pretty smooth from here on out? In other words, you don't expect any major catches at this point? There have been noises from the devs that they have some 2d/3d accel stuff working on the drivers now, so I gather the driver is pretty far along at this point?

I'm thinking of getting a new graphics card (a 4830 or 4850, if I can find a good deal on one), but I'd like to see working open-source drivers (2d/3d acceleration) before I do.

I'm just sort of wondering whether that time frame is more like "a month or two" or "about six months." I'm guessing, perhaps somewhat hopefully that it's the former.

I'm also wondering if these cards aren't architectured in a way that actually makes it *easier* to write Gallium3d mesa drivers from the start (because they're basically high-performance stream processors with a display adaptor attached, if I understand correctly). Is that what's being done? Or are the devs working from more of a "what we know works right now" point of view to get limited functionality more quickly.

[bridgman]

Anything that involves IP review is hard to predict because it's always possible to have someone identify a serious risk at the last minute, but we're definitely past the worst of it.

Timeframe should be more like "a month or two" than six months - I would like to see this out in the public in 2008.

re: ease of writing a Gallium driver I think it's a wash -- the new chips only have the 3d engine but on previous chips we would have only used the 3d engine for Gallium anyways. We can re-use a lot of the work done for a "mesa classic" implementation for Gallium so I expect to see work being done on a Gallium implementation almost immediately after we get 6xx/7xx to the same level as 3xx-5xx.

[Odur]

I just can't wait to get 3D-acceleration with radeonhd on my HD3870.

I just upgraded to Kubuntu 8.10 and had some serious issues with both the free drivers. The radon driver just put my monitor to sleep. The radeonhd (which worked perfect in 8.04) gave me a blank screen. I had to update radeonhd from tormodvoldens ppa, but even then it didn't work as it should. The screen flickered now and then. Very annoying. So I installed the fglrx from the repository. It works ok, but videos is flickering and I can't switch users (eg. start another x-session) without a total freeze of the whole system. To bad as we are two who use this computer simultaneously.
I tried to install the 8.11 fglrx, but that only gave me ABI errors in the Xorg.0.log file and wouldn't start X, so I had to revert to 8.10.

So you see. I have big expectations on the radeonhd to solve all my problem. I really don't want to have to sell my 3870 and buy an nvidia...

[lucx]

Me eder, dont want to sell my cards, and I'am an fan ati-amd and open sources. So I will wait , but I was nervous when I saw that news.