Nvidia vs ATI

[RealNC]

The irony here is that the ATI drivers worked right away with 2.6.28 :P Probably a one-time occurrence though

[t.s.]

The irony here is that the ATI drivers worked right away with 2.6.28 :P Probably a one-time occurrence though

eventhough, I still recommend nVidia over ATI, like others. IMHO, I think nVidia still won't open their driver cause they know they have none this far that can match their driver performance/stability. So, when nVidia open their driver, it is a signal for us to buy ATI card, heheh.

[bridgman]

I know it's a popular myth that we re-started support of the open source graphics community because we had "given up" on fglrx, but that just ain't the case.

As part of AMD, we picked up a number of new enterprise customers and those customers felt strongly that we needed at least basic open source driver support out-of-the-box with common distros. Something like the "nv" driver would probably have been sufficient at first, but with the transition to compositing desktops in full swing we felt it was worth going further and including both 3D and some basic video support in the open drivers.

[RealNC]

So we can keep saying that AMD is evil. Because it would be good to not have "3D and some basic video" but "high-performance, stable 3D and full featured video with nothing lacking".

[energyman]

I have a
X2 6000+
HD3870
4Gb of DDR2 800
Amd770+SB700 board
two Samsung 500GB harddisks.
A couple of usb devices
a hvd-scsi card
an audigy2 card
a tv card.
4 fans
I measured power consumption:

Desktop, KDE 4.2-svn: 106-108W
Desktop, KDE 4.2-svn, watching video: 106-110W
Desktop, KDE 4.2-svn, turned on composite effects: 106-108W
Desktop, KDE 4.2-svn, watching TV: 115-120W
Desktop, KDE 4.2-svn heavy compiling: up to 200W
virtual terminal, idle: 160W
ut2004: up to 295W

I am buying a new PSU soon - 450W. It will be enough.

[bridgman]

I don't get the difference. We are providing enough HW info and sample code to implement high performance, stable 3D. Whether that happens or not is a different story -- high performance, stable 3D is godawful expensive to develop if you are looking for performance parity with Windows.

What I think you will get realistically is "decent performance and stable", unless some billionaire decides to fund maybe $50M worth of development work into the open source stack out of the goodness of his/her heart. You know the story with video. We will open up what we can that doesn't push us into high-risk territory w.r.t. DRM and related obligations.

If you want to call that evil I guess that's up to you

[RealNC]

I usually don't attack the examples in conversations but the point instead, but in this case... $50M? Ah, come on! It can't be that expensive. No, seriously. I mean, you already have it working for Windows. Of course you (well OK, not *personally* you, but you know what I mean) don't want to open up the catalyst driver due to competition with NVidious. But in either case, calling you "evil" is justified anyway.

[Melcar]

I usually don't attack the examples in conversations but the point instead, but in this case... $50M? Ah, come on! It can't be that expensive. No, seriously. I mean, you already have it working for Windows. Of course you (well OK, not *personally* you, but you know what I mean) don't want to open up the catalyst driver due to competition with NVidious. But in either case, calling you "evil" is justified anyway.

The development that has gone into the Windows driver is enormous. I don't know why everyone underestimates the resources, time, and effort needed to achieve the same performance level as the Windows proprietary driver.

[RobbieAB]

As part of AMD, we picked up a number of new enterprise customers and those customers felt strongly that we needed at least basic open source driver support out-of-the-box with common distros. Something like the "nv" driver would probably have been sufficient at first, but with the transition to compositing desktops in full swing we felt it was worth going further and including both 3D and some basic video support in the open drivers.

And this is as blatantly as AMD are ever going to put it people: The open source effort wasn't really down to us, it was down to big OEMs leaning on them. Our joy and satisfaction with the effort is secondary at best.

[bridgman]

$50M is probably low, if you're looking for Windows parity. If you were funding the current activity level, including both corporate and volunteer devs, you're probably looking at somewhere between $3M and $5M/yr in terms of what that work would cost to fund. I think you would need a minimum of 5x that activity for maybe 2 years to get where you want to be, and that works out to $50M pretty quickly.

If you can live with what we're expecting to see, which is solid drivers without any significant optimization work, then $50M is obviously high and we think that can happen with the resources currently available in the community.

If you're saying that we are evil because we don't want to open up our proprietary code and risk losing our ability to compete in the much larger Windows and MacOS markets then I guess I can live with that. Seriously, though, if you ask the developers who are doing all the work today I think they will tell you clearly that they don't want to work in our proprietary code anyways.

And this is as blatantly as AMD are ever going to put it people: The open source effort wasn't really down to us, it was down to big OEMs leaning on them. Our joy and satisfaction with the effort is secondary at best.

Yes and no. You all want more or less the same thing. The original impetus for the project came from our enterprise customers, but the plan was worked out with community developers who understand your needs as well as enterprise needs. Obviously gaming support and video playback are more important to you than to a typical enterprise IT department, but we are providing enough hardware information to cover all the bases.

If we have the right plan then everyone will think we more or less planned around their needs, but I'm not going to blow smoke and tell everyone "we did this only for you"