If this is your first visit, be sure to
check out the FAQ by clicking the
link above. You may have to register
before you can post: click the register link above to proceed. To start viewing messages,
select the forum that you want to visit from the selection below.
can i have a pay to if i drop my ATI cart and telling FUD to ?
No, but it would be nice if they did!
Believe what you want, I only responded to bulletxt question since I do have experience with DCC apps this highend FirePro targets. This is not only the case on Linux, since I know a lot of people in VFX (primarily in Denmark and Sweden) who use Windows, and they have the same feelings about ATI. If this is going to change, the experience with an ATI card must be much better than current Nvidia offerings, and not only on price. It doesn't matter if ATI is almost as good, it must be clearly better overall, no matter platform. The damaged image ATI has must be cleared, but that can't happen until their driver does what it's supposed to (work reliably like their main competitor). I said I would give ATI another chance, because I truly respect them opening up stuff and dedicating resources on the open source development, and I'm a long time FLOSS advocate, but I still expect at least the same experience, preferably better, than Nvidia. I can't recommend hardware that are almost there...
Originally posted by bridgman
Numasan mentioned that his last experience with ATI workstation products was over 18 months ago, with the old driver stack we picked up when we aquired FireGL.
Yes, but the damage to your image is done. How many years was it since you aquired FireGL?
Just to clarify, the FireGL boards I have experience with, were from 2006-2007 (the series number escaped me). Surely you didn't use IBM GPU's then? But the codebase was the same? What you did with the new OpenGL driver and the new GPU architecture, was to reboot your whole effort in the workstation area. Frankly, you should have done this as soon as you changed from IBM to your own design. Now you face the troubles of convincing DCC users that this time you really mean it! We are in 2009 now, this situation should be a non issue, not a potential risk.
I think I will buy an RV7xx now (not FirePro yet), in a couple of months, to see if things are really better. Personally I think you should concentrate all your efforts into the open source driver, make it the best OpenGL performer out there with Linux stability. I don't care about DRM (nobody I know in the VFX business cares about digital restriction management. We make graphics, we don't deliver it). I just want to take advantage of the awesome hardware you made. I will give it a shot until the end of this year. If it is not better than Nvidia by then, I'm afraid the boat has sailed...
I don't know for sure, but my guess is that the benefit is much less with workstation-type apps. Consumer apps tend to be pixel-shader limited (so multiple cards can make a big difference even if the vertex processing is duplicated) but workstation apps are as likely to be vertex-shader limited and current multi-GPU implementations (ours or competitors) wouldn't help much there.
For workstation use there seemed to be more interest in things like frame-lock than Crossfire.
I re-read Michael's article and didn't see anything about crashes with >1 card in the system; is that from another test ?