new nVidia graphics processor type

[frogspasm]

http://gizmodo.com/gadgets/spectacular-techno_porn/nvidias-tesla-hardware-supercomputers-fo-reals-270650.php

[meagane]

;--------------)

[Koogle]

lol Holy crap.. that does look bad ass.. any cgtalk competitions coming up? :D

http://gizmodo.com/assets/images/gallery/4/2007/06/medium_576535755_c930c87af8_o.jpg

[aglick]

Tesla wil not have any graphics output. -No VGA, DVI or HDMI. Only an Ethernet jack.

This platform is for GPGPU applications and not realtime graphics rendering...

Adam
BOXXlabs

[UrbanFuturistic]

It won't do realtime raytracing then? Meh, I've lost interest :p

Seriously though, it's gotta be worth something for production rendering.

It's like the spirit of SGI is back from the dead.

[darkZ]

Just when I was thinking that this is as large as a video card can get ? :D

[Oddgit]

I think you would lek the Quadro Plex, those are the crazy big external video processors. They even have one based on 2 Quadro 5600s, i wonder how many FPS doom would get..... ;)

[Koogle]

It is a shame that such an awesome looking piece of kit has such limited uses for the majority of people. I suppose even it did work with your windows/linux whatever... and software(maya) it would have sucky drivers.

It does remind me of Asus XG Station:

http://thg.ru/technews/images/xg_station-080107.jpg

Definetly the way forward for high end workstation graphics cards, they are just getting too big for the pc cases and are better suited as an external unit with there own power supply etc

[revilo3D]

would gelato be able to tap into this power?

[enygma]

It is a shame that such an awesome looking piece of kit has such limited uses for the majority of people. I suppose even it did work with your windows/linux whatever... and software(maya) it would have sucky drivers.
Drivers aren't the big issue as Tesla is essentially an 8800GTX with the video outs ripped off of them. CUDA would most likely be the development environment which is currently at version 0.9 and working quite well.

The G80 is essentially a stream processor anyways, and doesn't even do data to video on the GPU any more. If you take the HSF off a G80 based video card, you will notice that there is an extra chip on there which happens to be used to interpret processed data from the G80 and make it displayable on screen.

The only advantage to Tesla I see is if they enable the WGL_NV_gpu_affinity extension in OpenGL for it, so that multiple GPUs for computing can be used much easier in Windows.

[ndat]

I thought the G92 architecture was coming out around then as well?

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.