PDA

View Full Version : SUPERCOMPUTING on CORE??????????



silviotoledo
03-03-2009, 07:57 AM
Someone said me NVIDIA now have a computer that costs around only $5K that process 60 Teraflops GPU based.

This means this computer is 1621 times quickier than an Intel Quad Core processor.

NVIDIA also have Ageia Physyx incorporated ( it was a plugin for lightwave in the past ) that have all the COOL dynamics we need in REALTIME.

http://www.youtube.com/watch?v=gMQDPLcqt8s&feature=related
http://www.youtube.com/watch?v=v8LpC0rrUC8
http://www.youtube.com/watch?v=o07KjfBx8GE&feature=related
http://www.youtube.com/watch?v=VuPN800KS6Q&feature=related

So, will Lightwave Core have some kind of partner development with NVIDIA?

Will we need to wait 1 year ( the core time ) to get this technology?

How difficult is to port lightwave to CUDA ?

How much will open CL help in the process?

May someone develop a plugin to export the lightwave solution for CUDA for rendering?

Too much power without use.

praa
03-03-2009, 08:04 AM
they should get seious about GPU computing... but in an open way not with CUDA but with openCL

i think they understand the potential of this but i also think they have very limited resources... as an underdog of the 3d industry you know :)

praa
03-03-2009, 08:09 AM
here is a thread about that...

http://www.newtek.com/forums/showthread.php?t=91652

mattclary
03-03-2009, 09:48 AM
( it was a plugin for lightwave in the past )

WTF are you talking about?

Was there really a Physx plugin for LightWave? I think the forum would have melted from the discussion were that actually true.

walfridson
03-03-2009, 10:08 AM
Think he means Dstorm liquid

iluxa
03-03-2009, 10:10 AM
Has trial LiquidPack,from DStorm, AGEIA Tech. drivers in installation pack?
I think, it possible!

walfridson was 1st ;)

silviotoledo
03-03-2009, 02:15 PM
MENTAL RAY will be available on this system soon.

the TESLA is 250 times more quick in processing than a quadricore intel.


We really need Lightwave working with mental ray

lots
03-04-2009, 07:40 PM
I wouldn't say its 250 times faster in EVERY situation. You have to keep in mind that not all computing benefits from the "stream" processors on the GPU the same way graphics do.

Get some code with alot of branching, and that 60 TFlop GPGPU package won't do you much good.