What would you do with this system?

[fafster]

I have a C2D motherboard, with processor and memory. The motherboard supports SLI, and handily I have two nVidia 8800GT cards I'm using to catch dust.

 

What would you do with a CUDA-enabled system? I want to make use of the system, and ideally I'd like to take advantage of both cards simultaneously.

 

I'm after ideas. Any suggestions?

[featured_spectre]

is it school based or home based?

[fafster]

Home based, but I'm open to having it in the school if it's interesting enough.

[featured_spectre]

Whats the full spec?

[dhicks]

I'm after ideas. Any suggestions?

 

Applications that I can see benifitting from a paralell processing approach to computation and being useful in a school context are video transcoding and speech recognition, both aimed at server-based applications. Both applications are currently tricky to get right in a school, requiring higher-end workstations and lots of audio / video data to be moved around. A server-based video editing solution could transcode all video given to it in to a format suitible for display in a web browser (Flash, or whatever) and a video editing application could be built around that. Speech recognition could, hopefully, be sped up to the point where it is actually useful to be used with special needs pupils.

 

Video editing would seem to be more practical to start with, requiring the programming of a web app rather than figuring out a new approach to speech recognition. I'd guess that an install of FFMPEG should take care of most of your video transcoding needs, although obviously you might need to do some adjustments to get it to use your graphics cards for acceleration.

 

--

David Hicks

Edited March 23, 2010 by dhicks

[fafster]

Full spec:

 

Core 2 Duo E6550 @ 2.33GHz

4GB DDR2 @ 1066MHz

ASUS motherboard with nForce 650 chipset

2x nVidia 8800GT

 

I've found a video CUDA-enabled transcoder in MediaCoder - more than a universal audio/video transcoder - MediaCoder official website

 

The latest CUDA drivers from nVidia apparently support SLI. Whether that means distributing the load across the cards, or simply switching off SLI in software to only function on one card, I'm not sure.

 

Folding@Home now supports CUDA, but whilst a worthy cause may not be most useful in a school (I could always run it overnight). For now, the video conversion sounds like the best bet as it could prove most useful.

[pete]

Rainbow tables / password cracking demo? (use the gpu for the processing).

[fafster]

Rainbow tables / password cracking demo? (use the gpu for the processing).

 

That definitely sounds like a cool idea. Any apps you know of to take advantage?