Welcome to MilkyWay@home

Extreme CUDA processing!

Message boards : Number crunching : Extreme CUDA processing!
Message board moderation

To post messages, you must log in.

AuthorMessage
pharrg

Send message
Joined: 16 Jan 09
Posts: 9
Credit: 167,133
RAC: 0
Message 8487 - Posted: 16 Jan 2009, 20:12:32 UTC

Hi. I am building a new machine that will have 3 nVidia GTX 295's running in SLI. Each board has 480 cores in parallel and almost 2Gb of dedicated RAM. Each is capable of 1.788 Teraflops of processing power. Altogether, by running 3 in SLI, this will give my machine 1440 cores and a whopping 5.364 Teraflops of power! Yes, it's like having a supercomputer sitting on your desk. But, unless your project is able to run CUDA (which the latest versions of BOINC support now), you can't make use of any of that power. It would represent a massive increase in processing power for work units. Most of the newer nVidia graphics cards support CUDA. Any plans to add CUDA support for your project?
ID: 8487 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile speedimic
Avatar

Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0
Message 8500 - Posted: 16 Jan 2009, 22:35:47 UTC - in response to Message 8487.  

AFAIK Cuda is not top of the ToDo-list here.

Have fun CUDAing over at Seti... LOL
(sorry, could not resist.)

mic.


ID: 8500 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cluster Physik

Send message
Joined: 26 Jul 08
Posts: 627
Credit: 94,940,203
RAC: 0
Message 8508 - Posted: 17 Jan 2009, 3:11:29 UTC - in response to Message 8487.  
Last modified: 17 Jan 2009, 3:12:33 UTC

If I had some time I would create a GPU app for MW (a very early version runs already). But it would be preferably for ATI cards as they have about 2.5 times the power of nvidia cards when it comes to double precision (as required by MW). A GT200 wouldn't be faster than a typical quadcore (which is unsatisfying for a GPU in my opinion), but a RV770 would add some significant juice.
Just as a cornerstone, the goal would be about 30s for a WU with a highly optimized GPU app for a HD4870.

PS:
Just for your information, to arrive at the peak power with double precision, divide the advertised flops numbers of your nvidia card by 12 and of your ATI card by 5.
ID: 8508 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Honza

Send message
Joined: 28 Aug 07
Posts: 31
Credit: 86,152,236
RAC: 0
Message 8511 - Posted: 17 Jan 2009, 8:26:13 UTC

Thanks for your insight, Cluster Physik.

@speedimic - running GPUs on SETI is good for heating, perhaps better that Intel's Presshot CPUs...considering performance per watt of oficial GPU app comparing to opt app running on Core i7.
BOINC Project specifications and hardware requirements
ID: 8511 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : Extreme CUDA processing!

©2024 Astroinformatics Group