Drexel dragonThe Math ForumDonate to the Math Forum



Search All of the Math Forum:

Views expressed in these public forums are not endorsed by Drexel University or The Math Forum.


Math Forum » Discussions » Software » comp.soft-sys.matlab

Topic: Using Cuda on a Server
Replies: 1   Last Post: Jan 22, 2013 8:06 AM

Advanced Search

Back to Topic List Back to Topic List Jump to Tree View Jump to Tree View  
Edric Ellis

Posts: 697
Registered: 12/7/04
Re: Using Cuda on a Server
Posted: Jan 22, 2013 8:06 AM
  Click to see the message monospaced in plain text Plain Text   Click to reply to this topic Reply

"Alafm " <abdoulaye-gory@edu.em-lyon.com> writes:

> I have a question regarding the use of Cuda and Matlab. I have a GPU
> (Tesla M2075) on a server (windows server 2008). I use another
> computer to develop some matlab applications and I would like to take
> advantage of the computer capabilities to develop some parts of my
> application at a lower level language (CUDA). I am able to use the
> Parallel Computing Toolbox from my computer and my aim now is to
> compile and develop some cuda applications from my computer. Is there
> any means to use Cuda without using such software like tigrisVCN ? For
> some safety policies I can't use directly the computer with the Tesla
> M2075. I can't access the CUDA toolkit from my computer for now.


Do you have MATLAB Distributed Computing Server installed on the
'server' machine? (http://www.mathworks.com/products/distriben/) If so,
you could use MATLABPOOL and SPMD to do this - if you are able to open a
matlabpool using the GPU server machine as a worker, the body of the
SPMD block can execute CUDA code using for example the CUDAKernel
interface (assuming that's what you're interested in, and that gpuArray
capabilities are not sufficient).

I must say though that you will almost certainly find life simpler if
you can develop the kernel locally on your desktop machine. The CUDA
toolchain can be downloaded without charge from NVIDIA
(https://developer.nvidia.com/cuda-downloads) and can run even on a
machine with no GPU (I must admit I haven't tried with the newer
releases where the toolkit and drivers are bundled together). Obviously,
if you have a capable GPU on your desktop (even if a poorly performing
one), you'll be able to develop and debug stuff there.

Cheers,

Edric.



Point your RSS reader here for a feed of the latest messages in this topic.

[Privacy Policy] [Terms of Use]

© Drexel University 1994-2014. All Rights Reserved.
The Math Forum is a research and educational enterprise of the Drexel University School of Education.