nMP (D700 or 500 specifically) and MatLab

[Boomhowler]

Hello!

Has anyone of you with the new D700/500 cards access to MatLab with the GPU/parallel processing toolbox?

If so, could you run the "paralleldemo_gpu_backslash(2)" benchmark included in that toolbox? Maybe post a figure or two (or some of the numbers) here?

// DL

 

[koban4max]

Hello!

Has anyone of you with the new D700/500 cards access to MatLab with the GPU/parallel processing toolbox?

If so, could you run the "paralleldemo_gpu_backslash(2)" benchmark included in that toolbox? Maybe post a figure or two (or some of the numbers) here?

// DL

the question is: could you?

 

[Boomhowler]

Ah, sorry for the mis-use of "could", I meant "can". As in: Can you run this and show the results here.

I have MatLab but no access to the hardware. I have run it on a 3.1 MP with a GTX 680 and it would be cool to see if MatLab on OSX can use both the GPU:s and at which ... darn it, just realized that MatLab only uses Cuda for the GPU-stuff now. Bah, have to write to them again

 

[ticotoo]

Need CUDA

Can not do it on those GPU's not CUDA:

"Speeding Up MATLAB Computations with GPUs (R2013b)

Parallel Computing Toolbox provides GPUArray, a special array type with several associated functions that lets you perform computations on CUDA-enabled NVIDIA GPUs directly from MATLAB. " I don't know about other toolbox's use of GPU in OpenGl, but its not listed.

My 680 GTX, hex 3.33 5,1:

 

[Boomhowler]

Can not do it on those GPU's not CUDA:

"Speeding Up MATLAB Computations with GPUs (R2013b)

Parallel Computing Toolbox provides GPUArray, a special array type with several associated functions that lets you perform computations on CUDA-enabled NVIDIA GPUs directly from MATLAB. " I don't know about other toolbox's use of GPU in OpenGl, but its not listed.

My 680 GTX, hex 3.33 5,1:

Yeah, I wrote that above, that it's only Cuda. I wrote to MatLab before, asking for OpenCL support. Guess I'll have to write again