MathGroup Archive 2009

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: CUDA and Mathematica


Hello,

There is indeed, but it's not easy. Right now you'll have to write a 
MathLink program in C using the NVIDIA CUDA SDK, which obviously 
requires knowledge of C. Last I talked to R&D, Wolfram and NVIDIA are 
currently working together to incorporate functions in Mathematica that 
make use of CUDA/GPU processing, but that won't be here until version 8 
at least.

-Zach

On 8/13/2009 3:22 AM, Filippo Miatto wrote:
> Dear all,
>
> I use a desktop pc with an nvidia graphic card, cuda enabled.
> is there a way to exploit the capabilities of parallel processing of
> cuda technology, as it is already possible with mathlab for example?
> apparently the benefits are really big, especially if one is using
> more graphic cards togehter, or some (expensive) particular models.
> Filippo
>
> ------------------------------------------------------------
>
> Via Scossore, 94
> 31029 Vittorio Veneto (TV)
> Italy
>
> msn:			dr.ziofil at hotmail.com
> Quantum Optics
> Group Mail:		miatto at molphys.leidenuniv.nl
>
>
>    


  • Prev by Date: Re: Recognizing finite groups in Mathematica 7
  • Next by Date: ParrallelDo and set::noval
  • Previous by thread: Re: Re: CUDA and Mathematica
  • Next by thread: DisplayFormulaNumbered AND "Equal Symbol Aligned"