Re: [ublas] ublas and nvidia cuda

Subject: Re: [ublas] ublas and nvidia cuda
From: Rutger ter Borg (rutger_at_[hidden])
Date: 2010-05-10 03:31:51

Tim Odenthal wrote:

> Hi!
> 
> Has anybody out there already played around with the (relatively) new CUDA
> Toolkit
> (http://news.developer.nvidia.com/2010/03/nvidia-releases-new-cuda-
> toolkit.html) and ublas? Does it support "enough" C++ to compile code
> which uses ublases (sparse) matrix types? I gather there is a big issue
> with virtual functions...
> 
> Does anybody have example code which uses ublas' types and runs on a GPGPU
> (compiling with nvcc) ?
> 
> Are there (official) plans to exploit GPGPU computing power for LA with
> ublas? (Bindings?)
> 

Hello Tim,

at the moment, the bindings support NVidia's CUBLAS library through the BLAS 
bindings. Define BOOST_NUMERIC_BINDINGS_BLAS_CUBLAS to enable it.

What is lacking (at least, I couldn't easily find it) are high-quality GPU-
based containers. Preferably ones that enable the writing of algorithms that 
minimize their memory traffic from/to GPU memory. (Isn't this a nice 
showcase of why we have C++ allocators?)

Cheers,

Rutger