CULA R12 (CUDA 4.0 support) now available GPU Accelerated Linear Algebra Package

EM Photonics and the CULA team are very pleased to announce that CULA R12, based on CUDA 4.0, is available immediately at our downloads page.

Besides CUDA 4.0 support, this release also introduces the new link-compatible interface, which allows for zero-effort porting of existing codes which use LAPACK routines. For more information, please see the documentation included or read our blog post about the feature.

We just added a blog post with information on using the new link interface in MATLAB.

To utilize this new interface in MATLAB you simply have to change a few environment variables – there are no MEX files to compile, no clunky gpuArray objects, and no changes MATLAB function names!

For example, the MATLAB commands:

>> tic; A = A*A'; toc;

Elapsed time is 3.414187 seconds.

>> tic; [q,r] = qr(B); toc;

Elapsed time is 11.318329 seconds.

Will be accelerated with zero effort from the user!