Wolfram Research, NVIDIA Collaborate to Provide Integrated GPU Programming in Mathematica

Wolfram Research has announced a technical collaboration with NVIDIA to integrate GPU acceleration into Mathematica 8.

Combining Mathematica's programming ease-of-use with the computational speed of the GPU-equipped hardware dramatically increases application performance and user productivity across industry, research, and education.

Tom Wickham-Jones, Wolfram Research's director of Kernel Technology, says, "With Mathematica, scientists and engineers can easily tap the enormous parallel processing power of GPUs through a familiar high-level interface. Thanks to Mathematica's full-featured development environment and CUDA integration, users can focus on algorithm innovation rather than spending time on repetitive tasks, such as GUI design."

NVIDIA's GPU architecture can transform Mathematica's computing, modeling, simulation, and visualization performance, boosting speed by up to a factor of 100. Mathematica's intuitive CUDA GPU programming features along with its built-in ready-to-use examples for common application areas, such as image processing, medical imaging, statistics, and finance, make these performance gains easily accessible.

"Mathematica users with GPU-enabled systems from laptops to high-end Tesla supercomputers will now be able to perform complex, data-intensive computations much more easily," says Andrew Cresci, general manager, vertical marketing solutions at NVIDIA. "Mathematica's intuitive CUDA programming interface eliminates the need to write C/C++ or FORTRAN code to take advantage of GPU computing, making Mathematica a compelling choice for anyone looking to harness GPU high-performance computing capabilities."

Mathematica's CUDA programming capabilities will be showcased at the GPU Technology Conference Sept. 20-23 in San Jose, Calif., and are the theme of session #2028 on Sept. 21 at 2 p.m. as well as Wolfram's exhibit at booth #31.