Mathematica - a Perfect Test Environment for OpenCL Code

By releasing UnRisk-Q we have unleashed the programming power behind all UnRisk solutions. It comes with a Mathematica front-end.

With our newUnRisk we will go far beyond this quant development environment. It will be
  • Multi Language
  • Platform Agnostic
  • Inherently Parallel
  • Cross Platform
A variety of applications in quant finance are computationally extremely demanding - if you for example introduce Montecarlo VaR or CVA/DVA you may need hundred of millions of single instrument valuations. newUnRisk will intelligently combine coarse and fine grain parallelization driving OpenCL over the Grid.

The concept is described here.

Our OpenCL code ist tested in Mathematica - the programs support of OpenCL (CUDA) makes Mathematica a highly efficient testing platform for algorithms and their execution on either a CPU or GPU.
In the picture above you see what Mathematica does for you - only the kernel code (in the picture CUDA but it is the same with OpenCL) remains your responsibility.