MATHEMATICS, SIMULATION AND MODELLING
11 July 2011
Intended to bridge the gap between prototyping and production phases for system engineering, Wild Cruncher, a combination of hardware and software, has been released by HPC Project. Numerical computation environments like Scilab are flexible and easy to use for system modelling, however high data volume processing or deployment of the model once it has been validated can lead to a loss of performance. A complete rewriting of the model in a more performing language like C is then necessary. As this is time consuming and creates a control gap between the engineer and the programmer, this rewriting phase is often skipped, except for critical projects.
Wild Cruncher automates this rewriting and once the model is validated, a ‘click’ will generate the C version. Another ‘click’ will generate a parallel version in order to take advantage of multi-core or hybrid hardware architectures. All these phases, including the compiled code execution, take place in a single working homogeneous environment. The Scilab to C translation software has been developed by HPC Project, while the parallelisation, code generation and optimised compilation for the associated hardware are performed with open source software, Par4All. The hardware is built with technologies from Intel and Nvidia embedded in an office environment compatible device.