HPC programming (using C, C++, FORTRAN, MPI) is a highly specialized expertise and only the large and well funded projects can justify modeling and simulation to be done by HPC programming and running the fully HPC-coded models on dedicated or shared supercomputers. Many R&D organizations today don’t have any HPC programmers on their teams at all. However, there is a great value in highly-specialized, critical portions of the model to be written in C, Fortran, and MPI.

Also, there is a great value in previously written, well tested, and time-proven HPC code if it can be easily re-used in new simulation models programmed in Very High Level Languages such as MATLAB®, Python, or R.

Star-P enables HPC programmers to:

• Increase the value and usability of their low-level programs (written for example in C++ and MPI) for the R&D team
• Dramatically simplifies the combination of a VHLL-written top-level algoritm with the sophisticated and optimized low-level HPC code, to create a complete simulation model
• Eliminate the need to rewrite the high-level algorithm logic in low-level code. This is especially useful under time-pressure, as the VHLL+Star-P application can quickly be used in production while the port to low-level code takes place.
• Speed up the interaction with the VHLL programmer or science/engineering domain expert to dramatically shorten the time to solution
• All that while using the familiar HPC programming environments, plus a simple set of Star-P Connect tools

View Larger Image >