Download PDF >

Accelerate MATLAB® code 10-100X faster computation 10-100X larger data sets No re-programming in C/C++/Fortran/MPI

Star-P is a new programming paradigm that enables desktop tools - such as the Python, MATLAB® from The MathWorks, R and others - to be run on accelerated multi-core systems for the development, testing and deployment of parallel algorithms. Servers and/or clusters that are equipped with ClearSpeed Advance™ accelerators can now be programmed using the Star-P parallel programming platform. Applications with underlying math functions such as large matrix multiplies benefit from ClearSpeed's floating point acceleration, and with Star-P, additional functionalities and applications can run in parallel on scalar systems.

Target Markets
Financial Services, Universities and National Laboratories, Life Sciences, Healthcare, and Manufacturing enterprises have 100's and even 1000's of engineers, scientists and analysts that use desktop tools, such as MATLAB to develop mathematical and statistical models for image processing, signal processing, financial quantitative analysis, and other modeling applications. IT and HPC executives buy large scale systems and accelerators based on applications that need the capacity and performance of these systems.

The Star-P Solution: Adding Software Capabilities for ClearSpeed Advance Accelerators
ClearSpeed acceleration can be realized through plug and play with standard math libraries and tools such as MATLAB, or by programming with ClearSpeed's software development tools. Star-P enhances this dynamic by offering a parallel programming platform, combining accelerator technology, that the large majority of engineers, scientists and analysts within high performance computing (HPC) can use to build custom applications without having to know all of the technical details of low level programming techniques required in C, C++, Fortran, Message Passing interface (MPI) and others.

Value Proposition: Accelerate Time-to-Solution
Reducing the time to develop large and complex algorithms from years to months, improving performance of existing models and algorithms written in MATLAB, and enabling engineers, scientists and analysts to benefit from ClearSpeed accelerated multi-core systems on a much larger scale than are possible on desktops - via Star-P & ClearSpeed Accelerator boards- are key value statements for the High Performance Computing community.