11 May 2010
Nema Labs has introduced FASThread, a parallel programming tool that enables easy and reliable parallelisation of code running on multicore platforms.
Most of today’s computers have multiple processor cores, but a majority of all software applications can only make use of a single processor. FASThread integrates into IDEs (Integrated Development Environments) and provides programmers the tools to get maximum performance from code, processors and computers.
FASThread from Nema Labs automatically optimises sequential source code for use with multicore processors: it profiles the source code to find performance bottlenecks, applies parallelisation and verifies that the parallel code gives the expected performance improvement. In addition to the automatic features, FASThread is unique in its ability to help developers locate those parts of the source code that may cause problems during parallelisation, and step by step explaining how to solve the problems. FASThread is based on advanced mathematical algorithms that preserve code correctness when moving from sequential source code to parallel executable.
The first version shipped is FASThread C/C++ for Microsoft Visual Studio and uses the OpenMP threading API. Versions supporting Eclipse/Linux will follow. Nema Labs also have plans to add Fortran support.