Promoting Polyhedral Compilation
Polly Labs is a multinational collaboration between academia and industry. Its mission is to advance research and software around the Polyhedral Model.
Polyhedral compilation is a powerful technique to automatically optimize loop nests which often occur in scientific and multimedia code. Optimizations include parallelization (OpenMP, OpenCL, CUDA, MPI, ...), vectorization, tiling, loop fusion, loop reversal, loop interchange and many more.
Polly Labs contributes software to the community itself, such as:
Polly Labs 2018 Projects
Collect Polyhedral BenchmarksCurrently, Polly’s performance is measured on PolyBench benchmark. PolyBench does not cover all of the possible kernels so we may miss out some...
Compiling Chapel with Polly/LLVMChapel is an emerging parallel programming language developed with the aim of providing better performance in High-Performance Computing as well as...
Metamorphic Testing for Math LibrariesRecently, metamorphic testing techniques have been applied with success in various fields to find domain-specific bugs, as presented in this survey....