The High Performance Open Community Runtime (P-OCR)
As new clusters and architectures are becoming more common across the sciences, plus the resurgence of heterogeneity in both HPC and other computing areas, the software stack becomes center stage on how to exploit the best performance out of the vast range of architectures. In adition to the vast availability of computational substrates, their ever changing conditions put extra requirements on the software stack designers. This requires intelligent, elastic and high performance software stacks. With this is mind, we introduced the Performance Open Community Runtime (P-OCR) which is a flavor of the OCR runtime that provides the elasticity of the ideas presented under OCR plus a high performance implementation and support for other areas of computing (such as Big Data computation). Its agility is derived from P-OCR introspective capabilities and the internal support to reduce data movement with selected compression and symbolic execution when required.
This webpage is the repository for the P-OCR source code. To download the current version, please follow the link below. For any comments, questions or suggestions, send an email to Joshua Suetterlein