ForestGOMP: An Efficient OpenMP Environment for NUMA Architectures

Exploiting the full computational power of current hierarchical multiprocessor machines requires a very careful distribution of threads and data among the underlying non-uniform architecture so as to avoid remote memory access penalties. Directive-based programming languages such as OpenMP, can greatly help to perform such a distribution by providing programmers with an easy way to structure the parallelism of their application and to transmit this information to the runtime system. Our runtime, which is based on a multi-level thread scheduler combined with a NUMAaware memory manager, converts this information into scheduling hints related to threadmemory affinity issues. These hints enable dynamic load distribution guided by application structure and hardware topology, thus helping to achieve performance portability. Several experiments show that mixed solutions (migrating both threads and data) outperform workstealing based balancing strategies and next-touch-based data distribution ...

• Type: Journal
• Year: 2010
• Where: IJPP
• Authors: François Broquedis, Nathalie Furmento, Brice Goglin, Pierre-André Wacrenier, Raymond Namyst
• Keywords: Current Hierarchical Multiprocessor, Hierarchical Thread Scheduling, IJPP 2010, Memory Access Penalties