Programming and Optimizing Task Data-Flow Applications with OpenStream

30 October 2015 2pm at RAL (VC to DL) Antoniu Pop, University of Manchester, Programming and Optimizing Task Data-Flow Applications with OpenStream and Aftermath Dynamic task-dependent parallelism is an increasingly popular programming model on shared-memory systems. Compared to data parallel loop-based concurrency, it promises enhanced scalability, load balancing and locality. These promises, however, are undermined by non-uniform memory access (NUMA). We show that it is possible to preserve the uniform hardware abstraction of contemporary task-parallel programming models, for both computing and memory resources, while achieving near-optimal data locality.

Date: 
Friday, 30 October, 2015 - 14:00 to 16:00