WORKSHOP "HPC UD ‘17”
WORKSHOP ON HYBRID MPI/OPENMP PROGRAMMING ON INTEL PLATFORMS
In the last years, novel computational architectures designed for high performance computing have been deployed as heterogeneous parallel computers, composed by processing nodes with several accelerators or coprocessors large vector register capacity to speed up scientific applications and also heterogeneous memory systems. The objective of this tutorial is teaching how to exploit MultiProcessing, multithreading and vectorization on multicore and manycore architectures, using OpenMP and profiling tools to perform top-down analysis of parallelization.
Todas las fechas
- 20 de Septiembre de 2017 02:00 PM - 05:30 PM
Powered by iCagenda