Introduction to Parallel Programming with MPI（介绍与MPI并行编程）.pdfVIP

Introduction to Parallel Introduction to Parallel Programming with MPI Programming with MPI PICASso Tutorial October 22-23, 2007 Stéphane Ethier (ethier@) Computational Plasma Physics Group Princeton Plasma Physics Lab Why Parallel Computing? • Want to speed up a calculation. • Solution: – Split the work between several processors. • How? – It depends on the type of parallel computer • Shared memory (usually thread-based) • Distributed memory (process-based) – MPI works on all of them! Shared memory parallelism CPU CPU CPU CPU CPU CPU CPU CPU 0 1 2 3 4 5 6 7 L1 L1 L1 L1 L1 L1 L1 L1 L2 L2 L2 L2 L2 L2 L2 L2 BUS MEMORY • “Classic” shared memory node • Renewed interest due to multi-core chips • Processors communicate through memory Shared memory parallelism • Program runs inside a single process • Several “execution threads” are created within that process and work is split between them. • The threads run on different processors. • All threads have access to the shared data through shared memory access. • Must be careful not the have threads overwrite each other’s data. Shared memory programming • Easy to do loop-level parallelism. Main Process • Compiler-based automatic parallelization threads serial work – Easy but not effective start parallel