40 likes | 151 Vues
This summary provides insights into the AMMPI implementation of MPI version 1.1, highlighting its trivial porting capabilities across various platforms. AMMPI is known for delivering high performance, thanks to vendor-specific tuning, and is compatible with systems like Linux/Myrinet, MPICH, IBM SP3, and Cray T3E. Built on the AMUDP codebase, it features robust error reporting, non-blocking communication, and a performance monitoring API. This overview underscores its strengths and considerations regarding latency and bandwidth performance.
E N D
AMMPI - Summary • Active Messages–2 (AM) implementation over MPI version 1.1 • Porting is trivial - works on virtually any platform that has MPI 1.1 • Often provides very high performance – vendors tune their MPI well • Linux/Myrinet, MPICH, IBM SP3, Origin 2000, Cray T3E, many others… • Based on the AMUDP code base, same cool features • Robust, clear error reporting for ease of debugging • SPMD bootstrapping library (but we use site-specific mpirun) • Network performance/utilization monitoring API • MPI Interface • Non-blocking sends, non-blocking receives • Uses MPI communicators to co-exist happily with other MPI-aware layers
Good Bad AMMPI – Latency Performance minimal small message, round-trip time measured from application
Bad Good AMMPI – Bandwidth Performance with 64 KB messages == MAX_MEDIUM == MAX_LONG