Title :
Concord: re-thinking the division of labor in a distributed shared memory system
Author_Institution :
Dept. of Comput. Sci. & Eng., Washington Univ., Seattle, WA, USA
Abstract :
A distributed shared memory system provides the abstraction of a shared address space on either a network of workstations or a distributed-memory multiprocessor. Although a distributed shared memory system can improve performance by relating the memory consistency model and maintaining memory coherence at a granularity specified by the programmer, the challenge is to offer ease of programming while maintaining high performance. Concord meets this challenge by carefully splitting responsibilities among the programmer, the compiler, and the runtime system. Concord has allowed a single programmer to port several real, large shared-memory parallel programs onto an Intel iPS/2 in a few weeks and achieve reasonable speedup
Keywords :
distributed memory systems; parallel programming; shared memory systems; storage management; Concord; Intel iPS/2; abstraction; distributed shared memory system; distributed-memory multiprocessor; division of labor; granularity; large shared-memory parallel programs; memory coherence; memory consistency model; shared address space; Coherence; Computer science; Debugging; Maintenance engineering; Program processors; Programming profession; Runtime; Scattering; Workstations; Yarn;
Conference_Titel :
Scalable High-Performance Computing Conference, 1994., Proceedings of the
Conference_Location :
Knoxville, TN
Print_ISBN :
0-8186-5680-8
DOI :
10.1109/SHPCC.1994.296694