Adaptive granularity : Transparent integration of fine- and coarse-grain communication

• Published in: International journal of parallel programming Vol. 25; no. 5; pp. 419 - 446
• Main Authors: PARK, D, SAAVEDRA, R. H, MOON, S
• Format: Journal Article
• Language: English
• Published: New York, NY Plenum Press 01-10-1997; Springer Nature B.V
• Subjects: Applied sciences; Communication; Computer architecture; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Data processing. List processing. Character string processing; Exact sciences and technology; Memory organisation. Data processing; Software; Software engineering; Studies; Compiler; Distributed memory multiprocessor system; Computer simulation; Storage management; Data transmission; Replication; Adaptive granularity; Performance analysis; Program analysis; Resource management; Software engineering; Execution time
• ISSN: 0885-7458; 1573-7640
• DOI: 10.1007/BF02699885

The granularity of shared data is one of the key factors affecting the performance of distributed shared memory machines (DSM). Given that programs exhibit quite different sharing patterns, providing only one or tow fixed granularities cannot result in an efficient use of resources. On the other hand, supporting arbitrarily granularity sizes significantly increases not only hardware complexity but software overhead as well. Furthermore, the efficient use of arbitrary granularities put the burden on users to provide information about program behavior to compilers and runtime systems. These kind of requirements tend to restrict the programmability of the shared memory model. A new connections scheme is presented, called Adaptive Granularity (AG). Adaptive Granularity makes it possible to transparently integrate bulk transfer into the shared memory model by supporting variable-size granularity and memory replication. It consists of 2 protocols: one for small data and another for large data.