Synchronizing Operations on Multiple Objects

Parallel programming on distributed memory systems is one of the most challenging research areas in parallel computingtoday. Objectbased parallel programming languages are an important class of languages for such systems. Shared objects allow the programmer to deal with data partitioning, communication, and synchronization in a highlevel manner. Synchronizing operations on a single object is well understood. Dealing with synchronization on multiple objects distributed over the available processors, however, is still an open issue. In this paper, we sent an abstraction, called weavers, that is used to synchronize operations on multiple objects, and show how weavers are applied in a runtime support system for atomic functions on multiple objects.

• Type: Conference
• Year: 1998
• Where: IPPS
• Authors: Tim Rühl, Henri E. Bal
• Keywords: Distributed And Parallel Computing, IPPS 1998, Multiple Objects, Parallel Programming, Parallel Programming Languages