Simulating asynchronous multiple-loop networks is commonly considered a difficult task for parallel programming. Two examples of asynchronous multiple-loop networks are presented in this article: a stylized queuing system and an Ising model. In both cases, the network is an n × n grid on a torus and includes at least an order of n2 feedback loops. A new distributed simulation algorithm is demonstrated on these two examples. The algorithm combines three elements: (1) the bounded lag restriction; (2) minimum propagation delays; and (3) the so-called opaque periods. We prove that if N processing elements (PEs) execute the algorithm in parallel and the simulated system exhibits sufficient density of events, then, on average, processing one event would require O(log N) instructions of one PE. Experiments on a shared memory MIMD bus computer (Sequent's Balance) and on a SIMD computer (Connection Machine) show speed-ups greater than 16 on 25 PEs of a Balance and greater than1900 on 214 PEs of a Connection Machine.
The full text of this article is premium content
No entries found
Log in to Read the Full Article
Please select one of the options below for access to premium content and features.
Create a Web Account
If you are already an ACM member, Communications subscriber, or Digital Library subscriber, please set up a web account to access premium content on this site.
Join the ACM
Become a member to take full advantage of ACM's outstanding computing information resources, networking opportunities, and other benefits.
Subscribe to Communications of the ACM Magazine
Get full access to 50+ years of CACM content and receive the print version of the magazine monthly.
Purchase the Article
Non-members can purchase this article or a copy of the magazine in which it appears.