Open Access
The Beta Operation: A Parallel Primitive
Evan R. Cohn
- 01 Nov 1988
3
TL;DR: The beta operation is examined, its efficiency is demonstrated, it is generalized in several directions, and its suitability for general use is shown.
read more
Abstract: : The ever-decreasing cost of computer processors has created a great interest in multiprocessor computers. However, along the increased power that this parallelism brings, comes increased complexity in programming. One approach to lessening this complexity is to provide the programmer with general purpose parallel primitives that shield him from the structure of the underlying machine. There are two contending goals that must be satisfied when designing primitives. On one hand we would like to make the primitives as general and abstract as possible. The more general a primitive is, the more easily it can be used as a building block for creating complex algorithmic constructs. On the other hand, we want to avoid making the primitive so general that it becomes inefficient. In The Connection Machine, Hillis suggests that the beta operation as a parallel primitives for his hypercube-based machine. We shall examine the beta operation, demonstrate its efficiency, generalize it in several directions, and show its suitability for general use.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Implementing the multiprefix operation on parallel and vector computers
Thomas J. Sheffler
- 01 Aug 1993
TL;DR: This paper presents a work efficient algorithm for the multiprefix operation on n elements that runs in __ __ S = O(V n parallel steps on a p = V n processor CRCW-ARB PRAM) and shows that by using the multip refix operator for sparse-matrix dense-vector multiplication, it is obtained performance exceding traditional FORTRAN-based approaches.
8
Multireduce and Multiscan on Modern GPUs
Marco Eilers
- 01 Jan 2014
TL;DR: This thesis proposes an algorithm which, despite its generality, performs at least as well as the best published histogramming algorithm for all inputs and provides a 18× speedup over the CPU algorithm for small numbers of buckets, making this solution the fastest existing histogram algorithm for GPUs.
4
Implementing the multiprefix operation efficiently
TL;DR: This paper presents an efficient deterministic implementation of the multiprefix operation for the hypercube and shows that this implementation can be modified to create an implementation for the multipRefix operation without increasing the use of any of the simple primitives by more than a constant factor.
3
References
Sorting networks and their applications
Kenneth E. Batcher
- 30 Apr 1968
TL;DR: To achieve high throughput rates today's computers perform several operations simultaneously; not only are I/O operations performed concurrently with computing, but also, in multiprocessors, several computing operations are done concurrently.
Scans as primitive parallel operations
TL;DR: A study of the effects of adding two scan primitives as unit-time primitives to PRAM (parallel random access machine) models is presented and it is shown that the primitives improve the asymptotic running time of many algorithms by an O(log n) factor, greatly simplifying the description of many technologies.
•Proceedings Article
Scans as Primitive Parallel Operations.
Guy E. Blelloch
- 01 Jan 1987
TL;DR: In this article, a study of the effects of adding two scan primitives as unit-time primitives to PRAM (parallel random access machine) models is presented, and it is shown that the primitives improve the asymptotic running time of many algorithms by an O(log n) factor.
353
•Proceedings Article
Routing, Merging and Sorting on Parallel Models of Computation (Extended Abstract)
Allan Borodin,John E. Hopcroft +1 more
- 01 Jan 1982
TL;DR: A variety of models have been proposed for the study of synchronous parallel computation.
289