Computational models for parallel computers

Abstract

Computational models define the usage patterns of a computer. They can be used to derive the architecture of the machine, provide guidelines for programming tools, and suggest how the machine should be used in applications. Identifying computational models is especially important for parallel computers, since their architectures and usages are still not well understood in general. This paper describes a number of computational models for parallel computers. These models characterize the communication patterns under which processors exchange their intermediate results during computation. Emphases are placed upon models for one-dimensional processor arrays, reflecting Carnegie Mellon's experiences with the Warp systolic array machine. These models include local computation, domain partition, pipeline, multifunction pipeline and ring.

<title>Real-time construction and rendering of three-dimensional occupancy maps</title>

“…Much of the motivation for its development was derived from issues in robotics [11], so the machine has a variety of attractive features, not the least of which is the ability to combine different models of computation [12J as appropriate in single applications [13]. Since this was our first exposure to the iWarp, we implemented the algorithms using simple strategies.…”

COLTHPF, a run-time support for the high-level co-ordination of HPF tasks

“…Many choices regarding resource allocation, such as the mapping of the various tasks on the machine and the degree of parallelism of each task, might be decided by the compiler tool on the basis of either suitable performance models or directives supplied by the programmer. As an example of this high-level approach, below we show how COLT HPF can be exploited by a high-level, template-based co-ordination language that facilitates programmers in organizing HPF tasks according to pipelines and processor farm forms of task parallelism [7,8]. We present templates which implement these forms of task parallelism and we discuss their instantiation by means of a simple compiler.…”

Model programs for computational science: A programming methodology for multicomputers

“…This time I chose matrix multiplication, which can be pipelined in a straightforward way as shown in Figure 2 [37].…”