Approaches for integrating task and data parallelism

Abstract

Languages that support both task and data parallelism are highly general and can exploit both forms of parallelism within a single application. However, integrating the two forms of parallelism cleanly and within a coherent programming model is di cult. This paper describes four languages (Fx, Opus, Orca, and Braid) that try to achieve such an integration and identi es several problems. The main problems are how to support both SPMD and MIMD style programs, how to organize the address space of a parallel program, and how to design the integrated model such that it can be implemented e ciently.

Endpoint Security in Networks: An OpenMP Approach for Increasing Malware Detection Speed

“…Parallel systems can be divided in task and data parallelism. Task parallelism refers to the system capable of executing different tasks simultaneously, while data parallelism indicates the execution of separated instances of the same task over different blocks of data at the same time [8]. In a simplistic view, parallel processing may have three main objectives: execute a work faster, increase the throughput of a program or reduce the latency of bottleneck codes.…”

A Realistic Network/Application Model for Scheduling Divisible Loads on Large-Scale Platforms

“…Finally, one could envision extending our application model to address the situation in which each divisible load application consists of a set of tasks linked by dependencies. This would be an attractive extension of the mixed task and data parallelism approach [20,28,3,33,17] to heterogeneous clusters and grids.…”

Streaming Applications

“…Moreover, Section 3.4 will show Hinch efficiently exploits parallel architectures. The major difficulties we encounter are combining task-and data-parallelism [19], supporting dynamic reconfiguration, and handling asynchronous events.…”

“…However, combining task-and data parallelism is difficult [19]. Hinch needs to perform load balancing, synchronize parallel tasks, and allow communication between tasks that concurrently use different resources.…”

“…Some systems support a fixed set of different parallelism types [75]. Bal et al describe several systems that combine task and data parallelism [19]. Only few systems allow extending the range of supported parallelism types [37,44].…”