arXiv.itsnick.co

an independent field guide to 23 million papers from arXiv & OpenAlex — not affiliated with arXiv.org

Computer Science · Computer Networks and Communications

UK.

S. Bani-mohammad, Glasgow G Rz, I. Ababneh, Lewis M. Mackenzie

See on the map

Abstract

Two strategies are used for the allocation of jobs to processors connected by mesh topologies: contiguous allocation and non-contiguous allocation. In non-contiguous allocation, a job request can be split into smaller parts that are allocated to non-adjacent free sub-meshes rather than always waiting until a single sub-mesh of the requested size and shape is available. Lifting the contiguity condition is expected to reduce processor fragmentation and increase system utilization. However, the distances traversed by messages can be long, and as a result the communication overhead, especially contention, is increased. The extra communication overhead depends on how the allocation request is partitioned and assigned to free sub-meshes. This paper presents a new Non-contiguous allocation algorithm, referred to as Greedy-Available-Busy-List (GABL for short), which can decrease the communication overhead among processors allocated to a given job. The simulation results show that the new strategy can reduce the communication overhead and substantially improve performance in terms of parameters such as job turnaround time and system utilization. Moreover, the results reveal that the Shortest-Service-Demand-First (SSD) scheduling strategy is much better than the First-Come-First-Served (FCFS) scheduling strategy.

More like this