Measuring Bandwidth for Super Computer Workloads

TL;DR

This paper discusses the importance of bandwidth measurement in supercomputers, demonstrating how real workload traces from LANL can be used to estimate bandwidth for better resource management and system utilization.

Contribution

It presents a method to estimate bandwidth using real supercomputer workload traces, aiding in resource management and performance analysis.

Findings

Bandwidth estimation is feasible using real workload traces.

Proper bandwidth measurement can improve supercomputer resource utilization.

The study provides a practical approach for bandwidth analysis in supercomputing environments.

Abstract

Parallel computing plays a major role in almost all the fields from research to major concern problem solving purposes. Many researches are till now focusing towards the area of parallel processing. Nowadays it extends its usage towards the end user application such as GPU as well as multi-core processor development. The bandwidth measurement is essential for resource management and for studying the various performance factors of the existing super computer systems which will be helpful for better system utilization since super computers are very few and their resources should be properly utilized. In this paper the real workload trace of one of the super computers LANL is taken and shown how the bandwidth is estimated with the given parameters.