Special Issue Paper

Computer Science - Research and Development

, Volume 29, Issue 3, pp 221-229

Automatic detection of power bottlenecks in parallel scientific applications

  • María BarredaAffiliated withDepto. de Ingeniería y Ciencia de Computadores, Univ. Jaume I Email author 
  • , Sandra CatalánAffiliated withDepto. de Ingeniería y Ciencia de Computadores, Univ. Jaume I
  • , Manuel F. DolzAffiliated withDepartment of Informatics, University of Hamburg
  • , Rafael MayoAffiliated withDepto. de Ingeniería y Ciencia de Computadores, Univ. Jaume I
  • , Enrique S. Quintana-OrtíAffiliated withDepto. de Ingeniería y Ciencia de Computadores, Univ. Jaume I

Rent the article at a discount

Rent now

* Final gross prices may vary according to local VAT.

Get Access

Abstract

In this paper we present an extension of the pmlib framework for power-performance analysis that permits a rapid and automatic detection of power sinks during the execution of concurrent scientific workloads. The extension is shaped in the form of a multithreaded Python module that offers high reliability and flexibility, rendering an overall inspection process that introduces low overhead. Additionally, we investigate the advantages and drawbacks of the RAPL power model, introduced in the Intel Xeon “Sandy-Bridge” CPU, versus a data acquisition system from National Instruments.

Keywords

Power efficiency High performance computing Profiling and tracing power consumption Scientific applications