Power- and Cooling- Aware Parallel Performance Diagnosis

Rashawn L. Knapp, Karen L. Karavanic, Sriram Krishnamoorthy, and Andres Marquez


power-aware computing, parallel performance tools, performance analysis, green data center


Increasing concern about the power consumption of data centers and computer laboratories, which in some cases matches or exceeds the resources required to power a small city, drive a need for a new, integrated approach to parallel performance diagnosis that integrates traditional application-oriented performance data with measurements of the physical runtime environment. We have developed infrastructure for combined evaluation of system, application, and machine room performance in the high end environment. We present a new intensity metric for use in automated performance diagnosis tools, and show results from a case study of the performance, power and cooling impact of the choice of physical location for a scientific application within the machine room.

Important Links:

Go Back