Energy Efficient High Performance Computing Working Group Infrastructure
 
 
 

Infrastructure

David Grant, ORNL and David Martinez, SNL are co-leads for this Sub-Group

Vision:

A consensus on specific clear and accessible HPC data center infrastructure metrics to facilitate continued improvement of energy performance without compromising mission.

Scope:

  • Encourage development and utilization of infrastructure efficiency metrics and benchmarking
  • Develop key energy performance metrics for HPC facilities (infrastructure)
    • Adapt/use industry standard metrics, e.g. PUE/DCiE
  • Promote utilization of waste heat
  • Address internal and external fans
  • Address internal and external power supplies
  • Discuss air management metrics.
  • Collect and share benchmark data for HPC centers

Documents:

"PUE Past and TUE Future Presentation" is a powerpoint summary of the PUE metric and two proposed new metrics that improve some issues with PUE. The new metrics are called iTUE and TUE.

"Dashboard Recommendations Presentation" is a summary of a paper that provides general recommendations to help select or tailor the energy elements or parameters of an HPC data center infrastructure dashboard. Dashboards will monitor and display energy consumption of various physical data center components in or near real time as well as trend data. The dashboard will display monitored, measured and calculated parameters.

"'Hot' for Warm Water Cooling" is a report that describes an analysis and recommendation for warmer liquid-cooling temperatures that can be used to guide future supercomputer procurements and to standardize the design basis for warmer temperature cooling systems. The vision is to build liquid-cooled solutions that do not require compressors, thereby making them more energy-efficient, lower carbon, and more cost-effective than their air-cooled predecessors. The net result will be significant cost savings, reduced capital expenditures, reduced energy bills as well as reliability improvements. Secondary goals are to reduce or eliminate water consumption (i.e. evaporation in cooling towers) and enable more productive use of heat recovered from the supercomputers. This report is written by the EE HPC WG.

"Thermal Guidelines for Data Processing Environments" is an ASHRAE whitepaper.

LBNL, NREL, and Green Grid created a metric to quantify the use of heat that is normally rejected called Energy Reuse Efficiency. See attached document below for a paper describing this metric entitled "Energy Reuse Efficiency Metric".

A task force representing 7x24 Exchange, ASHRAE, The Green Grid, Silicon Valley Leadership Group, U.S. Department of Energy, U.S. Environmental Protection Agency, United States Green Building Council, and Uptime Institute provides recommendations on measuring and publishing values for PUE at dedicated data center facilities. The task force continues to work towards a Version 2 document, which will provide corresponding guidance for mixed-use buildings. Documents includes an attachment called "Data Center Metrics Task Force Recommendations" which is a Version 1 draft of this document.

An interesting paper on heat re-use entitled "The Data Furnace: Heating up with Cloud Computing".

There are increasing numbers of data centers that have employed heat reuse. We are trying to keep a list of those implementations. Read the "Heat Reuse" document.

Data Center Programming Guide:

There are three new documents that may be of interest. See documents below.

  1. Recommendation for Incorporating Data Center Specific Sustainability Best Practices Into Executive Order 13514 Implementation.

  2. New Data Center Building Energy Programming Guide

  3. Recommendations for Meeting Energy Efficiency Requirements for New Federal Data Centers

 
Acrobat documents
 
 
  Lawrence Livermore National Laboratory
7000 East Avenue • Livermore, CA 94550
  Operated by Lawrence Livermore National Security, LLC, for the
Department of Energy's National Nuclear Security Administration
     
LLNL-WEB-670983  |  Privacy & Legal Notice