A Path To Energy Efficient HPC Datacenters

By Hayk Shoukourian

October 29, 2013

Energy efficiency is rapidly becoming a key factor for many modern high-performance computing (HPC) datacenters. It poses various challenges, which need to be addressed holistically and in an integrated manner, covering the HPC system environment (system hardware and system software), the hosting facility and infrastructure (cooling technologies, energy re-use, power supply chain, etc.), and applications (algorithms, performance metrics, etc.).

Most of the management schemes present in current HPC datacenters do not allow data to be shared between the HPC system environment, hosting facility, and infrastructure. But, it is important to collect and correlate data from all aspects of the datacenter in order to: better understand the interactions between different components of the datacenter; spot the improvement possibilities; and assess any introduced improvements. There are currently no tools that support a complete collection and correlation of energy efficiency relevant data, allowing for a unified view of energy consumption present in the datacenter.

That’s why a new energy measuring and evaluation toolset is being developed at the Leibniz Supercomputing Centre of the Bavarian Academy of Sciences (BAdW-LRZ) which is capable of monitoring and analysing the energy consumption of a supercomputing site in a holistic way, combining the HPC systems with data from the cooling and building infrastructure. The tool, named Power Data Aggregation Monitor (PowerDAM), allows the collection and evaluation of sensor data independently from the source systems and is capable of monitoring not only HPC systems but any other infrastructure that can be represented as a hierarchical tree. It monitors physical sensors as well as virtual sensors which can represent different functional compositions of several physical sensors.

PowerDAM provides a plug-in framework for defining the desired monitored entities such as IT systems, building infrastructure, etc. Two plug-in interfaces for each monitored entity are provided: one for sensor data collection and one for collecting application relevant data (e.g., utilized compute nodes, starting and ending timestamps of application, etc.) from system resource management tools.

PowerDAM is an underlying framework for energy efficiency related research at BAdW-LRZ.

Evaluating and Reporting

Energy-to-Solution (EtS) is an important metric for PowerDAM which denotes the aggregated energy consumption of an application consisting of the energy consumption of utilized compute nodes and partial sub-system components (e.g., system networking and system cooling).

Figure 1 presents the EtS report for an application executed on CoolMUC MPP Linux cluster. The first part of the report (part I) shows the sensor measurements for all utilized components in the order of timestamp, sensor name, value and unit.

Figure 1: EtS Report for an application executed on CoolMUC MPP Linux Cluster

Part II shows all approximations of source measurement data which were considered to be invalid (missing measurements, out of bounds data, etc.). Part III shows the aggregated energy consumption (EtS) of the executed application and provides information on the consumption percentages of computation, networking, and cooling.

The ability to calculate the EtS of an application allows for the further understanding and tuning of the application internally (via change of algorithms, memory access patterns, etc.) as well as externally through hardware adaptation (e.g., static/dynamic voltage frequency scaling).

PowerDAM provides various visualization options such as: the power draw, utilization rate, and averaged CPU temperatures of utilized compute nodes; correlation between power and load for these nodes; different EtS reports; and system power consumption for a given time frame (e.g. day, month, and year). Figure 2 illustrates one of these options – the EtS report (encompassing in parallel to the EtS, the percentages for computation, infrastructure, cooling, and networking) for all executed applications by a given user.

Figure 2: EtS Report for All Jobs Submitted by Given User

PowerDAM “node-map” view displays the dynamic behavior of compute nodes for a given sensor type. This view updates automatically after a customized amount of time and uses a color mapping to classify the behavior of the compute nodes (Figure 3).

Figure 3: Utilization Map of Compute Nodes for CoolMUC Linux Cluster. The color green illustrates the 96% to 100% utilization range. The color white illustrates the 0% and 90% to 95% utilization range. The color red illustrates the 1% to 89% utilization range. (not all compute nodes of the cluster are depicted)

The “node-map” view can be essential for understanding the interconnection between different sensor types. For example, correlating utilization rate (Figure 3) with CPU temperature (Figure 4) allows the investigation of the interdependency between utilization rates and CPU temperatures of defined compute nodes (nodes lxa130 and lxa17).

Figure 4: Temperature Map of Compute Nodes for CoolMUC Linux Cluster (2×8-core AMD CPUs per compute node)
(not all compute nodes of the cluster are depicted)

Further development will allow PowerDAM to: classify applications according to power draw, runtime, performance, and energy consumption; provide data necessary for the enhancement of the resource management systems; and report on datacenter key performance indicators (KPIs) such as PUE, ERE, DCiE, WUE, etc.

More detailed information on PowerDAM is available in the Proceedings of the First International Conference on Information and Communication Technologies for Sustainability under “Towards a Unified Energy Efficiency Evaluation Toolset: An Approach and Its Implementation at Leibniz Supercomputing Centre (LRZ)” and is indexed under DOI 10.3929/ethz-a-007337628.

The development of PowerDAM was made possible by the PRACE Second Implementation Phase project PRACE- 2IP in the Work Package “Prototyping” which has received funding from the European Community’s Seventh Framework Program (FP7/2007-2013) under grant agreement no. RI-283493 and within the SIMOPEK project which has received funding from the German Federal Ministry of Education and Research (BMBF) under grand agreement no. 01IH13007A. The work was achieved using the PRACE Research Infrastructure resources at BAdW-LRZ with support of the State of Bavaria, Germany.

The authors would like to thank Jeanette Wilde for her valuable comments and support.

Author Affiliations

Hayk Shoukourian(1,2); Torsten Wilde(1); Axel Auweter(1); Arndt Bode(1,2)

1Leibniz Supercomputing Centre of the Bavarian Academy of Sciences and Humanities (BAdW-LRZ)

2Technische Universität München (TUM), Fakultät für Informatik

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

AI-Focused ‘Genius’ Supercomputer Installed at KU Leuven

April 24, 2018

Hewlett Packard Enterprise has deployed a new approximately half-petaflops supercomputer, named Genius, at Flemish research university KU Leuven. The system is built to run artificial intelligence (AI) workloads and, as Read more…

By Tiffany Trader

New Exascale System for Earth Simulation Introduced

April 23, 2018

After four years of development, the Energy Exascale Earth System Model (E3SM) will be unveiled today and released to the broader scientific community this month. The E3SM project is supported by the Department of Energy Read more…

By Staff

RSC Reports 500Tflops, Hot Water Cooled System Deployed at JINR

April 18, 2018

RSC, developer of supercomputers and advanced HPC systems based in Russia, today reported deployment of “the world's first 100% ‘hot water’ liquid cooled supercomputer” at Joint Institute for Nuclear Research (JI Read more…

By Staff

HPE Extreme Performance Solutions

Hybrid HPC is Speeding Time to Insight and Revolutionizing Medicine

High performance computing (HPC) is a key driver of success in many verticals today, and health and life science industries are extensively leveraging these capabilities. Read more…

New Device Spots Quantum Particle ‘Fingerprint’

April 18, 2018

Majorana particles have been observed by university researchers employing a device consisting of layers of magnetic insulators on a superconducting material. The advance opens the door to controlling the elusive particle Read more…

By George Leopold

AI-Focused ‘Genius’ Supercomputer Installed at KU Leuven

April 24, 2018

Hewlett Packard Enterprise has deployed a new approximately half-petaflops supercomputer, named Genius, at Flemish research university KU Leuven. The system is Read more…

By Tiffany Trader

Cray Rolls Out AMD-Based CS500; More to Follow?

April 18, 2018

Cray was the latest OEM to bring AMD back into the fold with introduction today of a CS500 option based on AMD’s Epyc processor line. The move follows Cray’ Read more…

By John Russell

IBM: Software Ecosystem for OpenPOWER is Ready for Prime Time

April 16, 2018

With key pieces of the IBM/OpenPOWER versus Intel/x86 gambit settling into place – e.g., the arrival of Power9 chips and Power9-based systems, hyperscaler sup Read more…

By John Russell

US Plans $1.8 Billion Spend on DOE Exascale Supercomputing

April 11, 2018

On Monday, the United States Department of Energy announced its intention to procure up to three exascale supercomputers at a cost of up to $1.8 billion with th Read more…

By Tiffany Trader

Cloud-Readiness and Looking Beyond Application Scaling

April 11, 2018

There are two aspects to consider when determining if an application is suitable for running in the cloud. The first, which we will discuss here under the title Read more…

By Chris Downing

Transitioning from Big Data to Discovery: Data Management as a Keystone Analytics Strategy

April 9, 2018

The past 10-15 years has seen a stark rise in the density, size, and diversity of scientific data being generated in every scientific discipline in the world. Key among the sciences has been the explosion of laboratory technologies that generate large amounts of data in life-sciences and healthcare research. Large amounts of data are now being stored in very large storage name spaces, with little to no organization and a general unease about how to approach analyzing it. Read more…

By Ari Berman, BioTeam, Inc.

IBM Expands Quantum Computing Network

April 5, 2018

IBM is positioning itself as a first mover in establishing the era of commercial quantum computing. The company believes in order for quantum to work, taming qu Read more…

By Tiffany Trader

FY18 Budget & CORAL-2 – Exascale USA Continues to Move Ahead

April 2, 2018

It was not pretty. However, despite some twists and turns, the federal government’s Fiscal Year 2018 (FY18) budget is complete and ended with some very positi Read more…

By Alex R. Larzelere

Inventor Claims to Have Solved Floating Point Error Problem

January 17, 2018

"The decades-old floating point error problem has been solved," proclaims a press release from inventor Alan Jorgensen. The computer scientist has filed for and Read more…

By Tiffany Trader

Researchers Measure Impact of ‘Meltdown’ and ‘Spectre’ Patches on HPC Workloads

January 17, 2018

Computer scientists from the Center for Computational Research, State University of New York (SUNY), University at Buffalo have examined the effect of Meltdown Read more…

By Tiffany Trader

How the Cloud Is Falling Short for HPC

March 15, 2018

The last couple of years have seen cloud computing gradually build some legitimacy within the HPC world, but still the HPC industry lies far behind enterprise I Read more…

By Chris Downing

Russian Nuclear Engineers Caught Cryptomining on Lab Supercomputer

February 12, 2018

Nuclear scientists working at the All-Russian Research Institute of Experimental Physics (RFNC-VNIIEF) have been arrested for using lab supercomputing resources to mine crypto-currency, according to a report in Russia’s Interfax News Agency. Read more…

By Tiffany Trader

Chip Flaws ‘Meltdown’ and ‘Spectre’ Loom Large

January 4, 2018

The HPC and wider tech community have been abuzz this week over the discovery of critical design flaws that impact virtually all contemporary microprocessors. T Read more…

By Tiffany Trader

How Meltdown and Spectre Patches Will Affect HPC Workloads

January 10, 2018

There have been claims that the fixes for the Meltdown and Spectre security vulnerabilities, named the KPTI (aka KAISER) patches, are going to affect applicatio Read more…

By Rosemary Francis

Nvidia Responds to Google TPU Benchmarking

April 10, 2017

Nvidia highlights strengths of its newest GPU silicon in response to Google's report on the performance and energy advantages of its custom tensor processor. Read more…

By Tiffany Trader

Deep Learning at 15 PFlops Enables Training for Extreme Weather Identification at Scale

March 19, 2018

Petaflop per second deep learning training performance on the NERSC (National Energy Research Scientific Computing Center) Cori supercomputer has given climate Read more…

By Rob Farber

Leading Solution Providers

Lenovo Unveils Warm Water Cooled ThinkSystem SD650 in Rampup to LRZ Install

February 22, 2018

This week Lenovo took the wraps off the ThinkSystem SD650 high-density server with third-generation direct water cooling technology developed in tandem with par Read more…

By Tiffany Trader

Fast Forward: Five HPC Predictions for 2018

December 21, 2017

What’s on your list of high (and low) lights for 2017? Volta 100’s arrival on the heels of the P100? Appearance, albeit late in the year, of IBM’s Power9? Read more…

By John Russell

AI Cloud Competition Heats Up: Google’s TPUs, Amazon Building AI Chip

February 12, 2018

Competition in the white hot AI (and public cloud) market pits Google against Amazon this week, with Google offering AI hardware on its cloud platform intended Read more…

By Doug Black

HPC and AI – Two Communities Same Future

January 25, 2018

According to Al Gara (Intel Fellow, Data Center Group), high performance computing and artificial intelligence will increasingly intertwine as we transition to Read more…

By Rob Farber

US Plans $1.8 Billion Spend on DOE Exascale Supercomputing

April 11, 2018

On Monday, the United States Department of Energy announced its intention to procure up to three exascale supercomputers at a cost of up to $1.8 billion with th Read more…

By Tiffany Trader

New Blueprint for Converging HPC, Big Data

January 18, 2018

After five annual workshops on Big Data and Extreme-Scale Computing (BDEC), a group of international HPC heavyweights including Jack Dongarra (University of Te Read more…

By John Russell

Momentum Builds for US Exascale

January 9, 2018

2018 looks to be a great year for the U.S. exascale program. The last several months of 2017 revealed a number of important developments that help put the U.S. Read more…

By Alex R. Larzelere

Google Chases Quantum Supremacy with 72-Qubit Processor

March 7, 2018

Google pulled ahead of the pack this week in the race toward "quantum supremacy," with the introduction of a new 72-qubit quantum processor called Bristlecone. Read more…

By Tiffany Trader

  • arrow
  • Click Here for More Headlines
  • arrow
Share This