insideHPC » Green HPC

Energy Efficiency as a Worldwide Quest

Rich Brueckner — Sat, 05 May 2012 17:19:02 +0000

Over at the ISC Blog, HP’s Marc Hamilton writes that improving system energy efficiency is everyone’s business.

Today’s best process technology requires about 70 picojoules per floating point operation (FLOP). A US Defense Advanced Research Agency (DARPA) study projected that by the end of the decade, this will drop to 5-10 picojoules per FLOP. Doing the math, a 2020 Exascale system would require 5-10 megawatts to perform 1 ExaFLOP of calculations. While that is a lot of power, there are certainly many data centers today that can support 5-10 metawatts. So why all the concern about energy efficiency? Unfortunately, FLOPs alone are not the power driver in modern supercomputers. The energy cost of moving two 64 bit operands in and out of the processor is estimated at 1000 to 3000 picojoules per FLOP, and that means up to 1 gigawatt would be required for an Exascale system, well outside the capabilities of even the world’s largest data centers.

Read the Full Story.

The Green Brick Road to Exascale

Rich Brueckner — Wed, 04 Apr 2012 21:00:00 +0000

In a guest post on the ISC blog, Beth Sharp and Dr Tom Wilkie of Scientific Computing World write that the price of energy will dominate the future of HPC.

Ultimately though, it is on the road to exascale that the price of energy will dominate the future of HPC. And again, this is where energy efficiency will turn out to be a benefit rather than a cost to the community. The best performer in the November 2011 Green500 list was the IBM BlueGene/Q machine at the University of Rochester in the USA with a tally of just over 2000 Mflops/Watt. It is not the fastest machine in the world, coming in at place 29 on the Top500 list. But a straightforward extrapolation of its technology would indicate a power demand of 500 MW at exascale.

Read the Full Story.

Video: Scalable Cluster Computing with Nvidia GPUs

Rich Brueckner — Fri, 16 Mar 2012 09:17:57 +0000

www.youtube.com/watch?v=MH8RZX18yqA

In this video, Axel Koehler presents: Scalable Cluster Computing with NVIDIA GPUs.

Recorded at the HPC Advisory Council Switzerland Workshop on March 15, 2012. Download the slides (PDF).

Slidecast: Hardcore Computer – Relentless Performance through Liquid Submersion

Rich Brueckner — Sun, 26 Feb 2012 13:00:46 +0000

www.youtube.com/watch?v=gvVAQWyuPwk

In this video, Chad Attlesey presents: Hardcore Computer: Relentless Performance through Liquid Submersion. The company designs and manufactures top-tier, high-performance computer systems using a patented liquid submersion cooling technology.

It has been recently estimated that as much as six percent of the US power grid goes to data centers,” said Chad Attlesey, CTO and President of Hardcore Computer. “Roughly half of that power goes to air conditioning and air moving equipment needed for air cooling or hybrid cooling of servers and other rack-mounted equipment. We have an elegant approach to solving the problem, which delivers radically improved cooling efficiency and allows for maximum computational power in the smallest possible real estate.”

Learn more at Hardcore Computer * Download the MP3 * Subscribe on iTunes * If Dropbox is blocked, download from this Google page.

Podcast: Business Opportunities from the Climate Challenge

Rich Brueckner — Sat, 25 Feb 2012 13:00:34 +0000

In this Intel Chip Chat podcast, Allyson Klein speaks with Jon Koomey from Stanford about his new book Cold Cash, Cool Climate. Koomey makes the case that their is profit to be made with Greener computing. Download the MP3.

Project Will Meter Energy Usage of European HPC Rampup

Rich Brueckner — Thu, 23 Feb 2012 20:10:44 +0000

A new European Commission-funded project aims to tackle the energy-efficiency implications of increasing investment in new HPC datacenters. On the heels of the EC announcement of increased HPC investments in the region, the main goal of the three-year CoolEmAll Consortium is to increase understanding of how different factors impact energy-efficiency.

The project will develop two key tools to allow others to leverage the research done by the CoolEmAll Consortium.

Simulation, Visualisation and Decision support (SVD) toolkit. The SVD toolkit is a real-time CFD tool that will allow datacenter planners to model the energy efficiency implications of physical placement of servers within a facility, different approaches to cooling, as well as the role played by applications and workload.
Blueprints/designs of energy-efficient hardware. As a set of open source designs, Blueprints will be based on a high-density server known as the RECS|Compute Box developed by German start-up Christmann informationstechnik. These designs, along with the SVD toolkit, should al

Read the Full Story.

DARPA Shoots for Power Efficiency Revolution with PERFECT Workshop

Rich Brueckner — Mon, 30 Jan 2012 16:23:06 +0000

Clipped from: www.darpa.mil (share this clip)

One of the toughest hurdles on the road to Exascale is processing power efficiency. This week DARPA announced a new initiative called the Power Efficiency Revolution for Embedded Computing Technologies, or PERFECT. The initiative will kick off with a Proposer’s Day Workshop in Arlington, Virginia on February 15.

PERFECT aims to achieve the 75 GFLOPS/w goal by taking novel approaches to processing power efficiency. These approaches include near threshold voltage operation and massive heterogeneous processing concurrency, combined with techniques to effectively use the resulting concurrency and tolerate the resulting increased rate of soft errors. The program seeks to leverage and incorporate anticipated industry fabrication geometry advances to 7 nanometers. PERFECT does not plan to build hardware, rather it seeks to develop a simulation capability to measure and demonstrate progress. It plans to specifically address embedded systems processing power efficiencies and performance, and is not concerned with developments that focus on exascale processing issues.

Read the Full Story.

Podcast: Is There a Moore’s Law for Energy Efficiency?

Rich Brueckner — Mon, 02 Jan 2012 13:00:18 +0000

In this IEEE podcast, Stanford’s Jonathan Koomey discusses his recent paper looking at the Moore’s Law–like exponential progress in the energy efficiency of computers over the past six decades.

So, if these trends continue—and my friends at Intel say that they have at least another 10 years of these performance and efficiency trends improving—if they continue, they basically bode well for the increased sophistication and lower-power use of mobile computing centers and controls. So doubling every year and a half means over the course of 10 years, the efficiency of computers will go up by a factor of 100. So that means that we will have the ability to have more and more mobile centers that use less and less power to collect information and also to bring it back to some wireless base station. The people who are doing the most interesting work now are focused on ultralow power computing that was never before possible. And so Josh Smith, who’s at Intel and also the University of Washington, is working with mobile centers that have no power source. No battery. They use the power from stray television and radio signals to power themselves. And that means that you’re free from the constraint of having to have a power source.

Read the Full Transcript.

RSC Super is Greenest in Russia

Ralph — Fri, 09 Dec 2011 21:04:49 +0000

This week the Russian supercomputer vendor RSC Group announced that their liquid-cooled x86 system installed at South Ural State University is the most energy efficient Russian HPC system according to the Green500 list published in November 2011. The system took the 109th position in the rating. Currently there are only five Russian supercomputers in Green500, their number decreased more than twice compared to the previous edition of the rating.

We are happy that the most powerful supercomputing system built by RSC by this time is a de facto the most energy efficient HPC system in Russia and CIS, as shown in the new edition of Green500 rating. Undoubtedly, our energy efficient RSC Tornado architecture with liquid cooling for widely available standard server boards made a great contribution in this success. Most computing nodes of SKIF Aurora SUSU are based on this architecture”, said Alexey Shmelev, Chief Operating officer at RCS Group.

Read the Full Story.

GPU-Powered Tsubame 2.0 Ranked as only Petaflop System on Green500

Rich Brueckner — Wed, 23 Nov 2011 16:34:56 +0000

Let’s face it; when it comes to the TOP500, most people pay attention to the biggest systems on the list. But for the Green500, there’s a million ways to spin it.

Today Nvidia announced that, for the second year in a row, the fastest system on the Green500 is powered by the company’s Tesla GPUs. At #2 on the Green500 list, the Tsubame 2.0 supercomputer at the Tokyo Institute of Technology sustains 1.192 Petaflops of LINPACK performance with only 1398.61 Kw of power. The system is also the only contender in the Top5 of both lists (it is also ranked #5 on the TOP500).

The rise of GPU supercomputers on the Green500 signifies that heterogeneous systems, built with both GPUs and CPUs, deliver the highest performance and unprecedented energy efficiency,” said Wu-chun Feng, founder of the Green500 and associate professor of Computer Science at Virginia Tech.

Three other Tesla GPU-based systems made the Top 10. The National Center for Supercomputing Applications (NCSA) and Georgia Institute of Technology in the U.S. and the National Institute for Environmental Studies in Japan secured #3, #9, and #10 respectively. Tesla GPU-based systems also secured #11, #12, and #13 CSIRO in Australia, Tianhe-1A in Tianjin and Nebulae in Shenzhen, China. Read the Full Story.

Hydro Power Draws Green Supercomputers to Holyoke

Ralph — Mon, 10 Oct 2011 17:51:38 +0000

Clipped from: www.mghpcc.org (share this clip)

Five Universities are teaming to build the Massachusetts Green High Performance Computing Center using cheap water power from the Connecticut River.

Boston University, Harvard University, the Massachusetts Institute of Technology, Northeastern University and the University of Massachusetts formed the venture to boost academic research in protein structure, fluid flows, the dynamics of the earth’s atmosphere, human social interaction, the evolution of the galaxy and other issues. The universities are each spending $10 million, the state of Massachusetts has committed $25 million and technology companies EMC Corp. and Cisco Systems Inc. are contributing $2.5 million apiece.

The relatively cheap electricity is particularly important for the computing center, which is expected to consume up to 15 megawatts, the equivalent of powering as many as 15,000 homes. Read the Full Story.

Interview: AMD’s John Fruehe on Bulldozer’s Power Saving Features

Rich Brueckner — Fri, 07 Oct 2011 14:55:39 +0000

This week Joe Casad interviewed John Fruehe, AMD’s Director of Product Marketing, about the Bulldozer line of 16-core processors. One thing I found interesting was Bulldozer’s new TDP Power Cap.

TDP Power Cap lets you set a custom TDP (power limit) for the processor. Many customers are trying to optimize their data centers and get the best density possible. Since they have a predefined power budgets per rack, TDP power capping allows a customer to take the TDP down a bit, freeing up some of that power headroom so that they can maximize space in their rack. A good example was a customer I met with in Europe. They had power budgets for the rack, and always ran out of power first, so they had 3-5 slots free at the top of the rack. With TDP Power Cap, they can reduce the TDP, allowing them to populate the extra slots at the top of the rack, which allowed them to reduce the number of total racks grow their server count without taking up more floor space.

Read the Full Story.

Green Computing: An Implicit Thrust for SC11

Staff — Wed, 05 Oct 2011 11:45:35 +0000

In this guest feature, Natalie Bates, Co-chair of the Energy Efficient HPC Working Group, gives us a preview of sustainability topics at SC11.

Energy efficiency will again be a hot topic at SC11, with 27 Technical Program sessions explicitly focused on energy efficiency. A complete list of these sessions organized both chronologically and by topic can be found on the Energy Efficient HPC Working Group website. SC11, the annual International Conference for High Performance Computing, Networking, Storage and Analysis, will be held Nov. 12-18 in Seattle. For more information, see the SC11 website.

Here is a look at some of the key topics and related sessions;

Liquid cooling. There are six sessions on liquid cooling, four of which are case studies. The other two sessions describe an initiative involving the U.S. Department of Energy (DOE) supercomputing centers working collaboratively with industry representatives to develop guidelines for warmer liquid-cooling temperatures to guide future supercomputer procurements, and to facilitate the design of warmer temperature cooling systems. Their vision is to build liquid cooled solutions that do not require compressors and are more energy-efficient, more carbon-friendly and more cost-effective than the air-cooled predecessors. A paper describing this initiative will be presented at a State-of-the-Practice session on Wed., Nov 16.

Energy efficient architectures and trends. The Green500 Birds-of-a-Feather on Thurs., Nov. 17 is one of five sessions looking at this area and will likely build on the June List that showed two trends; aggregating low-power processors and using accelerators. Taking low-power processors for HPC further, there is another Birds-of-a-Feather session on Tue., Nov 15 that will announce a new website on using embedded computing technologies with processors developed for smart phones and tablets. There are also two technical papers. A team from Facebook will present their custom server design which significantly improves power efficiency as well as environmental footprint. Another team from Lawrence Berkeley National Lab compares three architectural alternatives: Intel CPU, NVIDIA GPU and an FPGA-simulation. The FPGA-simulation showed both best performance and power efficiency.

On a related note, there is a Birds-of-a-Feather on Tue., Nov 15 that gathers an existing community driving toward a commonly adopted set of methodologies, workloads and metrics for measuring the energy efficiency of supercomputer architectures. This is a collaborative effort between the Green500, the TOP500, The Green Grid and the Energy Efficient HPC Working Group. An immediate focus of this community is to improve the methodology for measuring the energy and/or power consumed while running workloads such as High Performance Linpack.

Energy efficiency of datacenters and systems. An all-day Workshop on Sun., Nov. 13, as well as a Panel on Wed., Nov. 16, emphasize networking, which is a topic that was only lightly covered in SC10 Technical Program sessions. For the networking expert, there is also an Exhibitor Forum that will present an overview of thermal solutions for pluggable I/O.

Another Workshop on Mon., Nov. 14, explores best practices as well as game-changing scenarios for energy efficiency of both data centers and systems. Among other topics, this Workshop will describe newly established sustainability metrics for Energy Reuse Effectiveness (ERE), Water Usage Efficiency (WUE), and Carbon Usage Effectiveness (CUE). A Birds-of-a-Feather on Wed., Nov. 16 evening provides a survey of state-of-the-art tools and techniques for optimizing system power at many levels; from component hardware to applications.

Eight sessions in Doctoral Research Showcase, Papers and Posters explore energy efficient supercomputing from various perspectives; proposals for job schedulers that consider electricity related factors, managing the trade-off between energy savings and increased application execution time, a novel energy saving synchronization technique, an automated multi-threaded power virus, and a power analysis framework for clusters.

Finally, there are three Technical Program sessions that focus on the use of high performance computing to help deliver renewable energy and make energy distribution systems more effective. A Masterworks session on Thur., Nov 17 will highlight the scientific and computational challenges of producing practical fusion energy (addressed through the DOE Fusion Simulation Program). There is a Workshop on Sun., Nov. 13, promoting the use of high performance computing and networking for higher fidelity power grid simulation and higher frequency measurement of their state. Similarly, a paper will describe a scalable approach to optimize energy systems for efficient integration of wind power into the power grid.

Although this is a list of sessions with a specific focus on energy efficiency, many more sessions will include energy efficiency as part of a broader focus.

Video: Microsoft – Datacenter Efficiency – Optimizing Performance and Power

Rich Brueckner — Thu, 29 Sep 2011 12:00:22 +0000

www.youtube.com/watch?v=PlCaRatudXA

In this video, Ryan Waite, GM of HPC for Microsoft presents: Datacenter Efficiency - Optimizing Performance and Power. Recorded at the HPC User Forum in San Diego on Sept. 7, 2011.

There’s still time to register for the HPC User Forum series coming to Europe next week:

At CEA in France, Oct. 3-4. Agenda and Registration site
At HLRS in Germany, Oct. 6-7 Agenda and Registration site

Registration is free of charge, including refreshments.

Video: New Sustainability Metrics Measure Datacenter’s Total Impact on the Planet

Rich Brueckner — Wed, 28 Sep 2011 15:38:33 +0000

www.youtube.com/watch?v=itEpgBz6Z7U

In this video, Nicolas Dube from the Green Grid describes newly established sustainability metrics for datacenters. Just as Power Usage Effectiveness (PUE) has become a standard for measuring power efficiency, the new metrics take into account the environmental impact of a datacenter in terms of Energy Reuse Effectiveness (ERE), Water Usage Efficiency (WUE), and Carbon Usage Effectiveness (CUE).

Establishing these metrics is a great step forward, but it is going to take a lot of work to get the results that let us all breathe easier. If you have interest in sustainability, I urge you to attend or send someone from your organization to the following workshops at SC11:

Workshop: Towards and Beyond Energy Efficiency, Monday, Nov. 14 9:00AM - 5:30PM
Birds of a Feather: Setting Trends for Energy Efficiency, Tuesday, Nov. 15 12:15PM - 1:15PM

These workshops are hosted by the Energy Efficient High Performance Computing Working Group, They have a number of meetings and webinars scheduled on their site, and you can keep up with it all by joining their Linkedin Group.

Moving Compute for a Greener Future

Rich Brueckner — Tue, 20 Sep 2011 11:33:01 +0000

In a recent interview in EarthSky magazine, AMD Fellow David Mayhew describes a research collaboration that’s helping plan how to build a network of data centers powered by wind and solar energy.

What if you could rent access to computation from a bunch of data centers sitting out in fields next to windmills? Then, first, you don’t own the computer anymore, and it’s not really your problem to keep it running and keep it from having viruses. And you’re not upgrading it. You would just have this ubiquitous access to computation. That would be a much better model, I think, for most users, whether personal users or business users. Let’s turn computation into something that’s a lot like electricity. We use it but leave its creation to a relatively small set of very expert people.

Read the Full Story or download the MP3.

DARPA Challenge to Inspire 40x Power Efficiency Improvement for Supercomputers

Rich Brueckner — Thu, 01 Sep 2011 12:00:39 +0000

Eric Smalley from Discover Magazine describes how researchers are stepping up to a DARPA Challenge to construct a Petaflop supercomputer by 2018 that consumes no more than 57 kilowatts of electricity.

The teams that survive the initial design, simulation, and prototype-building phases may earn a chance to build a full-scale supercomputer for Darpa. Making the cut will demand a total rethink of computer design. Nearly everything a conventional computer does involves schlepping data between memory chips and the processor (or processors, depending on the machine). The processor carries out the programming code for jobs such as sorting email and making spreadsheet calculations by drawing on data stored in memory. The energy required for this exchange is manageable when the task is small—a processor needs to fetch less data from memory. Supercomputers, however, power through much larger volumes of data—for example, while modeling a merger of two black holes—and the energy demand can become overwhelming. “It’s all about data movement,” said Sandia’s Richard Murphy

Four teams from Intel, Nvidia, Sandia and MIT are currently competing, but the final solution may incorporate multiple technologies derived from the competition. Read the Full Story.

NPR Podcast: Curbing Supercomputers’ Growing Drain on Energy

Rich Brueckner — Tue, 02 Aug 2011 17:48:01 +0000

In this podcast, LBNL’s John Shalf appears on NPR’s Morning Edition radio show on efforts to reinvent supercomputing processors to consume less energy.

Supercomputers have become a critical tool for scientists. Each year, they get bigger and faster — and use a lot more power. Soon, each one will need as much energy as a small city. That has researchers looking to reinvent the supercomputer — by using the technology inside cellphones.

Video: Panel Discussion – Has the PUE Issue Already Been Solved for the Datacenter?

Staff — Mon, 25 Jul 2011 11:30:10 +0000

www.youtube.com/watch?v=2zVMeTRn3KA

Has the issue of optimized PUE largely been solved already for large datacenters? In this presentation, you’ll hear from the following individuals in a panel discussion moderated by James Hamilton of Amazon.

Chris Malone, Google
Rob Coupland, Telecity Group
Ian Bitterlin, Ark Continuity Ltd
Brian Waddell, Norman Disney & Young

Recorded at the 2011 Data Centre Efficiency Summit held on May 24 in Zürich, Switzerland. A tip of the hat goes to Datacenter Knowledge for pointing us to this story.

Green Storage Technology Takes Off for the Sustainable Cloud

Rich Brueckner — Thu, 14 Jul 2011 15:16:06 +0000

This week Cutting Edge Networked Storage rolled out their new green storage technology for the cloud at the Esri International User Conference in San Diego. The new Power Saving Cloud Archive boasts “up to 90% power savings” and lower TCO for organizations struggling with mounds of unstructured data.

Our new Power Saving Cloud Archive sets a new paradigm toward the deployment of advanced unified multi-tiered storage solutions that provide access at both block level and file level, thereby giving users greater flexibility to handle the full range of data types within the same storage systems,” said Michael Ehman, president of Cutting Edge Networked Storage. “The bottom line is that the Power Saving Cloud Archive architecture is poised to revolutionize the industry by taking unified storage to the next level – into the “Green Cloud” – which will result in unprecedented levels of performance, scalability, reduced carbon emissions and cost savings.

Based on an open-systems approach, the Power Saving Cloud Archive uses EdgeWare for unified storage related functions. Read the Full Story.