Visit additional Tabor Communication Publications
July 15, 2010
Summer is heating up, and so are our supercomputers. The insatiable drive for more computational performance means servers are becoming ever denser, and correspondingly hotter. Today, a rack of high-end blades can dissipate 30 kilowatts or more. And with the era of coprocessor acceleration upon us, many HPC servers are being fitted with 200-watt GPUs, further adding to the heat load.
This wouldn't matter so much if we kept our machines in the pool, but air being what it is (a poor conductor of heat), the burden on the cooling infrastructure keeps escalating. Keeping the machinery at a comfortable temperature can represent from a third to a half of a facility's power consumption. Even with the most recent recommendations from the American Society of Heating, Refrigerating and Air-conditioning Engineers (ASHRAE) to crank up the datacenter thermostat from 77F to 80F, HPC machine rooms are reaching their thermal limits.
That's why liquid cooling has been such a big part of supercomputing. These machines, especially the proprietary designs, have always been on the leading edge of computational density, and sometimes wouldn't survive on air flow alone. In fact, since the days of the early Cray systems in the 1970s, a lot of the top-end supercomputers have had water or some other liquid coolant running through the hardware. That's why the father of supercomputing, Seymour Cray, referred to himself as "an overpaid plumber."
A couple of recent stories point to a new direction for liquid cooling. Instead of just running coolant through the racks, it's now being funneled directly onto the hottest components: the processors themselves. IBM's Aquasar supercomputer, which was recently delivered to the Swiss Federal Institute of Technology Zurich (ETH Zurich), is an example of one such system.
The 6-teraflop Aquasar machine uses customized water-cooled BladeCenter servers that sport both Intel Nehalem CPUs and IBM PowerXCell processors. Water is piped into a heat exchanger that sits right on top of the chips. Because of the intimate contact with the processors, the water does not need to be chilled, and can be as warm as 60C. That's 140F for those of you keeping score in the USA. The idea is to keep the processors below their critical maximum of 85C (185F).
At ETH Zurich, the heated (waste) water is piped away to help warm the buildings at the facility. IBM claims the carbon footprint of such a system is reduced by as much as 85 percent compared to a conventionally-cooled computer setup.
A more general case involves what Google is doing -- or thinking about doing. The company recently filed to patent a server assembly design in which two motherboards sandwich a liquid-cooled heat sink. In this setup, the processors are being cooled via the heat sink, while the other components, like the memory chips, are air cooled. According to a report in Data Center Knowledge:
The design is among a number of Google patents on new cooling techniques for high-density servers that have emerged since the company’s last major disclosure of its data center technology in April 2009. Several of these patents deal with cooling innovations using either liquid cooling or air cooling applied directly on server components.
In 2007, Google filed a patent for a different sort of liquid-cooling arrangement. The "Water Based Data Center" design outlined sea-based computing facility that floats on the water, employs the waves to help generate electricity, and uses the sea water to help provide cooling for the computers. That patent was granted in May 2009.
Perhaps an even more novel method is immersion cooling, in which the whole server is submerged into an inert liquid, such as mineral oil. That too, is not a new concept. Some of the early supercomputing systems, including the Cray-2*, used immersion cooling. A modern version is being offered by Austin, Texas-based Green Revolution Cooling, which claims its horizonal rack design and "GreenDef" oil coolant can manage power densities as high as 100 kilowatts per rack. Bring on the GPUs!
The company is claiming its immersion system uses 95 percent less power than conventional cooling. Some of that can be attributed to the fact that all the internal server fans can be yanked out, which alone should reduce the power draw by 5 to 25 percent. The company recently installed some test units at the Texas Advanced Computing Center (TACC). If the Green Revolution offering pans out as advertised, maybe we'll see more supers taking the plunge.
*The original post incorrectly specifed Cray-1 as one of the early supercomputers using immersion cooling. It was the Cray-2 design that introduced this cooling design. Hat tips to readers Richard Lakein and Max Dechantsreiter for pointing out the gaffe. -- Michael
Posted by Michael Feldman - July 15, 2010 @ 9:58 AM, Pacific Daylight Time
Michael Feldman is the editor of HPCwire.
No Recent Blog Comments
Contributing commentator, Andrew Jones, offers a break in the news cycle with an assessment of what the national "size matters" contest means for the U.S. and other nations...
Today at the International Supercomputing Conference in Leipzing, Germany, Jack Dongarra presented on a proposed benchmark that could carry a bit more weight than its older Linpack companion. The high performance conjugate gradient (HPCG) concept takes into account new architectures for new applications, while shedding the floating point....
Not content to let the Tianhe-2 announcement ride alone, Intel rolled out a series of announcements around its Knights Corner and Xeon Phi products--all of which are aimed at adding some options and variety for a wider base of potential users across the HPC spectrum. Today at the International Supercomputing Conference, the company's Raj....
Jun 19, 2013 |
Supercomputer architectures have evolved considerably over the last 20 years, particularly in the number of processors that are linked together. One aspect of HPC architecture that hasn't changed is the MPI programming model.
Jun 18, 2013 |
The world's largest supercomputers, like Tianhe-2, are great at traditional, compute-intensive HPC workloads, such as simulating atomic decay or modeling tornados. But data-intensive applications--such as mining big data sets for connections--is a different sort of workload, and runs best on a different sort of computer.
Jun 18, 2013 |
Researchers are finding innovative uses for Gordon, the 285 teraflop supercomputer housed at the San Diego Supercomputer Center (SDSC) that has a unique Flash-based storage system. Since going online, researchers have put the incredibly fast I/O to use on a wide variety of workloads, ranging from chemistry to political science.
Jun 17, 2013 |
The advent of low-power mobile processors and cloud delivery models is changing the economics of computing. But just as an economy car is good at different things than a full size truck, an HPC workload still has certain computing demands that neither the fastest smartphone nor the most elastic cloud cluster can fulfill.
Jun 14, 2013 |
For all the progress we've made in IT over the last 50 years, there's one area of life that has steadfastly eluded the grasp of computers: understanding human language. Now, researchers at the Texas Advanced Computing Center (TACC) are utilizing a Hadoop cluster on its Longhorn supercomputer to move the state of the art of language processing a little bit further.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?
Join our webinar to learn how IT managers can migrate to a more resilient, flexible and scalable solution that grows with the data center. Mellanox VMS is future-proof, efficient and brings significant CAPEX and OPEX savings. The VMS is available today.