Visit additional Tabor Communication Publications
November 30, 2011
Some of the most important supercomputing models aimed at climate change research have been developed by National Oceanic and Atmospheric Administration (NOAA), and in particular, its Geophysical Fluid Dynamics Laboratory (GFDL) at Princeton University. The GFDL researchers are experts in climate science, but as with many scientists, they are often less adept with the vagaries of supercomputing technology. That's where HPTi comes in.
HPTi (High Performance Technologies Inc.) was bought by DRC Company in July, but maintains its autonomy and mission as a federal contractor for high-end technology support. The company's strength lies in its HPC expertise and its ability to apply computational and science research to their clients' applications. For NOAA, HPTi provides high performance computing know-how, consulting and training. As part that contract, HPTi supports the GFDL climate work, helping researchers there upgrade their climate models as well as providing guidance for future supercomputing hardware and software tools.
A major reason the GFDL work is so important is that their results are incorporated into climate assessments composed by the Intergovernmental Panel on Climate Change (IPCC). Although IPCC does use results from climate research derived at other labs, the GFDL models are central to their assessments. And the reports themselves have become the de facto standard for climate policymakers, scientists, the press, and the public.
The niche HPTi has carved out with GFDL has allowed researchers at the lab to concentrate on the physics of the models, leaving the nitty-gritty of HPC to the HPTi staff of computational scientists, software developers, systems support people, and other consultants. Because the company provides a lot of the computation glue for the researchers, HPTi tends to maintain long-term relationships (and contracts) with agencies like NOAA. In this case, the company has been supporting the GFDL climate research effort since 2008.
The continuity of involvement is important. The HPC hardware for the climate work gets upgraded every few years, requiring a reassessment of the software as well as the software development tools. According to William Cooke, a senior associate who works with climate modeling team at GFDL, over the last four years, the HPC systems used to run the models have changed dramatically.
Cooke says as recently as a few years ago, GFDL was employing an 8,000-core SGI Altix supercomputer, with a shared-memory architecture. A couple of years ago, they move to a Cray XT6 machine, with 30,000-plus cores. In the next few months, they're going to upgrade that system to an XE6, adding 78,000 more cores in the process. When installed, that system will represent a peak petaflop of computational horsepower.
Ideally the scientists would like to just recompile their application software and run it on the new machine, but in practice, that's not what happens. The hardware upgrades, especially the greatly increased core counts, necessitate that the climate models be modified if they are to take advantage of the additional computational power.
The additional power also allows the researchers to consider adding extra features, such as atmospheric chemistry, CO2 feedback, phytoplankton blooms, more detailed landforms and so on. But more directly, the extra cores can be used to increase the fidelity of the existing models.
For example, the current climate models use a two-degree square resolution for the atmosphere and land and a one-degree resolution for the ocean and ice. To get more fine-grained results, the scientists would like to get the atmosphere/land model down to half a degree or better and the ocean/ice model to at least a quarter of a degree. The resulting simulation will be better at picking up smaller scale effects like hurricane activity and the intensity of regional rainfall or drought.
To make it all work though, the models need to be stitched together, which takes a special piece of software called the coupler. According to Cooke, that's another critical components that HPTi has been spending a lot of time with. And in this area, he says, the increased core count that came with the Cray supercomputer forced a rewrite of the underlying algorithms. The new version not only enabled the coupled model to run on over 10,000 cores, but it also cut the simulation time in half.
That's significant, given the amount of computer time devoted to these climate models. Running a 20- to 30-year simulation takes about a week on the current system, but forecasting hundreds of model-years can tie up the same machine for up to six months. The scaled up software translates into more runs for the researchers, allowing them to refine their results and create more "what if" scenarios.
At the backend of the simulation, the computation turns into a typical big data problem. According to Cooke, even at two degrees resolution, the models generate about half a petabyte per month, and this has been going on for the last couple of years. With finer resolution, these datasets will get even larger.
Currently the raw data is sent over NOAA's research network (N-Wave) every day to be post-processed at GFDL. But as the models generate more and more data, it tends to become stuck in place, which is why data lifecycle management is becoming a critical component of the research. This is yet another area that HPTi is providing guidance for.
The IPCC's Fifth Assessment Report, which will include the latest simulation work from GFDL, is now underway and is scheduled for completion in 2013-2014. The report, the research, and the data upon which it rests will be available in the public domain.
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.
Jun 13, 2013 |
Titan, the Cray XK7 at the Oak Ridge National Lab that debuted last fall as the fastest supercomputer in the world with 17.59 petaflops of sustained computing power, will rely on its previous LINPACK test for the upcoming edition of the Top 500 list.
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.