HPC
Analytics
Cloud
Monday, July 30, 2012
Wednesday, July 25, 2012
Reference Architecture for NGS HPC
A week ago (2012.07.18), I released the second version of my reference architecture for next-gen sequencing HPC infrastructure. It was based on the first version which was implemented in four projects (MU, NWU Feinberg, KSU and UofAz). For the new version, I incorporated many new features:
- Adding Hadoop cluster – to handle 2nd analysis workload ported to Hadoop
- Adding Compute cluster (CPU, GPU and FPGA) – to handle 2nd analysis workload
- Adding deep/active archive (adopted from refac of Petascale Active Archive)
- Adding workload management with Platform LSF, Application Center (PAC) and Process Manager (PM)
- Adding Cloud management with Platform PCM-AE
- Adding analytical engines (for statistical and predictive analysis, and integration with other -omics data)
Update:
- original post: 2012.07.25
- updated diagram: 2012.08.03
Thursday, July 12, 2012
BigCompute Series - Electronic Structure
The most widespread application of electron structure is in the rationalization of chemical properties, in both inorganic and organic chemistry.
The following are the most commonly-used computational methods for electronic structure:
- Valence bond theory
- Generalized valence bond
- Modern valence bond
- Molecular orbital theory
- Hartree–Fock method
- Møller–Plesset perturbation theory
- Configuration interaction
- Coupled cluster
- Multi-configurational self-consistent field
- Quantum chemistry composite methods
- Quantum Monte Carlo
- Linear combination of atomic orbitals
- Electronic band structure
- Nearly-free electron model
- Tight binding
- Muffin-tin approximation
- Density functional theory
- k·p perturbation theory
- Empty Lattice Approximation
Some of the popular electronic structure applications, that are typical challengers to any HPC systems, are:
- VASP
- ABINIT
- SIESTA
Update:
- 2012.07.12 - first post
Wednesday, July 11, 2012
Top Research Institutions Asking Congress to Preserve NIH Funding
Today, 19 leading U.S. medical research institutions nationwide are urging congressional leaders to maintain current funding levels for the National Institutes of Health (NIH). They say medical research funding is essential for the health of the nation and the economy. NIH-funded research happens in all 50 states, and occurs in large and small communities across the nation — strengthening the economy and creating jobs. It is estimated that every $1 million in federal research funding creates at least 16 jobs in a community.
The institutions also ask that a federal salary cap for NIH-funded researchers be returned to its previous level. The letter says the pay cut to scientists will drive some away and make recruitment of "the best and the brightest" even more difficult.
Most medical research at U.S. academic medical centers — whether laboratory studies or clinical trials — is supported by NIH dollars. That includes research in all key disease areas: all types of cancer; Alzheimer's disease; Parkinson's disease and other neurological disorders; heart and lung diseases; diabetes and obesity; genetic conditions; infectious diseases; mental illness and more. Almost any major advance or therapy in these areas has involved NIH funding.
In recent years, inflation and budget cuts have led to cuts in award amounts and the number of grants. In the past, one-third of grant applications received funding in many disciplines. Now only 12 percent in some areas win funding. Difficulty obtaining grants has driven some experienced researchers from their scientific careers and made many students think twice about research as a career choice.
Those signing the letter are:
Arizona State University
Cleveland Clinic Lerner Research Institute
Duke University School of Medicine
Henry Ford Health System
Mayo Clinic
New York University School of Medicine
Northwestern University Feinberg School of Medicine
Oregon Health & Science University School of Medicine
University of California Health System
University of Colorado Anschutz Medical Campus
University Hospitals Case Medical Center
University of Michigan Health System
University of Minnesota
The University of Texas at Arlington
The University of Texas System
University of Utah School of Medicine
University of Wisconsin School of Medicine and Public Health
University of Wisconsin System
Vanderbilt University
Text of the letter sent to the U.S. Congress follows:
The undersigned institutions are recognized as some of the leading research and academic medical centers in the United States. We appreciate the opportunity to bring to your attention two issues that dramatically impact the research function of our institutions:
*National Institutes of Health (NIH) — It is critically important that the NIH continues to be recognized as a national priority. We strongly urge your consideration of funding NIH at the $32 billion level in the FY 2013 Labor-HHS-Education appropriations bill.
*Executive Pay Scale/Salary Cap — The restoration of the salary limit imposed on extramural NIH researchers to Level I of the Executive Pay Scale is critical to our ability to attract and retain the most qualified researchers to our institutions and the field. We urge your inclusion of the salary cap restoration during your subcommittee's deliberations.
We trust you know of our deep commitment to delivering high-quality, patient-centered care to patients who come through our doors. At the same time, our organizations are conducting research to improve basic understanding of disease and wellness, to create innovations in health and health care delivery, and to discover new technologies and therapeutics that will advance medicine and our ability to heal.
We respect the difficult decisions you must make, and we urge you during those deliberations to view medical research as an investment in our nation's health, security and economy. NIH-funded research happens in all 50 states, and occurs in large and small communities across the nation — strengthening the economy and creating jobs. The investment in NIH not only improves the health of the nation, but also benefits our economy in the short term and long term.
An increase in the FY 2013 investment for NIH will provide researchers across our country the tools and resources necessary to continue their work to prevent, alleviate and cure diseases, including some of our nation's most debilitating and costly conditions. Equally important to our research function is the need to restore the salary limit imposed on extramural NIH researchers to Level I of the Executive Pay Scale. As you recall, the FY 2012 funding reduced the salary cap to Executive Level II. This is very important; this is a reduction of $20,000 (10 percent) to every NIH extramural researcher. While the reduction in the salary cap will have a negligible (less than 1 percent) impact on the NIH budget, it will have a dramatically disproportionate impact on individual investigators and their institutions.
As our country strives to improve basic, clinical and translational research, we must not hinder the recruitment and retention of our nation's most talented investigators, scientists and physician scientists. The United States must remain committed to improving the health of the nation and must be globally competitive in research and technology.
As a nation, we must value our scientists and provide them appropriate incentives to pursue and continue a career in biomedical research. We need the best and the brightest minds pursuing new knowledge that can be delivered into the clinical practice to help patients. Now is not the time to exacerbate the decline of medical research dollars and discourage our nation's most talented researchers.
We urge your careful consideration of our policy concerns and specific requests. We look forward to working with you on these very important issues.
Deep Computing for Industrial Competitiveness
Researchers at IBM and Lawrence Livermore National Laboratory (LLNL) announced June 27 that they are broadening their nearly 20-year collaboration in high performance computing (HPC) by joining forces to work with industrial partners to help boost their competitiveness in the global economy.
Under a recently concluded agreement, IBM and LLNL have formed an HPC collaboration called Deep Computing Solutions to take place within LLNL’s High Performance Computing Innovation Center (HPCIC).
Announced last June, the HPCIC was created to help American industry harness the power of supercomputing to better compete in the global marketplace (http://hpcic.llnl.gov).
Deep Computing Solutions will bring a new dimension to the HPCIC, adding IBM’s computational science expertise to LLNL’s own, for the benefit of Deep Computing Solution’s clients.
Links:
Monday, July 2, 2012
University Wyoming Picked IBM for HPC Solution
June 27 -- IBM has been chosen to design and build the University of Wyoming’s campus cluster or high-performance computing center, which may have as many as 100 UW faculty members using it for their computational science research starting this fall.
The campus cluster, formally known as the Advanced Research Computing Center (ARCC), will use approximately 150 square feet (for five racks of computer equipment) in the UW Information Technology Building. Its capacity will be roughly 3 percent of the 75,000 CPUs or core hours available to UW at the National Center for Atmospheric Science (NCAR)-Wyoming Supercomputing Center (NWSC) in Cheyenne. The CPU is essentially the brains of the computer, where most calculations take place.
“Overall, they (IBM) were the best value for the university,” says Tim Kuhfuss, UW’s director of research support for Information Technology. “Their price was competitive, and they certainly understand our view for a ‘condominium model’ and can deliver it.”
Nuts and bolts
Under a condominium model, the university will provide the basic infrastructure -- personnel to run it, basic networking and the basic computer architecture to keep it running -- for the campus cluster. In exchange, UW researchers will buy computing nodes (computers) or storage. That investment will come from faculty securing successful grant proposals, which are expected to include a request for funding for the computational resources needed for their particular research projects.
Under the broad strokes of the contract, UW will pay IBM $1 million for the initial hardware needed for the cluster. UW has budgeted $1 million annually in hardware in the second and third years of the contract, but may spend less or more depending on how much money UW researchers contribute.
The contract includes an option to renew annually -- in one-year increments -- for two additional years beyond the first three. IBM also offered the university access to its research divisions, which was “very attractive to our faculty,” Kuhfuss says.
“They were looking at this as a partner, not a customer-vendor relationship,” Kuhfuss says. “That was something that was important to us.”
“IBM’s Smarter Planet Initiative focuses on a number of industries like energy. Research partnerships with universities are one of the key incubators to implement improvements in a variety of industries,” says Kent Winchell, IBM’s deep computing chief technology officer. “The UW School of Energy Resources, combined with the new university-wide plan for high-performance computing, aligns with IBM goals for a smarter planet. There also is synergy with the recent NSF/NCAR supercomputer located in Cheyenne for climate and environmental science.”
Initially, seven companies bid for the project. That number was reduced to three before IBM was chosen, Kuhfuss says. From an architectural standpoint, any of the three finalists qualified, but it was IBM’s desire to be a partner rather than just a vendor that made the difference, Kuhfuss says.
IBM will develop and test the system in its development lab in Boulder, Colo., before delivering the hardware and storage racks to UW’s IT Building sometime in July. Kuhfuss expects the campus cluster to be operational between August and October.
Advancing computational research on campus
The campus cluster, nicknamed “Moran” after Mount Moran in western Wyoming’s Teton Range, will serve two purposes.
First, it will enable atmospheric and earth sciences faculty members -- who will be able to use the NWSC -- to learn what to expect with the software. The cluster provides the opportunity for that group of faculty members to work out issues caused by scaling up parallel algorithms from tens or hundreds of processors to thousands of processors, before moving up to tens of thousands of processors on the NWSC supercomputer.
Second, the cluster will provide a research resource for UW research faculty members -- such as bioinformaticists, social scientists, pure mathematicians and theoretical physicists -- whose research doesn’t fall within the scope of the NWSC.
Initially, Kuhfuss says there will be a trial month or “free-range period,” most likely October, when any UW faculty member can use nodes (one node essentially equals 16 desktop computers) on the cluster to conduct research. But there will be an organized resource allocation system created for ARCC, says Tim Brewer, end user support manager of research support for information technology, who reports to Kuhfuss.
Jeff Lang, a high-performance computing architect and administrator, who also reports to Kuhfuss, will handle on-site, day-to-day operations of the ARCC.
Winchell, who graduated with a computer science degree from UW in 1981, recalled his undergraduate days when UW’s Laboratory Information System (LIS) purchased CDC cyber-computer systems, which he said were state of the art at that time.
“Access to those systems created a passion in me for using IT to solve complex problems,” Winchell says. “It’s exciting to see UW keep up the tradition of providing state-of-the-art systems to researchers and students.”
Subscribe to:
Posts (Atom)