By Matthew Dublin

Earlier this week at a user group meeting in Fairbanks, Alaska, Cray announced that they have officially joined the "hybrid supercomputing" bandwagon with the rollout of their Cray XK6 supercomputer. The new Cray XK6 is comprised of their patented Gemini interconnect — their not-so-secret secret sauce — and high-performance AMD CPUs and Nvidia Tesla GPUs. If ordered with all the bells and whistles, the XK6 is capable of more than 50 petaflops of computing power.

The Swiss National Supercomputing Centre in Manno, Switzerland, is the first XK6 customer this week, upgrading from their Cray XE6 system called "Piz Palu." Researchers at CSCS have used Cray systems for quite some time for a range of disciplines, including biology, genetics, and experimental medicine.

According to CSCS director Tomhas Schulthess, "Given the remarkable interest in GPU technology from the Swiss computational science community, it is essential that CSCS adopt this technology into its high-end production systems soon. However, we are not looking for another GPU based stunt to place high on any Top500 lists. The Cray XK6 promises to be the first general-purpose supercomputer based on GPU technology, and we are very much looking forward to exploring its performance and productivity on real applications relevant to our scientists."

Part of the pitch for the new systems is that Cray XT4, Cray XT5, Cray XT6 or Cray XE6 systems are easily upgradeable to the Cray XK6 system, which is expected to be available in the second half of 2011. It can also be configured in a single cabinet or a multi-cabinet system with tens of thousands of compute nodes.

By Matthew Dublin

The National Center for Supercomputing Applications announced last week the roll-out of a new 153 teraflop supercomputer dubbed Forge. The new system has a "hybrid" architecture that combines GPUs and CPUs, and is set to replace NCSA's previous hybrid system called Lincoln.

Forge will combine 18 Dell blades that contain 36 nodes of dual-socket/eight-core AMD processors, Nvidia GPUs — including eight Fermi GPU units for each node, for a total of 288 —and will use an InfiniBand interconnect fabric. The system will have 700 terabytes of file system space with an I/O bandwidth its designers hope will surpass 16 GB per second at full operational performance.

The new cluster will be housed in a cutting-edge 20,000-square-foot machine room at the University of Illinois' National Petascale Computing Facility.

Forge's predecessor, Lincoln, was NCSA's first foray into large-scale hybrid core computing and its track record so far has validated the concept of hybrid-core cluster computing for life sciences research. Klaus Schulten's research team at the University of Illinois at Urbana-Champaign, used Lincoln to run NAMD to study the organization and function of proteins and protein complexes within cells. Schulten's team found that two of Lincoln's GPUs were equivalent of 24 of Lincoln's CPU cores, while eight of its GPUs were equivalent to 96 CPU cores. And the University of Utah's Thomas Cheatham, who is studying how proteins behave in solutions and how drugs interact with them, was able to accelerate AMBER on the system, achieving 15 times speedup per node.

As of July 1^st, the new system will be allocated through the National Science Foundation Translational Research in the Academic Community process.

By Matthew Dublin

A little over a year after announcing the start of their life sciences division, hybridcore hardware maker Convey Computer has rolled out a new addition to their bioinformatics suite aimed at accelerating common bioinformatics applications. The new Convey GraphConstructor (CGC) is a hardware and software solution designed for de Bruijn graphs which are used in short-read genome assembly applications like Velvet and Abyss. According to their announcement, the CGC is capable of providing users with performance speed-ups of 2.2 to 8.4 times using Convey’s FPGA-Intel x86 processor combo and a parallel memory subsystem.

The use cases for Convey’s CGC include the Department of Energy's Joint Genome Institute researchers sequencing and analyzing 268 GB of metagenomic DNA from microbes incubated in cow rumen. With the CGC, the researchers were able to speed up the discovery process by as much as 2.8 times and reduce the memory requirements by roughly 83 percent. The results so far have been the discovery of nearly 30,000 new enzymes that could improve biofuel production.

It’s worth noting that Convey’s “hybrid solution” is essentially a standard x86 processor with an FPGA as a “co-processor” - certainly not a novel combination for bioinformatics. Possibly a better name for what Convey’s hardware does is “coherent co-processing” as the secret sauce is their specific implementation architecture where they've placed the FPGA very close to the CPU so that both processors work in lockstep with the same memory system model.

By Matthew Dublin

Diagnosing performance issues in an IT infrastructure is never easy and as we’ve seen with the recent Amazon cloud crash, these issues are even more difficult to track in a cloud computing environment. But a new technology by AppNeta may help IT managers better obtain information about servers, networks, and desktop performance.

AppNeta's PathView microAppliance is being touted as an easy-to-use network administration monitoring tool that’s the size of a cell phone. It can be implemented at remote locations where you want to monitor network activity, requires very little power, and uses a standard Ethernet connection. Once the PathView microAppliance is plugged in, it uploads network performance data to an AppNeta cloud server.

John D. Halamka, chief information officer of Beth Israel Deaconess Medical Center and chief information officer at Harvard Medical School, writes on his blog Life as a Healthcare CIO, “as we all roll out EHRs to small provider offices, often with challenging internet connections, remote monitoring of cloud network performance becomes even more critical… Currently we have deployed these devices at our central EHR private cloud site and two of major remote practices The level of detail and depth of available metrics and reports is amazing. A low cost, zero administration, cloud-based, network sniffer that is truly plug and play. That's cool!”

As the post points out, for electronic health records cloud providers, configuration is simple and you can send it via UPS to a provide practice with installation instructions and gather performance data sans complex onsite network sniffing setups.

By Matthew Dublin

Earthsky.org has a podcast by Aleksei Aksimentiev, an assistant professor at the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign, describing his approach to personalized medicine with supercomputers.

90 second podcast:

8 minute podcast:

Aksimentiev is using the Ranger supercomputer at the Texas Advanced Computing Center to develop cheap DNA sequencing using nanopores. He and his colleagues are using Ranger to simulate in atomic detail the process of DNA transfers through these nanometer pores and develop a numerical model of the nanopore sensor. Recently, they carried out the first-ever atomistic simulations of DNA translocation through synthetic nanopores.

By Matthew Dublin

Harvard Medical School and Brigham and Women's Hospital are using IBM business analytics technology to study the effectiveness and potential safety of prescription drugs. Using IBM's Netezza data warehouse appliance, Harvard researchers are conducting pharmacoepidemiology studies to analyze data from millions of de-identified patient records that include insurance claims data to develop novel data-intensive drug safety research methods. Netezza data warehouse appliances architecturally integrate database, server and storage components into a single unit.

Netezza's appliances use a proprietary Asymmetric Massively Parallel Processing architecture that combines open blade-based servers and disk storage with a proprietary data filtering process using field-programmable gate arrays.

"We wanted a computing platform with massive analytics power, but was extremely simple to administer," said Sebastian Schneeweiss, Associate Professor of Medicine, Harvard Medical School and Vice Chief of the Brigham & Women's Hospital Division of Pharmacoepidemiology and Pharmacoeconomics in a release. "As global health care evolves toward a learning healthcare system with a need for ongoing comparative effectiveness and safety research integrated in routine care, it is imperative that research methods evolve in parallel. IBM Netezza will accelerate our ability to devise, test and publish new computationally intensive algorithms applied to ever larger longitudinal healthcare databases that we hope will become the gold standard for researchers globally."

This project will also use use the Netezza technology to study economics and outcomes research once the effectiveness of a particular medication is established. Pharmaceutical companies and health insurers will have teams of HEOR analysts then demonstrate the value of new medical products for pricing and coverage decisions to provide higher quality of care with the hope that by intelligently mining claims data with a powerful analytics platform like IBM Netezza, they may be able to provide a faster way to answer questions about the most effective therapies, which has obvious economic benefits for both patients and health care providers.

Using business data architectures and IT technologies is not uncommon in life sciences research. In the journal Bioinformatics, a paper by Lauren Boyd and her colleagues entitled "The caBIG Life Science Business Architecture Model" describes how the cancer Biomedical Informatics Grid, or caBIG has adopted "Business Architecture Models," or BAMs, models that describe what a business does and how activities are accomplished, in order to establish a "Life Science BAM," or LS BAM. The caBIG LS BAM provides a shared understanding of vocabulary and processes common in life sciences research. The latest version of the caBIG LS BAM includes 90 goals and 61 researchers within "Use Case and Activity Unified Modeling Language" UML diagrams.

The authors report that "to facilitate system interoperability across the cancer research enterprise, the LS BAM (and the LS DAM) can be used beyond caBIG to support their requirements definition efforts. Common business and information models provide a consistent understanding of business processes and data to be collected and exchanged, which should lower the barrier to interoperability. The model may therefore be a resource to train software engineers, facilitate development of standards and underpin software validation in many organizations, such as cancer centers, commercial tool providers or other large research institutions."