By Matthew Dublin

Finland has invested $6.7 million in a national project called Biomedinfra, which aims to improve biobanking, translational medicine, and high-performance computing infrastructures. The hope is that by implementing a robust biomedical IT infrastructure, not only will health care in the country be improved, but international collaborations will have a seamless way to collect and utilize massive amounts of data.

According to Anu Jalanko, head of the Public Health Genomics Unit at Finland's National Institute for Health and Welfare, the new funding will play a key role in establishing the Finnish national biobank network, BBMRI.FI. "Biobanking requires close collaboration locally, nationally and internationally. Finland has strong traditions and expertise in biobank-related research and has created several world-class population biobank resources that are often also used in international collaborations," he says.

Tommi Nyrönen, who leads the development Finland's CSC- IT Center for Science where they are currently developing cloud-based high-performance computing and storage services tailored to the needs of biomedical researchers, the new funding will enable them "to develop technologies and policies to ensure that distributed biomedical data can be managed and used efficiently and securely. Research institutes should be able to access supercomputing resources easily and securely."

By Matthew Dublin

In yet another example of how crowd sourcing and gaming can advance life sciences research, a group from Carnegie Mellon University and Stanford University has developed an online game to help non-experts uncover RNA design principles. EteRNA scores players based on how well their virtual designs can be rendered as real, physical molecules.

The goal of the project is to help investigators design RNA knots, polyhedra, and other shapes that have not be identified before. At the end of each week researchers synthesize the top designs in order to determine if the resulting molecules fold themselves into the three-dimensional shapes predicted by computer models.

"Putting a ball through a hoop or drawing a better poker hand is the way we're used to winning games, but in EteRNA you score when the molecule you've designed can assemble itself," says Adrien Treuille, an assistant professor of computer science at Carnegie Mellon and co-leader of the EteRNA project. "Nature provides the final score — and nature is one tough umpire."

Treuille also helped to design Foldit, another online game where players compete against each other to find new protein folding sequences.

In December, researchers from McGill released a similar online game called Phylo, which challenges players to find and compare similar regions of genetic sequences from several genomes.

And for those of you who would like to let all of your friends know what a great RNA designer you are, EteRNA is also integrated with Facebook.

Here are some videos made by Treuille and his colleagues to better acquaint you with EteRAN:

What is the EteRNA game?

What have we learned from EteRNA?

How was EteRNA created?

By Matthew Dublin

Dell’s John Fragalla has a series of informative posts that will walk you through some of the basic issues related to power and cooling in the data center. Everybody that runs a cluster or data center by now knows the mantra of an increasing number of multicore chips and GPUs, more storage arrays, and network switches, results in higher system densities (more and more gear crammed into rack units) which in turn results in more of a power drain to keep both the system running and keep it cool.

Fragalla suggests that the first step IT managers can take in grappling with these issues is to become intimately acquainted with the specific power and cooling requirements by measuring power consumption with LINPACK and familiarize themselves with metrics such as Amps, Cubic Feet per Minute (CFMs), Tons of Cooling, and British Thermal units per hour (BTUs/hr).

In part three of his series of cooling and power, he discusses rack Power Distributions Units (PDUs), which come in various configurations, ranging from single phase power, to three phase power, and different amps per phase, such as 30A to 60A.

In his next post, Fragalla will introduce readers to Cubic Feet per Minute (CFM), which is a measurement of air flow a HPC system produces from component fans.

By Matthew Dublin

Just when you thought you were safe from having to wrap your head around yet another new type of hardware, AMD announces a processing architecture that aims to bridge the bottleneck between interconnects on CPUs and GPUs called the Accelerated Processing Unit or APU.

The APU is an attempt to allegedly help the field of heterogeneous computing (which is usually FPGA co-processors couple with standard CPUs) reach its true potential. Hetergeneous computing has gained an increasing amount of momentum over the last two or three years due to the widespread popularity of general purpose computing with GPUs, which everyone wants to integrate into their clusters or workstation so they can be part of all the excitement...

In the video below, AMD explains how this new technology, which looks to be initially integrated on workstations and desktop computers, works and why this is idea worth marketing. Like every great new hardware idea, this looks great on paper (on a white paper that is) and in an animated promotional video with futuristic sounding music, but only the programming Gods know what hell may await those tasked with really getting some use of this new gizzmo.

By Matthew Dublin

The National Center for Supercomputing Applications has ported TeraChem 1.41, the open source ab initio quantum chemistry package, to run on the center's hybrid CPU/GPU supercomputer Lincoln.

"Before users were on their own to port their software to graphics-processing units," said Thom Dunning, the center's director. "By providing TeraChem on Lincoln, we've made it much easier for chemists to harness the power of GPUs to accelerate their calculations and improve their productivity.

TeraChem was initially developed by Stanford University's Todd Martinez, although NCSA's Alexey Titov has implemented d-functions for the energy calculation within Hartree-Fock and DFT methods that will be available in the next production version of TeraChem.

Lincoln is a comprised of both Nvidia Tesla S1070 accelerator units and Dell PowerEdge 1950 servers.

No mention is made of what the process of successfully porting TeraChem across a hybrid cluster is like, what types of snags they ran into along the way, or if it was fairly trivial, but certainly employing GPUs as accelerators in this capacity looks promising.

According to Nvidia, a workstation with 4 GPUs running TeraChem outperforms 256 quad-core CPUs running GAMESS.