The Week in Review
Here is a collection of highlights from this week’s news stream as reported by HPCwire.
Dell and the University of Cambridge Launch HPC Solution Center
Dell is in the news again this week, following on the heels of last week’s partnership with the NASA Center for Climate Simulation (NCCS). This week, Dell is expanding the computational abilities of the University of Cambridge. The new Dell | Cambridge High Performance Computing (HPC) Solution Centre combines Cambridge HPC experience and Dell technology leadership to provide academic and private sector research organizations with cost-effective and accessible HPC solutions. The center has amassed a considerable talent pool and significant compute power aimed at solving “real world” challenges.
Dr. Paul Calleja, director of HPC service at the University of Cambridge, highlights the project’s goal of being accessible:
“The Solutions Centre has been founded with the overriding mandate of providing accessible research computing services and technology to organisations that would otherwise not have the money or expertise to benefit from such advantages, whether they are from academic or private sector backgrounds.”
Troy West, vice president and general manager of public sector EMEA at Dell, explained how Dell is aiming to make HPC available to a broader user base:
“Using open, standardised x86 server and storage architectures, Dell brings simplified and flexible HPC to a wider end-user audience and the Dell | Cambridge HPC Solution Centre is a shining example of what can be achieved. By donating our time and resources to this project, Dell is not only contributing back to the HPC community, but also giving customers the tools and guidance to adopt HPC solutions in a more efficient manner to drive down the cost and complexity traditionally associated with this highly technical section of IT.”
The center launched this week and is currently operational. Dell plans to use the Solution Centre as a base for its European HPC account teams to demonstrate its HPC technologies to customers and prospects in the region.
University of Groningen Selects Bright Cluster Manager
Bright Computing announced this week that the University of Groningen selected Bright Cluster Manager to manage its new 33 teraflops supercomputer, one of the Netherlands’ fastest. The system, named “Millipede,” was delivered by Bright Computing partner ClusterVision, a specialist in high performance compute clusters.
From the announcement:
The supercomputer runs Bright Cluster Manager — Advanced Edition, which is a Linux-based cluster management software solution that has built-in functionality for large and complex clusters, such as failover head nodes and multiple load-balancing provisioning nodes. Bright Cluster Manager’s wide range of monitoring and management functionality, combined with its easy-to-use command line and graphical user interfaces, will save the system administrators at the university substantial time in operating the cluster. The cluster will also run ScaleMP software to serve the needs for applications that require large amounts of memory, as well as the Fraunhofer Global Filesystem (FhGFS).
Professor Dr. Cees Sterks, general director of the Donald Smits Center for Information Technology at the University of Groningen, highlighted the importance of research at the University of Groningen, explaining that the new supercomputer will enable the university to offer a powerful service to users, helping them achieve their research goals.
Millipede sports 3,280 AMD Opteron processor cores and 132 terabytes of storage. Scientists at the university will use the system to further research in diverse fields such as informatics, chemistry, medicine and genetics.
Fixstars Releases ‘Yellow Dog Linux for CUDA’
GPUs are hot right now, and by extension so is CUDA, the parallel computing architecture developed by NVIDIA. So it was just a matter of time before Fixstars optimized its Yellow Dog Linux distribution for GPU computing. Yellow Dog Linux for NVIDIA CUDA version 6.2.1, released this week, includes more tools for CUDA developers, such as CUDA for Fortran and the Portland Group’s PGI accelerator.
From the release:
Yellow Dog Linux for NVIDIA CUDA v6.2.1 bundles NVIDIA’s CUDA SDK 3.1 and the updated packages found in RHEL/Centos 5.5. A whole host of other improvements and bug fixes have been made, including improved Intel chipset support, simplified NVIDIA toolkit version switching, as well as several improvements to Fixstars’ CUDA Plugin for Eclipse.
The release is now offered free to all users and can be downloaded here. More information on Yellow Dog Linux for CUDA is available here, and more information on CUDA programming is available via NVIDIA’s CUDA programming guide (PDF).