insideHPC.com

Congratulations to the University of Alberta for taking home the gold at the first annual Supercomputing Cluster Challenge.  Of all the competitors, U of Alberta didn’t have the fastest cluster design on paper.  However, good planning, preparation and a wonderful stroke of luck following a power outage on the show-floor powered the team to win the competition.

To all the teams …

Now that most of have finally recovered from last week’s annual Supercomputing show, InsideHPC has decided to compile a list of the various bits we [and others] covered during the show. This list is by no means comprehensive. If your event/press release/etc doesn’t appear here, we’re not picking on you. If you …

Always interesting to see an outside publication cover our little hot house ecosystem. Case in point: story at Wired, with pics.

So, I missed this on the show floor, but evidently IBM was showing off a liquid cooled version of a Power6 system. ZDNet’s George Ou has some details, with pics.

At the SC07 supercomputing conference (image gallery) in Reno Nevada, IBM previewed the liquid cooled 32-core 4.7 GHz System p 575 Power6 supercomputer last night at the opening gala.

…

In case you missed it, a team from Livermore and IBM won the Gordon Bell at SC07. Complete news from HPCwire.  Here’s the nut

Using groundbreaking computational techniques, a team of scientists from Lawrence Livermore National Laboratory and IBM earned the 2007 Gordon Bell Prize for a first-of-a-kind simulation of Kelvin-Helmholtz instability in molten metals on BlueGene/L, the world’s fastest

…

SC07: If you got one of the limited edition “My other computer is super” shirts at SC07 this year, take a snap of yourself wearing it and send it in. Pose in front of your organization’s supercomputer, with a hot co-worker, cool pet, or just hanging with the gang.

Bonus points if you’re one of the guys from a distant …

SC07: During the conference this week UK cluster company Concurrent Thinking announced a new series of appliances meant to make building HPC clusters snap together easy. From the release

Concurrent Thinking’s cluster appliances offer a simple and cost-effective solution to cluster deployment, management and support. Developed around an extensible service-oriented architecture and web services, individual appliances provide key services such

…

Sc07: Univa UD unveiled a new online community for users, administrators, and developers interested in an open source grid and cluster stack.

Grid.org is designed as a destination for community members who want to connect easily and productively with those who have similar interests and who want to engage in a vibrant, functioning community of active participants. At Grid.org, community

…

SC07: From SC Online we learned that a new book was announced at SC07 this week called “Petascale Computing: Algorithms and Applications,” edited by David A. Bader.

“High-performance computing will enable breakthrough science and engineering in the 21st century,” said Bader. “My goal in developing this book was to inspire members of the high-performance

…

SC07: The team lead by Indiana University with members from Technische Universitaet Dresden, Rochester Institute of Technology, Oak Ridge National Laboratory and Pittsburgh Supercomputing Center, took first place in the Bandwidth Challenge at SC07. Competitors in the Bandwidth Challenge were challenged to create methods for “utilizing a high-speed network path to support end-to-end network applications running across a grid …

SC07: From HPCwire we learn that Platform has a major rev of its Open Cluster Stack in store for users by the end of December

This new version of Platform OCS is built on an open source toolkit for cluster deployment, provisioning and configuration, providing users with maximum reliability, speed and flexibility.

…

SC07: As we mentioned would happen back in September, ClearSpeed has announced its ClearSpeed Accelerated Terascale System (CATS), 1 TeraFLOP in 1U

This small format system, taking up just 1U in a standard server rack, delivers up to one TeraFLOP of performance, achieving the world’s highest ever compute density to date.

At the conference, ClearSpeed is demonstrating 12 CATS, each connected

…

SC07: NVIDIA has announced a new version of its CUDA API for harnessing the power of GPU’s in computation.

This latest version of the CUDA software suite includes support for 64-bit Windows XP and starting with the release of CUDA 1.1, NVIDIA’s standard display drivers will include the CUDA driver, eliminating

…

SC07: DataDirect Networks today announced the release of their 8th generation Silicon Storage Architecture [S2A]. The S2A9900 will deliver a sustained bandwidth of 6 GB/s [yes… gigabyte!] per appliance. The 9900 is the first storage system to support 8Gbps Fibre Channel as well as 20 Gbps DDR host connections. The system …

SC07: The MathWorks announced yesterday that they’ve added support for multithreading on multicore processors

MATLAB now includes support for multithreaded computation for multicore systems and 64-bit Solaris platforms. Distributing Computing Toolbox now offers capabilities for developing applications that interleave parallel and serial code and interactively prototyping parallel algorithms on a desktop computer by running four local MATLAB sessions.

…”Our group of biochemists,

…