T-Platforms, the Russian HPC services and systems manufacturer that fielded a system at number 12 on last November’s list, announced today from ISC that they are expanding their presence into Western Europe with an office in Hanover, Germany. T-Platforms is also collaborating with PRACE and STRATOS workgroups on initiatives to spur development for the entire […]
Archives for May 2010
insideHPC puts its ISC clothes on
May 30, 2010 by
We’ve officially started our dedicated coverage of the International Supercomputing Conference in Hamburg, Germany. When we asked you many moons ago what events you wanted to see us spend more time on, the ISC was near the top of your requests. This year we are very excited to finally be able to make that happen. […]
Link and Run for 05/29/2010
May 29, 2010 by
OpenCL Optimization Case Study: Diagonal Sparse Matrix Vector Multiplication This article discusses performance optimizations for AMD GPUs and CPUs using as a case study a simple, yet widely used computationally intensive kernel: Diagonal Sparse Matrix Vector Multiplication. We look at several topics which come up during OpenCL performance optimization and apply them to our case […]
COMPETES Reauthorization passes House, finally
May 28, 2010 by
Today I received word by email from Congressman Dan Lipinski’s office (Democrat from Illinois 3) that the COMPETES reauthorization act finally passed as H.R. 5116 (not as 5325, Rep. Gordon’s Hail Mary attempt to get around the Republicans evil partisan play to kill the bill by turning it into a vote about federal employees looking […]
Microsoft unwraps technical computing initiative, leaves most to the imagination
May 28, 2010 by
In an email posted on the “executive e-mail blog” earlier this month Microsoft Server and Tools Business president Bob Muglia announced Microsoft’s Technical Computing Initiative. That first announcement and flurry of press coverage was nearly devoid of any articulated strategy or accomplishment. So I dug into this with Microsoft to find out what’s going on in Redmond and what they are really thinking about with this initiative.
While it is still very clear that Microsoft has the bulk of the work on this ahead of them on this — including the definition of a clear strategy for actually having an impact — I did find reason to believe that something good may happen here. Read on to find out why.
Node down? Toss it.
May 28, 2010 by
Doug Eadline explores the idea of tossing inexpensive cluster nodes when they go down instead of repairing them Designing a cluster with slower processors means that the node cost can be quite low. Of course there are interconnect issues, but for the sake of my argument, let’s assumed [sic] we are building a cluster out […]
Intel picture gallery: from sand to silicon
May 28, 2010 by
Intel has posted a new set of 40-ish pictures that follow a pile of sand on its journey to becoming a chip. Entertaining.
Bull powers up new super, outpaces JUGENE as Europe's leading super
May 28, 2010 by
Bull announced yesterday that they have fired up the French Atomic Energy Authority’s (the CEA) new 1.25 peak PFLOPS super The result of a collaborative program between Bull and the CEA which began in 2008, Tera 100 is the first petaflops-scale supercomputer ever designed and developed in Europe. Its theoretical maximum power of 1.25 Petaflops […]
Euro project aims to simulate life, the universe, and everything
May 28, 2010 by
(With apologies to Douglas Adams.) Science Daily reported this week on a new effort in Europe to simulate life on Earth, and although they haven’t yet added “the universe and everything” it is still pretty freaking ambitious Scientists are planning to use the largest supercomputers to simulate life on Earth, including the financial system, economies […]
Mellanox officially outs NVIDIA GPUDirect, NVIDIA says only the first step
May 28, 2010 by
When I reported on the PFLOPS super that Chinese Academy of Science has built with NVIDIA GPUs (the new Fermi-enabled Teslas), I mentioned it was using NVIDIA’s GPUDirect technology to improve the performance of GPU-to-GPU transfers. Mellanox is the lead partner with NVIDIA in developing that technology, which according to information I got in conversations […]
