Archives for July 2011

Podcast: Hot Interconnects Conference Looks to the Future of Networks

In this podcast, the Co-Chairs on the Hot Interconnects Symposium discuss their program agenda and why Hot Interconnects continues to be the premier international forum for high performance networking at all scales. The event takes place August 24-26 in Santa Clara, CA. Two days of technical sessions led by Fabrizio Petrini from IBM TJ Watson, […]

Former NASA CTO Rolls Out Cloud Appliance

This week Chris Kemp, the former CTO of NASA and co-founder of OpenStack unveiled a new appliance designed to make it easy for enterprises to build private clouds. Kemp is now CEO of Nebula, a cloud appliance startup that wants to enable small enterprises to adopt that same architecture that we’re seeing in large Internet […]

White Paper: Exascale Research: Preparing for the Post-Moore Era

This is an important piece co-authored by Marc Snir, William Gropp and Peter Kogge.

We know this has appeared in several other places, but we have also come across a number of people who weren’t aware of it, so as a service to our readers, we decided to include this.

View as a PDF

Video: 10gen Delivers Agility with Open Source MongoDB

httpv://www.youtube.com/watch?v=xwzwzHa6wTQ In this video, 10gens’s Jared Rosoff and Max Schireson discuss how the MongoDB database software delivers agility and scalability for the enterprise. MongoDB is a scalable, high-performance, open source, document-oriented database. Founded by former DoubleClick Founder and CTO Dwight Merriman and former DoubleClick engineer and ShopWiki Founder and CTO Eliot Horowitz, 10gen offers production […]

Report: Accelerating HPC Apps Through MPI Offloading

In this guest feature from the HPC Advisory Council, authors Gilad Shainer, Tong Liu, Pak Lui, and Richard Graham explore the advantages of offloading MPI collectives communications from the CPU to the cluster interconnect on various applications’ performance.

Abstract
In the past, performance tuning of parallel applications could be fairly accomplished by separately optimizing their algorithms, communication, and computational aspects. However, as we continue to scale future larger machines, these issues become co-mingled and must be addressed comprehensively. MPI collectives communications are frequently being used for processes synchronization and their performance is critical for scalable, high-performance applications. Optimizing collectives communication performance can be achieved by offloading these communication to the network therefore minimizing the negative effect of system noise and jitter as well as separating them from the rest of the CPU activities. Throughout application profiling we can identify applications which will greatly benefit from such offloading and can determine the associate performance and productivity benefits.

"Memory Advantage Program" to Help Researchers App-Up on World's Largest Memory Super

To help researchers take advantage of Blacklight, the largest shared-memory system in the world, the Pittsburgh Supercomputing Center (PSC) has launched a Memory Advantage Program. Shared memory” means that a system’s memory can be directly accessed from all of its processors, as opposed to “distributed memory” (in which each processor’s memory is directly accessed only […]

GPU.NET Version 2 Allows Developers to Code in Any .NET-Supported Language

This week TidePowerd announced the Version 2 release of their GPU computing solution for the .NET framework, GPU.NET. The platform allows developers to quickly write GPU-accelerated applications completely in .NET-based languages. Our core goal is to provide developers with powerful, yet easy-to-use tools for numerical and high-performance computing (HPC) – and to provide those tools […]

Gluster Rolls Out First Unified File and Object Storage for OpenStack

John Kreisa writes that today’s release of the Gluster Connector for OpenStack is big news for the cloud. So what are we about to release for OpenStack? This line from the press release says it all: “The Gluster Connector for OpenStack …supports the virtual motion of the VMs within the OpenStack compute environment.” This is […]

Job of the Week: Systems Software Developer at DE Shaw

D.E. Shaw is seeking a Systems Software Developer in our Job of the Week: Exceptional software developers sought to design, develop, and deploy systems software for a special-purpose supercomputer. Successful hires will join an interdisciplinary research group pursuing an ambitious, long-term project aimed at achieving major scientific advances in the fields of biochemistry and molecular […]

Open Cloud Initiative Launches at OSCON

This week the non-profit Open Cloud Initiative (OCI) announced its official launch at the OSCON 2011 Open Source Convention. The primary purpose of the Open Cloud Initiative (OCI) is to define “Open Cloud” by way of community consensus and advocate for universal adoption of Open Standard formats and interfaces” said Sam Johnston, founder and president. […]