Sign up for our newsletter and get the latest HPC news and analysis.

Nvidia to Offer Full CUDA Support for OpenPOWER in 4Q2014

sumit

“Nvidia is adding CUDA software support for GPUs with IBM POWER CPUs. IBM and Nvidia are demonstrating the first GPU accelerator framework for Java, showing an order of magnitude performance improvement on Hadoop Analytics applications compared to a CPU-only implementation.”

Video: Dynamic LNET Config

lustre-logo2

In this video from LUG 2014, Amir Shehata from Intel presents: Dynamic LNET Config.

Windows RDMA File Storage

Talpey

“Utilizing Mellanox’s RDMA-based interconnect solutions and Microsoft Windows Server 2012 R2, end-users can increase their application performance, improve their data center system efficiency, as well as consolidate their server and storage network infrastructure over a single-wire network for optimal CAPEX and OPEX savings.”

Video: Performance Analysis and Optimizations of CAE Applications

Fisnik Kraja

In this video from the HPCAC Swiss Conference 2014, Fisnik Kraja from science + computing ag presents: Performance Analysis and Optimizations of CAE Applications.

Xyratex Update from LUG 2014

xyratex

“We wanted to send out a special thank you to Seagate for showing such a strong commitment to Lustre. Their announcement at LUG – intent to transfer ownership of lustre.org to the community – is great news.”

Penguin Ups GPU Density with Relion 2808GT

relion

“The Relion 2808GT was designed to deliver maximum compute power with a minimal datacenter footprint and maximum efficiency. The system supports eight GPUs or coprocessors in two rack units and provides a higher compute density that any other server on the market. Configured with eight NVidia GPUs or eight Intel Xeon Phi co-processors a single Relion 2808GT can achieve over 8TFLOPs of double precision floating point peak performance.”

OpenFabrics Interfaces Working Group Architecture

grun

“Verbs and RDMA-based networks will continue to be important focus areas, but with the new OFWG, the OFA is expanding its focus to include additional application-centric APIs that could be better suited to other areas and technologies,” said Jim Ryan, chairman, OFA. “Application-centric I/O is an extension of the approach historically taken by the OFA, and we will use our expertise in this area to guide the open source development of APIs and I/O libraries.”

NVM as a Disruptive Technology

nvm

NVM is coming. In this series of videos from the OpenFabrics International Developer Workshop 2014, Jim Ryan and Doug Voight describe Non-Volatile Memory as a disruptive technology for HPC.

Thomas Sterling Presents: ParalleX Execution Model for Exascale Astrophysics

Thomas Sterling

“This talk describes an experimental methodology, ParalleX, that addresses Exascale challenges through a change in the fundamental model of parallel computation from that of the communicating sequential processes (e.g., MPI) to an innovative synthesis of concepts involving message-driven work-queue execution in the context of a global address space.”

Video: Paul Grun Kicks Off the OpenFabrics Workshop 2014

paul grun

In this video from the OpenFabrics International Developer Workshop 2014, Paul Grun from OFA presents: Welcome, Introductions and Workshop Goals. “The 10th Annual OpenFabrics International Developer Workshop held on March 30-April 2 offered a multi-day event dedicated to the development and improvement of OpenFabrics Software (OFS). The Workshop kicked off Sunday, March 30, with a keynote from Dr. Thomas Sterling, Executive Associate Director and Chief Scientist at CREST. Presentations began the following morning covering topics focused on I/O for Exascale systems and Enterprise applications and included interest areas such as distributed computing, storage and data access and data analysis applications.”