HPC News: Cray, EMC, Supermicro

It's been a busy week for the high performance computing industry. Here's some of this week's news announcements:

Cray awarded $16 million Defense Department contract.  Cray announced that the Department of Defense High Performance Computing Modernization Program (DOD HPCMP) has awarded a $16 million contract to Cray to significantly upgrade the program's three existing Cray XE6 supercomputers. Cray will double the number of processor cores in each of the three systems by upgrading the Cray XE6 supercomputers to the AMD Opteron 6200 Series "Interlagos" processors. These three systems will be combined into one petascale supercomputer, which will be located at the DSRC at ERDC. "Breaking through the challenges faced by the Department of Defense research, development, test, and evaluation communities requires increasingly powerful supercomputers in addition to the unique expertise that the HPCMP provides," said John West, director of the DOD HPCMP. "Cray has been a technology partner of the HPCMP for many years, and we look forward to continuing that relationship through the deployment of these systems." Cray also announced that its next-generation supercomputer code-named "Cascade" will be available with Intel Corporation's new Intel Xeon Phi coprocessors.

EMC expands VNX with HPC series.  EMC announced the new VNX HPC series, a preconfigured hardware and software appliance designed to simplify complex Lustre environments for commercial HPC customers. The VNX HPC series includes metadata and object servers, EMC VNX5100 metadata storage, VNX7500 object storage and the Terascala LustreStack software suite—delivering 10s of GB/sec bandwidth and scales from 720 Terabytes to 10's of Petabytes of capacity. "Lustre is one of the most powerful file systems on the market for handling HPC workloads and a natural fit for emerging commercial HPC applications," said Terri McClure, Senior Analyst at ESG. "But the management complexity, especially in fast growing environments, can be daunting. Many of these organizations don't have the resources to take full advantage of complex Lustre based implementations. EMC is helping to change that with its VNX HPC series solution, which is appliance based for easy deployments and integrated with Terascala for simple end-to-end management and EMC global support."

Supermicro launches FatTwin architecture.  At ISC 2012 Supermicro (SMCI) debuted FatTwin, a high-capacity powerful new server platform based on Supermicro’s highly successful Twin Architecture.  Its patented Twin architecture integrates independent systems side-by-side in a single chassis, multiplying compute density while reducing solution costs through shared resources such as the chassis, fans, power supplies, cabling and rack mount hardware. The architecture incorporates airflow optimized designs, air-shrouds and streamlined cabling which reduces fan speeds and power consumption resulting in improved power utilization and overall energy efficiency. “Energy consumption is one of the greatest challenges facing Data Centers as computing demands increase exponentially around the world,” said Charles Liang, President and CEO of Supermicro. “Supermicro is addressing this growing challenge by developing new generation server technologies such as FatTwin which maximizes performance with support of up to 135W processors while eliminating costly air-conditioning and cooling methods. The innovation of FatTwin is in its high-density, airflow optimized design that integrates up to eight dual-processor nodes in a standard 4U rackmount server and operates at temperatures up to 47°C. Our FatTwin platform will set new industry standards for performance and efficiency while offering lowered TCO and an accelerated return on investment.”