ISC13: NVIDIA Technology Powers Brain Simulator

After Monday's big announcement of Tianhe-2 becoming the world's most powerful supercomputer, the buzz continues at the International Supercomputing Conference in Leipzig, Germany. NVIDIA uses its GPUs to build an artificial neural network, Cray launches a x86 storage solution, and Intel, Mellanox and Xyratex all tout their impact on the Top500 supercomputers.

SGI and DoD launch supercomputer Spirit

SGI announced that the United States Department of Defense (DoD) has deployed the SGI ICE X high performance computing (HPC) system for its supercomputer Spirit, making it the 14th fastest supercomputer in the world according to TOP500. The SGI ICE X system has been deployed as part of the DoD's High Performance Computing Modernization Program (HPCMP), which provides compute resources for the Air Force Research Laboratory (AFRL) DoD Supercomputing Research Center (DSRC). ICE X products now comes with the newly-released Intel Xeon Phi 5120D coprocessors

"Our customers are already flocking to the fastest system in the Department of Defense, finding that their applications are performing significantly better on the new system," stated Jeff Graham, the director of the AFRL DSRC. "SGI has delivered a system that is exceeding their standard benchmark performance on DoD applications by an average of over 27%, which translates into higher productivity for scientists and engineers across the DoD."

NVIDIA GPUs to help build artificial neural network

NVIDIA (NVDA) announced that it has collaborated with a research team at Stanford University to create the world's largest artificial neural network built to model how the human brain learns. Google had previously developed a 16,000 CPU core large-scale neural network, and NVIDIA hopes to build one 6.5 times bigger. A team led by Stanford's Andrew Ng, director of the university's Artificial Intelligence Lab, created an large network with only three servers using NVIDIA GPUs to accelerate the processing of the big data generated by the network. With 16 NVIDIA GPU-accelerated servers, the team then created an 11.2 billion-parameter neural network.

"Delivering significantly higher levels of computational performance than CPUs, GPU accelerators bring large-scale neural network modeling to the masses," said Sumit Gupta, general manager of the Tesla Accelerated Computing Business Unit at NVIDIA. "Any researcher or company can now use machine learning to solve all kinds of real-life problems with just a few GPU-accelerated servers."

Cray launches x86 Linux Lustre storage solution

Cray  announced the launch of Cray Cluster Connect -- a complete Lustre storage solution for x86 Linux clusters. A compute agnostic storage and data management offering, Cray Cluster Connect allows customers to utilize their Linux compute environment of choice. The solution provides customers with a complete, end-to-end Lustre solution consisting of hardware, networking, software, architecture and support.

"Cray has a long, rich history in the HPC storage space, and we have built some of the largest and fastest Lustre file systems in the world," said Barry Bolding, Cray's vice president of storage and data management. "With Cray Cluster Connect, we are applying our Lustre expertise and innovation, and taking all that we have learned, developed and invested in parallel storage solutions to an expanded customer base. We can now deliver end-to-end, Lustre storage solutions for customers' existing x86 Linux environments. With the launch of Cray Cluster Connect, our storage and data management solutions are no longer limited to Cray supercomputer customers."

Companies showcase command over Top500 Supercomputers