GPU Cluster for Molecular Modeling

February 2011 - The latest cluster to come out of Nor-Tech's lab is a Rocks/Linux-based system designed for use in molecular modeling. The cluster, commissioned by a major Midwestern university, features the NAMD simulation package as well as CHARMM and InfiniBand.

NAMD (Nanoscale Molecular Dynamics) is an open source parallel molecular dynamics simulation package based on Charm++. Designed to conduct high-performance simulation of large biomolecular systems, NAMD scales to a multitude of processors on each cluster.

The cluster also uses NVIDIA Tesla Fermi cards, which enable seamless integration of GPU computing with host systems for high-performance computing and large data center, scale-out deployments.

In addition to GigE, Nor-Tech's GPU Cluster for Molecular Modeling uses InfiniBand to flow data between its processors. InfiniBand offers significantly higher throughput low latency as compared to GigE.

Rocks in an open-source Linux cluster distribution for HPC clusters developed by NPACI and SDSC. Rocks includes integral applications that tie together the components of a cluster.

GPU acceleration dramatically increases processing power:
Nor-Tech's benchmark results show how dramatically GPU's can accelerate applications compared to CPU only configurations.

More about GPU Clusters

GPU computing refers to the use of a GPU (graphics processing unit) to accelerate scientific and engineering computing. GPUs are quickly gaining industry prominence due largely in part to the ease of programming of the CUDA parallel programming model. The massively parallel CUDA architecture, developed by NVIDIA, consists of 100s of processor cores

Because GPUs appear on video cards and motherboards, GPU clusters can combine a CPU and GPU for numeric computation, so that portions of an algorithm can run on the CPU while the highly parallelizable parts are ported to a GPU.

Nor-Tech has extensive experience developing and deploying GPU Clusters, and worked with the ORNL Future Technologies Group to improve their benchmarking software SHOC for GPU’s. SHOC is a collection of benchmark programs that test performance and stability of systems using computing devices with non-traditional architectures.

Check out these links for more information:

Nor-Tech deploying GPU cluster to Ireland for micromagnetic modeling

GPU Clusters maximize power via GPU acceleration

Further reading: Nor-Tech HPC Clusters | Nor-Tech Case Studies | More about Nor-Tech

Talk to a Nor-Tech representative: Todd Swank 901 Cliff Road, Burnsville, MN 55337 Toll free: 877-808-7438 | Local: 952-808-1018 Send an email