ANSYS Expands HPC Benchmarking Program

By DE Editors

An enhanced HPC benchmark suite from ANSYS can assess performance on the latest HPC systems and technologies. This 4-million cell CFD benchmark model from ANSYS was used by its partners to demonstrate performance on cluster sizes up to 64 core. Each color represents a partition for parallel processing, with each partition mapped to a separate core.

ANSYS, Inc. (Southpointe, PA) announced an enhanced high-performance computing (HPC) benchmark suite that can guide customers who are adopting HPC systems for simulation. Used by ANSYS partners to assess performance of software from ANSYS on the latest HPC systems and technologies, the improved benchmark suite provides updated simulation workloads for structural analysis and CFD, exercising a simulation solution methodologies, and addressing bigger simulation tasks.

Including 15 simulation models in the mechanical modeling area and seven in the fluids modeling area, the suite provides customers with performance metrics for a variety of model sizes and solver methods, while reflecting the trend toward larger models and comprehensive multiphysics simulation.

Performance results for the suite of benchmarks have been generated by AMD, Intel, Mellanox Technologies, and other OEM partners — check out the ANSYS website for these benchmarks. The posted results demonstrate the performance scalability of software from ANSYS on the latest HPC technologies, including new quad-core processors from Intel and AMD, as well as the new ConnectX InfiniBand adapters from Mellanox. The company says the results yielded improvements to distributed memory parallel solvers and I/O speed in the ANSYS 11.0 and FLUENT 6.3 releases.

Intel used the new ANSYS benchmark suite to demonstrate the performance of its Intel Xeon Processor 5400 series, released in November 2007.

For more information, go to ANSYS, Inc.

Sources: Press materials received from the company and additional information gleaned from the company's website.