"We've worked in tandem with Intel to optimize ANSYS software for the architectural changes in the new processor generation," said Wim Slagter, Lead Product Manager for HPC at ANSYS. “So customers can leverage HPC technology to incorporate more-complex models, whether using finer meshes or investigating nonlinear behavior.”
ANSYS 16.0 will support the shared-memory and distributed-memory parallelism of the Xeon Phi chips in both a Windows and Linux environment. Using this setup to reduce the simulation times should allow large and medium-sized engineering firms to run more simulations—enabling faster innovation and reduced time to market.
ANSYS has tested simulations comparing the previous Xeon generation of processors (Intel Xeon E5-2697 v2) with the current generation (Intel Xeon E5-2697 v3) coupled with Xeon Phi (7120P) co-processors. Each system was given the same workload on ANSYS 16.0. The results show that the solution time in ANSYS Mechanical was reduced by a factor of 3.1.
"Dramatic performance increases can transform entire engineering workflows," said Frank Soqui, general manager at Intel. "The collaboration between Intel and ANSYS has delivered just that, tremendous new performance levels, without altering the end-user experience or complicating the IT environment for great price/performance results."
Is reducing your processing time by a factor of 3.1 enough for your organization to buy a new HPC cluster? Will these jumps in speed become irrelevant in the era of cloud computing? Debate in the comments below.
Source Ansys.