This technical talk will focus on the benefit of running Altair Radioss™ on 3rd Gen Intel® Xeon® Scalable Processor. Crashworthiness, impact and multiphysics simulations processed by Radioss are some of the most compute intensive numerical calculations. Such simulations are used in product manufacturability to assess and optimize product performance at the early phase of the design. For optimal performance, software like Radioss remains coded in Fortran language and therefore uses Intel® OneAPI compilers (classic Fortran & C/C++ compilers). It is highly optimized using tools like Intel® Advisor, Intel Trace Analyzer & Collector. Radioss takes advantage of AVX512 for vectorization as well as Intel® MPI Library and OpenMP for parallelization to reduce time to solution. In this presentation we will discuss the performance obtained using this new generation of Intel Xeon compared to previous version and focus primarily on how the additional memory bandwidth enables some breakthrough in term of performance improvement running Radioss. This allows to run bigger models, do more design iterations, and therefore accelerate the pace to innovation for Altair and Intel customers.
Expert in software optimization and parallelization on multi-core architectures and accelerators, Eric developed the Hybrid MPP parallel version of Altair Radioss finite element software. After receiving an MSc degree in computer science, Eric started his career in 1994, working at CNRS, then Mecalog and Altair since 2006. Since 2017, Eric is Vice President of Radioss Development and Altair Solver HPC. Eric also holds an Executive MBA from HEC business school obtained in 2007. LinkedIn profile: https://www.linkedin.com/in/ericlequiniou/