Platform

High Performance Computing

  • Extended AMD GPU support
    • Benefit from more hardware options to leverage GPU-accelerated computation
      • AMD Instinct MI300 series is now supported
      • Includes MI300X
    • Additionally Radeon Pro W7x00 GPUs now supported
      • W7800 and W7900 are recommended
  • GPU-native Grid Sequencing Initialization
    • Up to 17% faster steady state simulations of vehicle external aerodynamics on GPUs with a GPU-native implementation of the Grid Sequencing Initialization (GSI) method for the Coupled Flow Solver​
    • Note: when running on GPU, the number of Maximum Grid Levels is internally set to 6
    • GPU-native segregated fluid isothermal and partial slip model
      • Significantly speed up (up to 5x faster) microfluidics and rarified flow applications with GPU-native segregated fluid isothermal and partial slip models​
        • Maxwell Slip​
        • Von Smoluchowski​
      • CPU-equivalent flow solutions ensured by maintaining a unified code base​
    • GPU-native S2S radiation model
      • Speedup (up to 8x faster) full vehicle thermal management (VTM) simulations with a GPU-native S2S radiation model
      • CPU-equivalent solutions ensured by maintaining a unified code base
  • Improved -benchmark options for GPU usage
    • Easier use of in-built benchmarking for coupled flow cases on low GPU counts
      • These cases may require more than one CPU process per GPU card
      • A single benchmark run can now be configured to run multiple sets of CPU processes per GPU
      • "-gpgpus" option added to allow separate benchmark inputs for each number of CPU processes evaluated
  • Clearer GPU error messaging for out-of-memory errors
    • Better understanding of error messages with more informative output
      • If a scenario is encountered where GPU memory has run out, GPU memory utilization is reported as a percentage and explicitly highlighted as an out-of-memory error
  • GPU-native Derived Part monitoring
    • Improved performance of Derived Part reporting/monitoring with GPU-native implementation
      • Valid for monitoring on probes and plane sections
      • Reduced turnaround time due to less CPU-GPU data migration
  • GPU Coupled Flow performance improvements
    • Reduced turnaround time for coupled flow problems thanks to AmgX enhancements
    • Up to 10% improvement for steady state automotive aerodynamics

Deployment

  • Scheduled Message Passing Interface (MPI) versions support changes for Simcenter STAR-CCM+ 2410
    • To be certified on Linux: Intel MPI 2021.12
    • To be certified on Windows: Intel MPI 2021.12
    • To be retired on Linux: Cray MPI 7

User Experience

  • Trigger sequences of Simulation Operations
    • Automate sophisticated workflows with the possibility to trigger simulation operations sequences
      • Trigger different sequences based on desired scenario
      • Better organized operations
      • Easier to troubleshoot
  • Print Simulation Guide
    • Quickly review and share information relevant to the simulation, with the possibility to print (also to PDF) the simulation guide
    • Check the simulation guide content without opening the simulation file