Any relatively detailed structural mechanics model – e.g., mesh size in excess of ~500K DOF – is going to take a very long time to solve on a single CPU core. ANSYS HPC reduces the time to solution significantly. For larger models (in excess of ~10 MDOF) parallel processing with ANSYS HPC is needed to make the simulation feasible; a single CPU core typically cannot access enough computer memory to handle these models.
- Reducing turnaround time
- More design evaluations per day
- Simulating larger or more complex models
- Good scalability of distributed solvers up to 128 cores and above.
- A large variety of structural analysis problems can benefit from hardware accelerators (i.e. NVIDIA GPUs and Intel Xeon Phi coprocessors). Besides the sparse direct solvers, the PCG iterative solvers are – for example – also supported by GPUs.
- Hardware accelerators bring a strong speed-up to bulky 3D, linear models possibly involving (large) 3D assemblies, and bonded contact (welded, glued or bolted connections).
- Each HPC license can enable either a CPU core or a GPU socket (or coprocessor). This adds a lot of value to HPC in terms of flexibility and speed – i.e., get the highest speedup with whatever combination of CPU cores and hardware accelerators is available.
- ANSYS provides flexible and scalable HPC licensing.