Release 2025 R1 includes the following improvements to the solution process:
Enhancements include:
The default support for message passing interface (MPI) software has been upgraded to Intel MPI 2021.13.0 for both Linux and Windows. Previously, the default was Intel MPI 2021.11.0.
When using AMD processors, the default support for message passing interface (MPI) software is now Open MPI 4.0.5.
For details and other optional supported MPI software versions, see MPI Software in the Parallel Processing Guide.
Updates include:
The ROCm libraries have been updated to 6.2.0. Prior releases used version 5.7.0.
The CUDA libraries have been updated to 12.6. Prior releases used version 12.0.
The performance of the PCG solver has been improved when running with GPU acceleration as more equation solver calculations are offloaded on the GPU card(s).
For more information on GPU cards and their updates, see Requirements for the GPU Accelerator in Mechanical APDL in the Ansys, Inc. Installation Guides and Requirements for the GPU Accelerator in Mechanical APDL in the Ansys, Inc. Installation Guides.
For Linux and Windows, the following cards have been tested and added to the list of recommended GPU devices for the GPU accelerator:
AMD Instinct MI325X
AMD Instinct MI250X
NVIDIA B100
NVIDIA A800
AMD Radeon Pro W7900
AMD Radeon Pro W7800
NVIDIA RTX 5880 Ada
NVIDIA RTX A4500 Ada
The following were removed:
NVIDIA Quadro GV100
NVIDIA Tesla Series V100
For the complete list of recommended GPU accelerators, see Requirements for the GPU Accelerator in Mechanical APDL (Linux) and Requirements for the GPU Accelerator in Mechanical APDL in the Ansys Installation Guides.
A new solver called the Mixed equation solver is now available. It combines the best of both the SPARSE and PCG capabilities, improving speed and reducing memory usage. For more information, see EQSLV and Solution in the Basic Analysis Guide.
For the PCG solver, the level of difficulty (
Lev_Diff) has been changed to optimize their values. For more information, see Change in Level of Difficulty Values for the PCGOPT Command and PCGOPT.The PCG solver requires less memory than in previous releases to solve models when:
Running on higher core counts (more than 64 cores).
Constraint equations and/or coupling equations exist in the model.
The PCG solver performance has been improved for certain models containing many contact pairs that were split using the contact splitting feature (CNCHECK,SPLIT).