-
-
April 6, 2024 at 4:48 amAmmar AhmedSubscriber
Hello everyone, I am trying to speed up my simple CFD (flow over body), I have dell precision tower with dual xeon e5 2690 v4, with 28 cores and 56 threads, and RTX 3050 24GB. The issue i am facing is that while performing CFD, usig SIMPLE method the simulation runs perfectly but GPU utilization is zero and when I shift to COUPLED method it utilizes GPU but consupmtion time is very large (about 50% more as compared to before). Kindly guide as to how I can shift to full GPU utilization with minimal time. For reference I am attachng a time comparison data.
-
April 8, 2024 at 11:34 amRobForum Moderator
The gpu offload is only compatible with a few models so you may find it's not giving much of a speed up. Depending on cell count and models more cpu may give a good speed up, but that also assumes the hardware can handle the message passing between the cores.Â
-
- The topic ‘GPU acceleration in CFD analysis’ is closed to new replies.
- Non-Intersected faces found for matching interface periodic-walls
- Unburnt Hydrocarbons contour in ANSYS FORTE for sector mesh
- Fluent fails with Intel MPI protocol on 2 nodes
- Help: About the expression of turbulent viscosity in Realizable k-e model
- Cyclone (Stairmand) simulation using RSM
- Mass Conservation Issue in Methane Pyrolysis Shock Tube Simulation
- Script Error
- Facing trouble regarding setting up boundary conditions for SOEC Modeling
- UDF, Fluent: Access count of iterations for “Steady Statistics”
- RIBBON WINDOW DISAPPEARED
-
1421
-
599
-
591
-
565
-
366
© 2025 Copyright ANSYS, Inc. All rights reserved.