Learning Forum

[h.a.maldonadodeleon]

Dear all,  I would like to know if someone has successfully enabled the GPU native solver for Fluent 2023R1 within Workbench. I would like to automate some runs via the DOE module and use the GPU acceleration options from Fluent.  Thanks in advance for your support. 

Update: As of January 2025, I've 'discovered' some workarounds: 

• If you set an environment variable in the Standalone Fluent Launcher to pick the GPU devices: FLUENT_GPU_DEVICES=all
• An alternative is to use Discovery and set your case (from geometry to solution) there. You can take advantage of the LiveGX solver which uses GPU-acceleration for solving single phase simulations. 

Hector 

[Rob]

The GPU solver requires an Enterprise level licence, so I'm not sure it will unless you can switch the Fluent licence call in WB. Given the WB Fluent launcher doesn't have this option I'm not hopeful. You may find Optislang/Twin Builder to be a better choice for controlling the runs. 

[h.a.maldonadodeleon]

By any chance, is there a way to change the specs for the Fluent Launcher? I do have access to an academic license with the Enterprise Level. Also, thanks for pointing out the Twin Builder alternative. I will take a look at it. 

[Rob]

Standalone, yes, but I can't see an option in Workbench. Read the restrictions in the User Manual, I suspect even if you can change the licence the residual/monitor reporting will be an issue. 

[Wenwei Qiu]

Hi Hector, I was facing the same question as you did, but after I read the user manual of 2024R1, I finally gave up!

It says :

The following list contains the current limitations of the Fluent GPU Solver:
• The GPU solver is not available in the Ansys Workbench environment

•...

I'm using workbench to parameterize my model so I can get the calculations to run automatically, but now it seems that I have to keep on using CPU, or find another way to calculate automatically, until ANSYS make it possible to run GPU solver in workbench????