Once you have the prerequisites above in place, MSC Apex Generative Design can use your GPU to run super fast but you need to know a few tricks to ensure it stays as fast as possible.
1) Use your Task Manager to identify how much Dedicated Memory your screen card has (not the shared memory).
2) In Apex GD, set the solver settings to use the GPU and limit the amount of memory to this value (you can increase this value a bit, but see further down how to tell).
NB: If you do not have the GPU option, you missed the option to install the GPU solver during installation. To fix this you will have to re-install Apex GD and ensure you tick the option.
4) Apex GD will choose an internal mesh resolution to ensure that this set limit is not exceeded, but this seldom mean that it will ever reach this limit. Therefor you can increase the limit a bit if you want a higher result resolution in the final iteration.
If you pick a too high value the solver will automatically switch to your CPU when reached but then the analysis will run much slower. It is best to avoid this.
5) Run your analysis and find the active result folder location which will look like:
"...Apex Generative Design WorkspaceModelNameMeshless Generative Design Scenario 1Opti_Date_Time"
Find the "log_FEM.log" file, open it with a text editor and go to end of the file:
The maximum of the values indicated above (near the bottom of the file) is the maximum amount of dedicated GPU memory that was used for your specific model
If you wish the simulation to use more of your available dedicated memory, you can increase the value in the application settings.
Please note: Each model is unique and the actual amount of used memory will be dependent on the volume of material left in the design which is a function of aspects like strut density, number of attachment points etc..
