I uploaded my computer performance here. It has 40 logical cores and 60 RAM. What is the best (fastest) setting to be used in Fluent as solver processes and GPGPU per machine?
I am using Academic version of Ansys2019R2
Thanks for your time!
I believe there is a limit on the number of cores in the academic license
Assuming you have enough licences the answer is "it depends".
40 cores and 60GB RAM isn't a good combination for large models, as you'll run out of RAM at 20-40M cells depending on the models used. Our clusters are typically spec'd at 3-4GB per core for most racks with a few (very) large memory machines for other tasks. The exact spec varies with each new cluster based on what's available and what we think we'll need.
Solver efficiency will start to tail off as the number of cells per core drops below 50-100k but again this is slightly model and domain dependent. We've got good scaling to 30-40,000 cores for some models and KAUST ran out of hardware at around 200k cores https://www.ansys.com/about-ansys/news-center/07-18-17-ansys-saudi-aramco-kaust-shatter-supercomputing-record
Distribution of CPU and GPU is discussed in the manual. If a Fluent run last more than 80% on LE Solver GPU can be advantageous.
Ansys customers with active commercial software licenses can access the
customer portal and submit support questions. You will need your active account number to register.