[lenzo]

So, I have attached the features of my computer.

If i want to take the maximum advantage of my features, in ANSYS, for example, how is the best way to use it?

Runing the Processing options in serial or Parallel? How many Processers and how many GPGPUs per machine?
In double precision?

I really want to use all the computer for this purpose, so I don't mind using all its potential just for CFD.

Thank you in advance!

[Antanas]

Originally Posted by lenzo

So, I have attached the features of my computer.

If i want to take the maximum advantage of my features, in ANSYS, for example, how is the best way to use it?

Runing the Processing options in serial or Parallel? How many Processers and how many GPGPUs per machine?
In double precision?

I really want to use all the computer for this purpose, so I don't mind using all its potential just for CFD.

Thank you in advance!

How many? No more that your license allows you. You do have license, don't you?

[lenzo]

Yes,
but, if my computer is an i7, i don't see the point of using 64 processors. Because I don't have that many

[flotus1]

So much for the obvious part
You should get the best performance running simulations on 4 threads since you have a cpu with 4 cores. You can try to increase the amount of threads beyond the amount of cores with hyperthreading activated, but you won't see much of a benefit or even decreasing performance.
GPU computing: your spec sheet does not give away any information about GPUs. I doubt that you have a suitable GPU so lets leave this aside. Ansys is quite restrictive on the GPUs they allow for GPU computing thanks to their partnership with Nvidia.
DP vs SP: single precision is slightly faster and uses less memory. But if your case fits into the RAM with double precision I would use it anyway. One thing less to worry about when validating your results.
I assume you have 2 DIMMs installed and use fast memory, e.g. DDR4-3000 or faster. It helps a lot.