Hi, dear developer,
I want to know which type of CPU shows better performance because I am assembling a workstation.
Is FDS optimized for Intel or AMD? There is rare information in this field.
I tested Intel Xeon 2680V4 (Dual) and AMD 3990X. But I found Intel is relatively faster when running multiple cases.
For example, I would run 28 cases for 28 cores Intel and 64 cases for 64 cores AMD at the same time for a total of 500 cases using Python scripts. This is because I want to use a single core to run one case instead of using MPI. This is faster and I want to do sensitivity analysis.
Intel costs about 5000 seconds while AMD costs 3000 seconds for 500 cases, respectively.
This means Intel would cost 10 seconds while AMD would cost 6 seconds for 1 case on average, respectively.
Even though AMD has a higher main frequency (2.9 GHz versus 2.8GHz) but it doesn't show better performance than expected.
Considering AMD has 2.3 times more cores than Intel, AMD is relatively slower. Since second-hand Xeon is extremely cheaper now, maybe using a new AMD is not a good choice.
Is FDS optimized for Intel?
By the way, do you have any suggestions about choosing a CPU?
Yours, Wang.