CFX processor and memory tweaks
I know you can select local parallel at the solver menu and split the solver work load to the amount of physical cores you have (n.b. I don't mean hyperthreading. I.e. for an i7 CPU I know you should run a local parallel job between 4 processes and not 8!)
However are there any memory tweaks I should also consider? Up until now I've just left the memory settings to default. For example I've kept the memory allocation factor to 1.0.
To give you an insight my computer has:
- Intel Xeon E3-1240 V2 @ 3.40 Ghz CPU (4 cores + hyperthreading = 8 threads)
- 16 GB ECC memory
Would I experience significant gains if I were to tinker with the memory allocation factor? Are there any other tweaks I should consider to get the most out of my computer?
It is very unusual for tweaks like this to make any difference. Give it a go because it is easy to test, but I will be very surprised if it makes a difference worth the effort.
Quite simply - if there were tweaks available to speed up simulations, wouldn't you think they would be implemented in the software? One of the key selling points of CFD software is speed so ANSYS has done a lot of work to get all possible speed out of the solver. It is unlikely you will be able to better.
I guess I can do a simple pipe flow simulation with and without memory allocation tweaks and compare the wall clock simulation times.
Watch this space.
|All times are GMT -4. The time now is 19:38.|