If i have a mesh of half a million hexa cells. Is there a way to calculate how much RAM i need to run this problem in serial. My solver keeps crashing complaining about memory allocation. I do have enough memory in my hard disk so it must be RAM it is complaining about.

Thank you

What for a CFD-software do you use!? ... 0.5 Million sounds not much for a modern 2GB RAM Hardware!

Regards, Stephan

I am using CFX. I have only 512MB in my computer. Do you think it is crashing due to lack of RAM memory?

Regards alpha

Definitively you have lack of RAM. How much RAM is needed for specific problem you can read at the start of solver run in text window under "Memory Allocated for Run".

But still if you have more RAM than needed on single processor computer you would be able to solve only problems which need 1-1,5 GB of memory . The problem is 32bit OS. In this case parallel run must be used.

Yes. Memory is cheap, get more.

Definitely! Except you are a pure student ... !

But generally also each CFD-Code have different requirements to the amount of RAM needed for a specific grid size! Unfortunately CFX is one of the more hardware consuming codes than others due to it's unstructured approach - if I remember right!? ...

As a last chance you can try to downscale the amount of RAM alredy covered by WINDOWS ( due all this interactive, colorfull but unnecessary garbadge / use Google for more )! Also try to reduce all system background activities to a minimum ( Virus scan ... ).

Good Luck!

Biggest issue with CFX is its coupled solver, which is a big memory hog. Fluent's segregated solver is much less memory intensive, but the only code I know of that can do big problems with little RAM is CFD-Fastran, which is multi-block structured. I guess other multi-block structured codes would be similar.

Thank you so much for your answers guys. I have solved the memory issue by moving my problem to a 2 Gig RAM computer, but now its crashing due to a "floating point exception: Overflow" error. Any advice with regards to new problem! Thank you again

Regards alpha

I would guess bad programing is the cause of your problem. I suggest that you redefine your problem - from the beginning.

Over the years I have found a number of memory related crashes that could be solved by redefining the problem.