CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   memory problem (https://www.cfd-online.com/Forums/cfx/21103-memory-problem.html)

wilson March 23, 2005 13:28

memory problem
 
Hi CFX users I am doing a project on Hydrofoil of a submarine the dimensions of the Hydrofoil is up to the scale 1:1 . Now I have to find out the cavitation occurring due to flow of water. The project is similar to the tutorial on Hydrofoil. I am giving boundary condition values just as given in the tutorial. But I am facing a very unusual problem.

ERROR #001100279 has occurred in subroutine ErrAction. | | Message: | | *** Run-time memory configuration error *** | | Not enough free memory is currently available on the system. | | Could not allocate requested memory - exiting!

This is the message that I receive in the out file generated. I made sure that ample memory is available in the hard disk but still the problem remains the same. In my hard disk free space left is 2 GB. RAM size is 1 GB. Is more space required than this or the problem lies somewhere else. Can any of u guys help me out?

Regards wilson


caty March 23, 2005 14:35

Re: memory problem
 
I don't think that the message reffers at the space avaliable on the disc, but to the RAM. 1GB may not be enough is you have a very fine mesh. This is what happend to me as well.

There is here somewhere on the forum a post that says how to calculate the Ram required by a certain number of nodes.

Akin March 23, 2005 16:24

Re: memory problem
 
I think the ratio from my experience is about 0.55 using tet and wedge elements, that is you will need 550mb to solve 1 milliion volume mesh. However, I also found out that it depends on the solution strategy as well. Finally, I like running on Linux machines, I get about 1.8Gb of ram out of 2 on each computer in the cluster, when I use WinXP its about 1.5-1.6Gb and sometimes less. Try running in Parallel, of course this are my experiences, some one may feel differently.

Glenn Horrocks March 23, 2005 17:35

Re: memory problem
 
Hi,

CFX is a node centred solver, not an element centred solver so it is more meaningful to use the number of nodes rather than elements.

In my experience I can run about 1 million nodes in two seperate 1GB partitions on a dual processor Xeon/WinXP machine for a single phase compressible simulation. Obviously things like multiphase will reduce that as more variables are stored per node. I would guess with 1GB ram you could run about 400k nodes.

As Akin says, I suspect you will need to go parallel to get a realistic simulation size.

Glenn Horrocks

wilson March 24, 2005 06:47

Re: memory problem
 
thanks all of u guys. its working!! i appreciate your help.

regards wilson

wilsondeo March 30, 2005 07:36

Re: memory problem
 
guys i am now facing new problem in the same cavitation problem. after 10 iterations i get a message of fatal error and the solver stops. i managed to get the number of nodes within limit. can't figure out why this fatal error comes. anyone have faced this problem before.

regards wilson

Glenn Horrocks March 30, 2005 18:34

Re: memory problem
 
Hi,

If the simulation solves for a while then crashes then you have diverged the solver. Have a look in the documentation regarding tips for obtaining convergence.

Glenn Horrocks


All times are GMT -4. The time now is 08:10.