|
[Sponsors] |
March 10, 2004, 06:17 |
solver error #001100279. Memory Problem
|
#1 |
Guest
Posts: n/a
|
Hi guys,
I got a solver error at the beginning of my simulation and it quit as a result. I have about 4millions hex elements. How much memory do you think I should at least have for my problem? Is there a limit for the number of elements that can be solved in CFX? +------------------------------------------+ 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! |
|
March 10, 2004, 06:24 |
Re: solver error #001100279. Memory Problem
|
#2 |
Guest
Posts: n/a
|
Try to see section of your out file where is located memory requirements - at the begining of solver section. Memory requirements determined by available computer memory and for large cases you can use parallel mode, in which memory requirements is mainly depends on partitioning requirements and number of available licenses.
|
|
March 10, 2004, 21:43 |
Re: solver error #001100279. Memory Problem
|
#3 |
Guest
Posts: n/a
|
For a single domain hex grid CFX-5 uses about 1400-1500 Bytes per node (maybe a bit less, but I'm going by memory, it's not far off anyways).
For a 4e6 grid that's ~1400*4*10^6 / (1024*1024) MB = 5340 MB. So you need access to around 5-6 gigs of RAM. If you are running this in serial you will need a big 64 bit workstation (eg: SGI/HPUX/Compaq), if you are running on 32 bit Linux/Windows then you must run in parallel. Neale |
|
March 10, 2004, 23:53 |
Re: solver error #001100279. Memory Problem
|
#4 |
Guest
Posts: n/a
|
Thanks a lot, Neale.... This is very helpful.
|
|
March 11, 2004, 00:44 |
Another question, Neale
|
#5 |
Guest
Posts: n/a
|
If I have about 4 millions unstructured tet mesh, can I use your figure to estimate the amount of memory required. It doesn't need to be exact though. I have to specify the memory requirement before I run the script. For large problem, I will have to put a request to the administrator. Your information on this issue will certainly be very useful. Thank you once again.
|
|
March 17, 2004, 01:17 |
Re: Another question, Neale
|
#6 |
Guest
Posts: n/a
|
A Tet/Prism hybrid mesh is a bit more per node than a hex mesh due to increased connectivity between nodes on an unstructured grid. So, something more like 2000-2200 Bytes per node would be a ballpark estimate. You can probably do the math for your problem size.
Neale |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Polyflow Simulation with ViscoElastic Model memory problem | sid2909 | FLUENT | 0 | May 10, 2010 06:29 |
Polyflow Simulation with ViscoElastic Model memory problem | sid2909 | ANSYS | 0 | May 10, 2010 06:27 |
Problem with implicit unsteady solver | CCMuser | STAR-CCM+ | 2 | March 3, 2010 12:20 |
gridgen memory allocation problem | ziad | Pointwise & Gridgen | 5 | February 27, 2010 21:16 |
memory problem in solver | rao | CFX | 4 | October 15, 2003 15:58 |