[ataha]

I received the following problem with one of the Fluent users I am supporting on our Supercomputer. We have a shared memory big machine which has (1GB) per core. Any inputs?
We have constructed a very large size mesh (about 45M, in future
application, even bigger one will be needed) and have successfully been
imported into Fluent running our local Win64Xp workstation (32G RAM, 8
cores). Preliminary simulation has been running and worked OK, but one
iteration took about eight hours. Then we resorted to NCSA parallel
computers.
But the problem is when we submitted the task, the following error appear:

_999999: Unable to allocate storage._

We have tried many methods, including :
(1) increase the node/core number;
(2) increase the parallel process of Fluent;
(3) change from Fluent6.3 to Fluent12;
(4) using -himem option;

Until now, the maximal node/process used is:

#PBS -l nodes=60pn=8:himem
Fluent121 3ddp -g -t48

But the grid can't be imported due to the error of "Unable to allocate
storage ".

[samy902]

Hi ataha,

I am going through a similar case. Did you find any solution?