[FelixD]

Hi Kamath,

I have recently been able to access a linux machine to run my simulations with a lot more memory so I was hoping that I would not have the malloc() error when running DiveMESH for the CFD module.

My machine has the following parameters:
MemTotal: 65866644 kB
So I was not expecting the error, and am confused as to why this is still happening given it is only a 2D simulation.

Do you have any ideas what may be causing the issue? I have attached the files I am trying to run. The FNPF simulation runs very quickly now and I have no issues with running it.

[FelixD]

If it helps at all, the only directory that wasn't empty after I got the error is the DIVEMesh_Decomp directory, which contained teh text file DIVEMesh-Decomposition.txt

I wasn't able to copy the text file out of my linux machine however I took a screenshot of the contents hoping that it might reveal something problematic that I am not seeing.

I've attached the screenshot and hopefully it can be of some help!

Cheers,
Felix

[kamath]

There has been another round of RAM usage optimisation in the DiveMESH RC about a couple of days back.
Could you compile that version and see if the problem still persists?

[Felix_Sp]

Hey Felix,

I just had a look at your files and think that you have to change P40 from 1 to 2 in the FNPF simulation.

With P40 1 you only have a single file printout for the latest time/iteration step (P41 P42). However, the CFD-module needs the continuous state-file prints as input data.

The temporal resolution is then based on your print-out frequency (again P41 P42). Time steps in between are interpolated in some way, I guess.

Maybe Arun can confirm this (if this solves the problem, you can ignore the private messages).

Cheers,
Felix

[kamath]

@Felix S. When I last tested P 40 1 and 2 give the same result so technically not a problem though you are correct in that P 40 1 would be the "correct" input.

However, if that affects the memory allocation, I am not sure.
I find it hard to replicate the memory overload