CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Parallel Simulation with fluent (https://www.cfd-online.com/Forums/fluent/200893-parallel-simulation-fluent.html)

esi1520 April 16, 2018 08:44

Parallel Simulation with fluent
 
Hi dear friends,

I have a PC with 16 core (2.3 GHz) and 32G RAM. I am trying to simulate a furnace with large dimension. Speed of simulation runs is very low. Can anybody to help me for parallel simulation to accelerate speed of simulation like MPI or GPU or other way.

Thanks

Mohamed Mousa April 16, 2018 09:11

1 Attachment(s)
Did you choose the parallel option ?

esi1520 April 16, 2018 09:44

3 Attachment(s)
Quote:

Originally Posted by Mohamed Mousa (Post 689059)
Did you choose the parallel option ?

Dear Mohamed thanks for reply,

Yes i do. Also I set GPGPUs value to 2 or 4 and in parallel setting tab select intel in MPI type.
in past i install MPI setting from ansys setup.exe however when i open fluent message
"Inhomogeneous process distribution on multiple machines.
Or processes per machine not an exact multiple of GPGPU's per machine.
Or not enough GPGPU's per machine.
GPGPU computing disabled." appeared.
Is there any way to accelerate fluent simulation speed?

Thanks

Mohamed Mousa April 16, 2018 12:10

Dear Esmaeil,

Check the link bellow. I hope it helps you solving the error massage.
https://studentcommunity.ansys.com/t...gpu-in-fluent/

I would also mention that:

1- Installing MPI from ANSYS setup.exe is for distributed memory only (More than 1 machine connected together by a network). So, if you are performing your simulations on a single machine, you don't need to install the MPI individually.

2- I see in your uploaded picture that you specified 4 GPUs. That means you have 4 GPU cards. That doesn't mean you want to decompose you domain into 4 parts.

3- I read before, that FLUENT supports only Quadro and Tesla Cards. So please double check that your card is supported if you are using something else.

Regards,

LuckyTran April 19, 2018 23:55

Before you do anything else.... try just using parallel Fluent with 16 processes and none GPU's. This will probably be the fastest calculation.

I seriously doubt that you have 4 GPU's that are all Quadro's or Tesla's. Apparently, Fluent agrees with me. GPU's only make the calculation faster if you're using the COUPLED P-V algorithm and the entire case fits on the GPU's. If your case is really big, then I doubt it fits on your GPU. What is the output of solve -> parallel -> gpgpu? Does Fluent recognize 4 GPU's?


All times are GMT -4. The time now is 01:00.