[vulcan]

Hi,

I am trying to use a 12 core (6 Hyperthreaded) and when running
fluent 2ddp -t12

I am getting the following error/warning

Process affinity not being set. core ids failed check (maxid>cores, ht>1)

Does anybody have any idea what it means? The simulation starts but I am getting this warning.

Thanks

[vinerm]

Try with lesser number. And when you say 6 hyperthreaded, it might mean that there are 6 dual core processors.

If you have unix machine, check using cat /proc/cpuinfo in a terminal, what is the last number of processor. If it is 11, then your command is perfectly alright. However, if it shows some other number, say n, then your command can be fluent 2ddp -tn+1

[yilufeihua]

When I started fluent to read file,I got the following error/warning
Note:Rank=0:Process affinity not being set. Offsets failed sanity check.
Note:Rank=1:Process affinity not being set. Offsets failed sanity check.
Note:Rank=2:Process affinity not being set. Offsets failed sanity check.
.......
(My workstationo has 12 core, even I use 2 core ,I'll also receive this message)
How can you solve the problem?

Originally Posted by vinerm

Try with lesser number. And when you say 6 hyperthreaded, it might mean that there are 6 dual core processors.

If you have unix machine, check using cat /proc/cpuinfo in a terminal, what is the last number of processor. If it is 11, then your command is perfectly alright. However, if it shows some other number, say n, then your command can be fluent 2ddp -tn+1

[sunilpatil]

Hi,
Have you solved this problem

[vinerm]

This is not an error or even a warning. This is just an information to tell user that the processor is already in use by other programs. If you make sure that no other program is using these cores then you'll observe that Fluent gives no such warning or information.

[sunilpatil]

Thank you for your reply. Recently I have posted a question (http://www.cfd-online.com/Forums/flu...rocessing.html) do you know how to solve this parallel processing problem.

[idfccc]

If your CPU is AMD's product, you can have a try to set the MPI types to "msmpi". I used this mothed, then the time cost of my simulation is decreased obviously.

[waynetsu]

Originally Posted by idfccc

If your CPU is AMD's product, you can have a try to set the MPI types to "msmpi". I used this mothed, then the time cost of my simulation is decreased obviously.

My CPU is Intel. I also feel my parallel simulation is much more time-consuming than before. Does this "msmpi" connected to the CPU type?

[@Arash]

Originally Posted by idfccc

If your CPU is AMD's product, you can have a try to set the MPI types to "msmpi". I used this mothed, then the time cost of my simulation is decreased obviously.

Hi Dear friend,

I use an Intel processor that has 24 physical cores. When I use the number of cores for example 14 or 16 or... in order to parallelize and reduce the solution time, my solution goes slowly. What should I do ?

Thankyou

[@Arash]

Originally Posted by waynetsu

My CPU is Intel. I also feel my parallel simulation is much more time-consuming than before. Does this "msmpi" connected to the CPU type?

Hi Dear friend,

I have exactly the same problem as you and my processor is Intel.
Did you manage to fix this problem and use more cores for parallelization?