Linux optimization for FLUENT
 

Hello everybody,

I would like to ask you, is there any linux software optimized for FLUENT? How can I optimize linux for run FLUENT calculations? Where can I find something to do this?

My problem is that this pc, which I use to calculations has RedHat 6.2 and FLUENT 5.5.14. These two programs don't work as I expected at the begining.

Regards Jacek

Re: Linux optimization for FLUENT
 

Hi,

I'm running Fluent_6.0 on RedHat7.2. Fluent is running very well. I do not have any problems.

Maybe you need to migrate to Fluent6.

Regrds

Re: Linux optimization for FLUENT
 

It should be OK.

Re: Linux optimization for FLUENT
 

Thank you,

I will try. But maybe Debian is better than RedHat 7.2?

Jacek

Re: Linux optimization for FLUENT
 

Linux is ok. The question is: Is Fluent compiled with the intel-compiler or with gcc ?

Re: Linux optimization for FLUENT
 

Hi,

I have no idea about compiler. What does mean gcc? I am planning to buy new pc mashine based on Athlon processors. Is it good way of thinking? Maybe they will work better than Pentium III 733 MHz, hope.

Regards

Jacek

Re: Linux optimization for FLUENT
 

Here have been some discussions about Atlons ans P4. Look for them. I`d prefer P4 >2GHz with rdram. What about the new 64bit cpu from Intel (I forgot the name..) ? benchmarks ?

Re: Linux optimization for FLUENT
 

I'm using Redhat 7.1 with Fluent 6.0. I've found that Fluent 5.5 doesn't like the glibcc version in 7.1, so this doesn't work properly with the gui. Fluent 6.0 has no such problems.

regarding compiling - I just use the standard make file and that works fine. I imagine it uses gcc as the compiler but haven't checked.

I have noticed that Fluent tends to run a bit slower if you use compiled udfs rather than inbuilt Fluent routines. For example when I use a udf to calculate an ideal gas density (same as Fluent) it seems to run a bit slower than the inbuilt model. I haven't done a rigourous comparison, but it seems a bit slower - which suggests the use of shared libraries etc may not be optimised during the compilation. The differences however are small and I doubt its worth worrying about too much.

Choosing the right models and setting up your problems etc is probably a better use of available time.

Greg

Re: Linux optimization for FLUENT
 

Yes, the udf's use gcc. But is it possible, that the fluent-code is compiled with other compilers ? Intels own c-compiler seems to produce faster code (10-30% I heard somewhere.. ?)

There had been a change of the glibc in suse 7.1 to 7.2 .

I also found, that the memory-requirements and the speed also depend on the nonconformal interfaces (I often have them).

The best way to get faster is: bay a new cpu :)

Re: Linux optimization for FLUENT
 

I'm not sure of any of these details. I suppose you can ask Fluent, though I'm not sure whether there's much point.

Speedup of 10-20% really isn't that much is it. I doubt whether its worth worrying about, given all the other costs and factors etc with modelling.

Greg