Possible reasons for not working parallel computing

My question is this: Are there any possible reasons that parallel computing is not working.
I followed the guide of butterfly and I figured out that parallel computing component is not working well.
I wonder why is it like this but still cannot find the reason. If I remove the component and run it with non parallel, it works. can anyone let me know why? Any possible reasons??

Hi,@Junoh Please post the further information about your question. Is it a question about daylight analysis or energy analysis or CFD analysis? And you should upload some images or screenshots about your question.

As shown here, when I link decomposing component to snappyhexmesh component, snappyhexmesh.log is not generated.
It is about CFD analysis.

Good afternoon.

I reopen this topic to ask for a easy tutorial for dummies to how create the parallel computing for butterfly. Using the outdoor example and increasing the grid, the calculation spends 2min per iteration.

Thanks.

1 Like

Hello,

I have the same problem in butterfly. when I run without parallel, it works well, but when I run with decomposeParDic scotch, it shows these errors, could anyone help ?

Thanks a lot !

Resolve process report output failure
Take a look at this. This should help you. I’ve talked about this in one of my online classes .

Tao

For blueCFD-Core 2017-2, the MS-MPI version must be 7.1 or 8.1. Is your version of MS MPI compatible with blueCFD?

Hi Tao,

Thanks for your help. I’ve browsed the topic that you share, but the problem still exist. Actually I am beginner with butterfly, I don’t really understand why do we need to change name of file ‘mpirun’ to simplefoam as what LvXiao suggest, knowing that the simplefoam file exist already in my log folder and the file ‘mpirum’ is empty. Here is my screenshot.

I’ve also take a look at, https://vimeo.com/showcase/7101158/video/373217727, site shared by suelikescats, but I don’t know why do we need to define a surface file as he show, as we have already defined ‘probe’ in the outdoor air exemple file, I think that’s the same.

BTW, could you share me your online classes ? It’s ok even though it would be in Chinese or in English.

Thanks again !

Yiqi

HI,

Thanks for your answer,

I’ve tried both of 7.1 or 8.1, and it still doesn’t work.

Ladybug Tools1.4.0 Tutorial (Chinese)
I noticed that there were a lot of Chinese speaking engineers and scholars in the forum, maybe I should open my course to the public.

Thanks a lot !

I love your channel !

How about running the contents of ir.bat line by line at the command prompt? You may get a more detailed log.

The solver for isothermal calculations is simpleFoam. To run it in parallel, we use mpirun -np cpunum simpleFoam. Butterfly create a log with the name of the command you run. When you read the residuals, you need to change the log file name from mpirun.log to simpleFoam.log because it tries to read simpleFoam.log instead of mpirun.log, even though you did a parallel calculation.

Hi !

Thanks for your apply. In fact, I’ve tried to change the name of mpirun, but after that, I discover my mpirun file is empty, so it doesn’t work.

Have you checked the logs with command prompt?

Did decomposePar fail?
What about changing the method from scotch to simple?

Hi,

I’m sorry, but could you tell me how to check the logs with command prompt ?

I think my decomposePar fail also, here is the message

Same for simple decompose

Thanks a lot

Hi,

BTW, I did the samething on another computer, it works well, so is it possible that was a problem link to my pc environment or my GPU ?

I have 2 Core and 4 processors on this pc and 4 cores and 8 processors on that one works. For all of them, I’ve just decomposed the calculation on 2 by decompose_par_dic.

Thanks !

I’m glad if the problem has been solved. The following information is for reference only.

The contents of ir.bat would be something like this. Do this line by line in a command prompt.

call "C:\Program Files\blueCFD-Core-2017\setvars.bat"
set PATH=%HOME%\msys64\usr\bin;%PATH%
cd C:\Users\yourname\butterfly\projectname
decomposePar  | tee log\decomposePar.log
mpirun -np 4 simpleFoam  -parallel | tee log\mpirun.log
reconstructPar  | tee log\reconstructPar.log
rm -r proc* | tee log\rm.log

One processor is usually allocated to an OS, etc. Therefore, the number of threads is less than the maximum number of processors.
However, I think all processors can be used for parallel computation. I’m not sure why a 4-processor PC could not run 4 parallel calculations. But we don’t usually do such a calculation, there is no problem.

Hi,

I try to do it in a command prompt, it works until decomposePar | tee log\decomposePar. log
then when I tap mpirun- np 4 simpleFoam -parallel | tee log\mpirun. log
there is the errors :

You seem to be typing the same command twice, as shown below.
mpirun -np 4 simpleFoam -parallel | tee log\mpirun.logmpirun -np 4 simpleFoam -parallel | tee log\mpirun.log

To be correct, try the following.
mpirun -np 4 simpleFoam -parallel | tee log\mpirun.log

Thanks a lot, I didn’t realize that, really sorry about this.

I try again but the error exist.