[Nek5000-users] Getting Error when compiling a large problem

nek5000-users at lists.mcs.anl.gov nek5000-users at lists.mcs.anl.gov
Fri Jul 13 17:29:00 CDT 2018


Dear Ambrish,


Postnek will certainly fail -- it is a serial code.  You will need to use

VisIt (or something comparable) for parallel post-processing.


Also, as a matter of course, I strongly recommend starting out with

10^4--10^5 elements and running your simulation at lower Reynolds

number to go through all steps of the workflow.  Then step up to 10^6.

Only after that would I consider > 10^7.


As I say, you'll be in record territory at that point and thus completely

on your own w.r.t. to support.   Only a few of our most experienced

users have pushed to that high of resolution.   If I were tackling the problem

you are attempting I would most definitely take these smaller steps first.

They take no time and provide a chance to fully debug the whole process,

including forcing conditions, boundary conditions, etc.   Since they will

run a 100 times faster, it make sense to gain insight in this way.


hth,


Paul


________________________________
From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
Sent: Friday, July 13, 2018 12:15:38 PM
To: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Getting Error when compiling a large problem


Dear Paul,


Thanks a lot for your prompt response.

I am trying to generate the grid with 642*642*42 elements. And most probably I have to go to lx1 = 10 or 12. So the problem size will be something like 6400*6400*420. I will have to use 32768 or 65536 cores on SuperMUC.

I generated this grid with earlier version of Nek (svn), and had to change MAXNEL in genbox and genmap directories.

In the new version, it appears that changing MAXNEL in maketools enforces this value everywhere.


I am getting error in the postnek. Prenek appears to have compiled.


The error is as follows:


/usr/app-soft/intel/v2017/bin/ifort -o  /usr/wrk/people6/ampa9999/Nek_v17/Nek5000/bin/postx postnek.o postnek2.o postnek3.o postnek5.o postnek6.o tsort.o postnek8.o postnek9.o plot.o getfld.o legend.o userf.o revert.o trap.o animate.o genxyz.o screen.o g3d.o subs.o xinterface.o locglob.o postnek5a.o blas.o xdriver.o scrdmp.o coef.o postnek7.o speclib.o mxm.o byte.o ssyev.o iolib.o -L/usr/lib/X11 -lX11 -lm
make[1]: Leaving directory `/usr/wrk/people6/ampa9999/Nek_v17/Nek5000/tools/postnek'

ERROR: postnek failed to compile!





Dear Ambrish,


What polynomial order are you using?

How many processors?


This seems like a very large number of elements (a record number).


What resolution, in terms of grid points, are you targeting.


Best,


Paul





Dear Nek Users,


I need to generate a rectangular grid with 17,310,888 elements. However, when I am changing MAXNEL in the maketools and recompiling, I am getting error. I am attaching the compilation output and my maketools file here. Could you please help me regarding that? Compilation works fine for MAXNEL=4,000,000.

Thanks in advance for your help.


With best regards,
Ambrish


What tools failed? I guess prenek and postnek??

On 13 Jul 2018, at 17:19, "nek5000-users at lists.mcs.anl.gov <mailto:nek5000-users at lists.mcs.anl.gov> " <nek5000-users at lists.mcs.anl.gov <mailto:nek5000-users at lists.mcs.anl.gov> > wrote:




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20180713/2a649028/attachment.html>


More information about the Nek5000-users mailing list