Issue in running the Standalone nek.i tutorial input file #600
Replies: 7 comments 8 replies
-
|
Hmm, I am not sure. This tutorial is part of our test suite, so we know it runs fine and is up-to-date. Can you try re-running without threads? Nek only uses MPI. That first line is deleting the JIT (just-in-time) compiled files that Nek creates when running an input file. Normally you don't need to do anything special with them, but it's good practice if input files are being troublesome to delete those so you're 100% sure there's not any lingering build settings/etc. from previous runs. |
Beta Was this translation helpful? Give feedback.
-
|
Strange, I get the same error. My installation passed all of the checks, run OpenMC and Neks separately, as well as the Cardinal test suites. I'll keep looking into the issue. |
Beta Was this translation helpful? Give feedback.
-
|
The test suite could be skipping this particular test, because we currently require at least 8 ranks to run it. I'd suggest proceeding to the other tutorials/examples, and I'll focus on fixing whatever is wrong with this one. If the whole test suite passes, then your install is fine and it's probably some issue on our end (like the case is using too much memory, or something we should fix). |
Beta Was this translation helpful? Give feedback.
-
|
Okay, I went to the standalone channel simulation in the test directory (ie test/tests/nek_standalone/channel/nek.i ) and was able to run the simulation successfully. I will work with these input files for now. Thank you very much for your help! |
Beta Was this translation helpful? Give feedback.
-
|
Hi All !
test:nek_errors/invalid_settings.executioner ................................. FAILED (EXPECTED ERROR MISSING)
|
Beta Was this translation helpful? Give feedback.
-
|
Hi @ME-20244, can you try running steps 1 and 3 from the "Checking the Install" section here: https://cardinal.cels.anl.gov/without_conda.html and then please report whether those cases ran here? Can you also try re-running the test suite, but save the screen output like |
Beta Was this translation helpful? Give feedback.
-
|
Hi! Yes, the three first tests run without issues (see below). For the whole tests, I attached the output file.TEST-1 OK k-effective (Collision) = 0.94693 +/- 0.02293 openmc0: Extracting OpenMC tallies... done TEST-2 OK TEST-3
finished with exit code 0 |
Beta Was this translation helpful? Give feedback.
-
|
Hi @ME-20244 - it looks like you're running into issues on both the OpenMC and NekRS sides. OpenMC errorsYour OpenMC errors look like small exo/csv diffs (i.e. the code is running fine, but there's small differences in the results). What OS are you running on? The test suite should run fine on a Linux-based OS, but it's known that OpenMC's test suite runs but does not give 100% matches for the simulation outputs on MacOS (openmc-dev/openmc#801). In practice, this isn't something to be that concerned about because these differences in output get smaller as you increase the number of particles (the differences should be within statistical error). Nek errorsThese look like a NekRS-specific error, not something in Cardinal itself. Can you try running NekRS by itself, without Cardinal? To do so, you'll run Can you post what the output of that is here? For instance, to run the |
Beta Was this translation helpful? Give feedback.
-
|
Hi! |
Beta Was this translation helpful? Give feedback.
-
|
Oops, I accidentally told you to run the wrong case (I needed to tell you to run a NekRS standalone case). Can you try the following instead? |
Beta Was this translation helpful? Give feedback.
-
|
Here it is ! |
Beta Was this translation helpful? Give feedback.
-
|
Ok that looks normal - can you also please try the following? This is a test which is failing in your run of the test suite, so I want to see if it runs okay outside the test harness. |
Beta Was this translation helpful? Give feedback.
-
|
This is the output required. |
Beta Was this translation helpful? Give feedback.
-
|
That looks good! think there may be something wrong with just the test harness itself, but not the Cardinal install. You can run Cardinal simulations fine with the Cardinal executable, but hitting a WIN_Allocate error when running those same input files through the test harness. If you are looking to do simulations and analysis, you should be ready to go with your Cardinal install. If you were going to be changing Cardinal's source code and needing to run the test suite frequently for test-based design, then we should still try to figure out the |
Beta Was this translation helpful? Give feedback.
-
|
As I am new in using the code, your conclusion reassures me to begin the simulation and analysis phase. |
Beta Was this translation helpful? Give feedback.
-
Hi all,
I am now beginning my journey of cardinal simulation with my first task, simulating a simple turbulent N-S problem. To accomplish this I have created my own directory in cardinal called "Projects" and within that directory "NS". I copied all of the files from the standalone tutorial into the NS directory as such:
cp -R /home/anthonyb08/cardinal/tutorials/standalone/* /home/anthonyb08/cardinal/Projects/NS
I then cd into the NS directory to run the simulation using the following command:
mpiexec -np 2 ../../cardinal-opt -i nek.i --n-threads=2
The simulation begins to run up until this point :
verify mesh topology
-0.50000000000000000 0.50000000000000000 Xrange
-0.50000000000000000 0.50000000000000000 Yrange
0.0000000000000000 20.000000000000004 Zrange
done :: verify mesh topology
mesh metrics:
GLL grid spacing min/max : 6.31E-04 6.98E-02
scaled Jacobian min/max/avg: 3.68E-01 9.35E-01 7.98E-01
aspect ratio min/max/avg: 2.69E+00 3.39E+01 1.14E+01
call usrdat3
done :: usrdat3
gridpoints unique/tot: 2263717 3317760
dofs vel/pr: 2187780 2258340
nek setup done in 2.5278E+00 s
set initial conditions
nekuic (1) for ifld 1
call nekuic for vel
xyz min -0.50000 -0.50000 0.0000
uvwpt min 0.45434E-10 -0.99997E-01 -0.21000 0.0000 0.0000
used config: pw+host (MPI min/max/avg: 4.50e-06s 1.22e-05s 8.82e-06s / avg bi-bw: 14.6GB/s/rank)
min 0% of the local elements are internal
meshParallelGatherScatterSetup N=7
..timing gs: 8.42e-03s 9.65e-03s 9.98e-03s 9.86e-03s
used config: pw+host (MPI min/max/avg: 9.60e-06s 1.94e-05s 1.45e-05s / avg bi-bw: 26.7GB/s/rank)
copying solution from nek
calling udf_setup ... .--------------------------------------------------------------------------
Primary job terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.
mpiexec noticed that process rank 1 with PID 0 on node DESKTOP-LTJ793R exited on signal 9 (Killed).
What does the community suggest?
-Thanks AB
Beta Was this translation helpful? Give feedback.
All reactions