Number of Depletion Regions effecting Max NPG / MPI ratio

Report all good and bad behavior here
User avatar
Jaakko Leppänen
Site Admin
Posts: 2387
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by Jaakko Leppänen » Thu Nov 12, 2015 5:02 pm

I'm getting the same error now. I'll get back to you...
- Jaakko

User avatar
Jaakko Leppänen
Site Admin
Posts: 2387
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by Jaakko Leppänen » Thu Nov 12, 2015 5:20 pm

Can you post the run-time output here?
- Jaakko

User avatar
Jaakko Leppänen
Site Admin
Posts: 2387
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by Jaakko Leppänen » Thu Nov 12, 2015 5:34 pm

Another thing to try is to adjust the batching interval, i.e. the number of generations for collecting the results (by default 1). This is done using the pop card:

Code: Select all

set pop <npg> <act> <skp> <k0> <btc>
where <k0> is the initial guess for k-eff (can be set to 1) and <btc> is the batching interval. Try setting the value to something like 10. This might remove the error, but you may get another one about not being able tto initiate a self-sustaining chain reaction. If you get this error, then the problem is most likely that there were no fissions within that generation.

Just to be sure... When you say you are running 1200 nodes and 16 cores, you are using MPI only between the nodes, and intra-nodal parallelization is done with OpenMP, right?
- Jaakko

cgentry7
Posts: 24
Joined: Mon Jun 03, 2013 10:59 pm
Security question 1: No
Security question 2: 92

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by cgentry7 » Thu Nov 12, 2015 5:52 pm

That's correct, internode parallelization using MPI and intranode parallelization using OPENMP. The output files are too large to attach to the post so I'm going to email them.

cgentry7
Posts: 24
Joined: Mon Jun 03, 2013 10:59 pm
Security question 1: No
Security question 2: 92

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by cgentry7 » Thu Nov 12, 2015 8:41 pm

Running with batching set to 10 produced the "Unable to initiate self-sustaining chain reaction" error.

I dropped the batching number to 5 and got the following warning:

***** Thu Nov 12 12:34:18 2015 (seed = 1447349464, MPI task = 877)
Warning message from function SampleReaction:

Failed to sample target nuclide (U074z31 35200748 -1)


I get this warning multiple times throughout the run with the MPI task and depletion zone changing with each call. It always appears to concern isotope "35200748" though.

I also noticed for the first couple of cycles that I got the following:

Serpent 2.1.24 -- Static criticality source simulation

Title: "TEST LATTICE"

Transport calculation: step = 1 / 73 (predictor)
BU = 0.00 MWd/kgU
time = 0.00 days

Active cycle 1 / 250 Source neutrons : 3058

Running time : 0:03:16
Estimated running time : -:--:--
Estimated running time left : -:--:--

Estimated relative CPU usage : 234.7%

k-eff (analog) = 0.00000 +/- 0.00000 [0.00000 0.00000]
k-eff (collision) = 0.00000 +/- 0.00000 [0.00000 0.00000]

Options : (O1) (UNR) (MPI=1200) (OMP=16) (CE/LI + 2 SS)
------------------------------------------------------------



It managed to get through the predictor step, but then died on the corrector step with "Unable to initiate self-sustaining chain reaction".

User avatar
Jaakko Leppänen
Site Admin
Posts: 2387
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by Jaakko Leppänen » Sat Nov 14, 2015 1:40 pm

The zeros result from the tallies not being updated until the batch is completed.
- Jaakko

cgentry7
Posts: 24
Joined: Mon Jun 03, 2013 10:59 pm
Security question 1: No
Security question 2: 92

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by cgentry7 » Thu Dec 03, 2015 6:28 pm

It appears that the revised reinitrng.c file you provided me worked. So the problem was related to the random number seed exceeding the range of a long integer. I'm not sure if you wanted to add the revisions to this post or not, so I left them off.

Sorry for the delay in response, I just wanted to confirm that the fix actually worked and it took a little while for my jobs to make it through the queue and for me to parse the results for comparison to our benchmark results.

Cole

User avatar
Jaakko Leppänen
Site Admin
Posts: 2387
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: Number of Depletion Regions effecting Max NPG / MPI ratio

Post by Jaakko Leppänen » Thu Dec 03, 2015 8:07 pm

Great, thanks for testing this. I'll include the fix in the next update.
- Jaakko

Post Reply