[pjoynt]

Hi All,


I am trying to run a simple irregular wave train in REEF3D SFLOW in preparation for real domain. I've noticed that with my current control and ctrl files, if I set F 60 to 0.0 then the simulation crashes before the first time step but if I set it to 0.01 it is fine. Note that my domain extents for a Z value of -60 to 10. All file are attached and I am using DiveMESH v_220405 and REEF3D v_220419.


Any idea if this is bug?


Cheers,
Patrick

[kamath]

Hi Patrick,

I am unable to reproduce the error you report.
The file works for me even with F 60 0.0
(I ran only up to 6 iterations)

However, I did change M 10 from 12 to 8 (to match the max on my machine)

Could you check if changing the number of procs helps in your case?
(That will point towards a decomp problem, which we can investigate)

What error do you get when reef3d crashes?

[pjoynt]

Originally Posted by kamath

Hi Patrick,

I am unable to reproduce the error you report.
The file works for me even with F 60 0.0
(I ran only up to 6 iterations)

However, I did change M 10 from 12 to 8 (to match the max on my machine)

Could you check if changing the number of procs helps in your case?
(That will point towards a decomp problem, which we can investigate)

What error do you get when reef3d crashes?

Hi Arun,


Looks like it fails with the following error (attached). Changing number of procs didn't fix it, note that I am using the newer intel processors that have 8 fast cores and 8 slower cores - maybe this has something to do with it (12th Gen Intel Core i9-12900k)? I know that it has 16 cores but 24 threads so you can't actually run the model with that combo (or at least it fails whenever I try), it seems to like the 1:2 ratio not 1:1.5 ratio...


Patrick

[kamath]

Ok, so it is a seg fault as expected.

Could I bother you to carry out a tedious procedure of finding the maximum procs this might run for?


Starting with M 10 1? Need to just look for the simulation to startup, say 1-2 outer iterations.

That will probably answer your question

[valgrinda]

To me it looks like F 60 is the problem, which is 0 here. Try setting up the case so that F 60 will be equal to the still water depth. In addition, A 211 should be 4 and N 47 0.25.

Given your wave conditions, FNPF maybe be the overall better choice for you.

[Chalwwz]

Hi,

Similar to Arun, I couldn't reproduce the error. However, I do suggest three options that may help:

1) In your control.txt, you commented out S 1 comment, however 'S 1' is still in capital letter. Make it a lower case letter, because otherwise the program might still consider it as a valid command even after the // sign. This might lead to some unexpected solid object that might influence you simulation;

2) If possible, shift you coordinate system upwards by 60 so the vertical coordinate starts from 0, it should work when you set your domain from 0 to 70 and set the water depth at 60.

3) try to use a newer code, if you are using Github to get the source code from REEF3D, then try to use 'git checkout release_candidate' to use the release_candidate version.

Hope one of the solutions solve your problem.

[OceanMan]

Hi REEF3D team and everyone.
I run wave propagation over a submerged bar using the REEF3D::NSEWAVE model (modified from the REEF3D::CFD tutorial). I don't know why the height gauges (P 51) is an abnormally large value as shown on the attached image file. While height gauges still work with the REEF3D::CFD model.
Has anyone encountered a similar problem?
I would appreciate your help!