Message by Enrico Sain:
Hello,
I´m having a problem with my project while performing the simulation. As suggested in previous threads, I ran SPIS-NUM under Eclipse in order to understand where the error comes from.
This is the log I get in the debugger console:
opening log file for SPIS/NUM: ./SpisNum.log
Default output path for SPIS/NUM set to ./
Default input path for SPIS/NUM set to ./
Serialising object of type [Lspis.Top.Default.GlobalParameter
in file ./globalParOut.gp
Serialising object of type [Lspis.Top.Default.LocalParameter
in file ./localParOut.lp
GC Warning: Repeated allocation of very large block (appr. size 4194304):
May lead to memory leak and poor performance.
GC Warning: Repeated allocation of very large block (appr. size 33554432):
May lead to memory leak and poor performance.
Starting building simulation from UI parameters
Simulation was built from UI parameters
Exception in thread “main” java.lang.NullPointerException
at spis.Vol.VolMesh.ThreeDUnstructVolMesh.mapToInternalNodeIndexing(ThreeDUnstructVolMesh.java:1597)
at spis.Vol.VolMesh.ThreeDUnstructVolMesh.mapToInternalNodeIndexing(ThreeDUnstructVolMesh.java:1684)
at spis.Vol.VolMesh.ThreeDUnstructVolMesh.setInternNodeNumbering(ThreeDUnstructVolMesh.java:1769)
at spis.Solver.Poisson.ConjGrad3DUnstructPoissonSolver.init(ConjGrad3DUnstructPoissonSolver.java:325)
at spis.Solver.Poisson.ConjGrad3DUnstructPoissonSolver.(ConjGrad3DUnstructPoissonSolver.java:169)
at spis.Top.Simulation.SimulationFromUIParams.init(SimulationFromUIParams.java:500)
at spis.Top.Simulation.SimulationFromUIParams.(SimulationFromUIParams.java:235)
at spis.Top.Top.NumTopFromUI.(NumTopFromUI.java:79)
at spis.Top.Top.SpisTopMenu.launchMenu(SpisTopMenu.java:146)
at spis.Top.Top.SpisTopMenu.main(SpisTopMenu.java:1553)
Since I don´t know Java I´d very much appreciate any help in intepreting the error messages!
Many thanks in advance,
Enrico
P.S. I´m using SPIS 3.7RC09
Message by Jean-Francois Roussel:
Hi,
I never saw that (null pointer at that stage).
Neither did I see"Repeated allocation of very large block (appr. size 4194304): May lead to memory leak and poor performance"
=> are you running a very big model?
Do you have the same issue with the official version 3.6?
JFR
Message by Enrico Sain:
Hello,
Thank you for your answer. Even if I still don’t know the exact reason, the issue was probably related to a “wrong” 3D Unstructured Volume Mesh. I simply used the GMSH option “Optimize quality of tetrahedra” in Tools/Options/Mesh/Advanced and the problem disappeared!
I’m running a model with around 100000 tetrahedra and a ratio between the biggest element size and the smallest one of around 100. I’m also experiencing some memory issues with the same model: while the solver is building the simulation and exactly after the S/C circuit txt file is read (I can see it in the Standard Log window), the memory usage increases in a few seconds from around 300 MB to a peak of 2000 MB. Then, when the simulation starts running, it rapidly decreases towards 800 MB and it remains almost constant until the end of the data extraction process. Is this a normal memory usage profile?Because of this software behaviour I cannot increase the mesh resolution more than the actual one, otherwise the memory limit is reached (in my case 2500 MB) and SPIS crashes and closes. I read on the forum that SPIS can theoretically reach 1.e6 tetrahedra for every GB of RAM and I’m wondering why I cannot reach more than 1.e5 tetrahedra with 2.5 GB available.
I’ve haven’t tried the 3.6 version yet, but if you say that it’s more stable I will do it soon!
Best regards,
Enrico