Dependency of the results on the used CPU

Physics Processes, Models and Cross Sections
1

Good morning! I am running a simulation where I need to estimate the number of charged particles generated and releasing energy in a series of aluminum blocks, which contain a layer of boron-10. While doing this, I used two different computers with different processors: one with an i7, and the other with an Intel Xeon. I downloaded the code from the same GitHub repository, but I’m getting completely different results, without errors in both cases. Why might this be?

  • i7 pc:
    Geant4 Version: Geant4 11.2.0
    Operating System: Ubuntu 24.04.1 LTS
    Compiler/Version: gcc version 13.2.0
    CMake Version: cmake version 3.28.3

  • Xeon pc:
    Geant4 Version: Geant4 11.1.1
    Operating System: CentOS Stream release 9
    Compiler/Version: gcc (GCC) 11.5.0
    CMake Version: cmake version 3.26.5

figura

2

The results are quite different, but you’re not comparing like with like if I understand correctly. The i7 results are with Geant4 11.2, whilst the Xeon are with 11.1.1? What do you get if you run 11.2 on both the i7 and Xeon (and similarly with 11.1.1 on both)?

3

Thanks for the information! I tried to repeat the simulation on the same machine, with an i7, using both versions of Geant4, 11.2 and 11.1.1, and the results seem to show that there is a problem with the source in the newer version. In the graph I report the results, the code and the computer are the same.
figura

4

Thanks! I’ve moved this to the physics category as this is likely a difference/bug in settings between 11.1 and 11.2. Could you post which physics list you are using (or the code if you’ve created this yourself), and any settings you are applying please?

5

Thank you for the response. In my simulation, I used the QGSP_BERT_HP physics list.

1 Like
6

@civanch, @ribon, @pico, could you take a look at this please?

7

The physics list QGSP_BERT_HP in G4 11.2 has bugs that are not yet fixed (in G4 11.1 or earlier versions, it is fine. The bugs are in 11.2).
Please use another _HP physics list, e.g. QGSP_BIC_HP or FTFP_BERT_HP.

1 Like
8

Thank you very much for the response. Now I will try to use one of the suggested libraries.

9

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.