B1 running error in runbeam

msmaran · April 7, 2021, 4:09pm

My laptop config :
intel® Core™ i3-7100U Processor (3M Cache, 2.40 GHz )
4 GB DDR4 RAM
1 TB SATA HDD
I’ve run the B1 example Beam on 10^6 and got doserate. When i run the particles 10^7, my laptop got hang. I need Appox 10^11 to 10^13 particles to get result.
What should i do to get solution?

weller · April 7, 2021, 4:32pm

/vis/disable
/tracking/storeTrajectory 0
/run/printProgress 1000000
/run/beamOn 1000000000
wait for results…

measure the time for smaller numbers (say, 1e6 particles), before launching the “big run” to know what to expect… if 1e6 particles takes 1 second, 1e11 particles takes more than 1 day.

keep in mind that you might have to stitch different runs to get your desired number of 1e11 particles, since the event number is an int-value (maximum is around 2e9).

you should also consider multi-threading, you will find examples if you look for it

allison · April 7, 2021, 4:53pm

Why do you need so many particles? Can you not run, say 10^6, and scale the result to your experiment?

msmaran · April 7, 2021, 5:27pm

I know, int-value couldn’t exceed 10^10. I tried to check how large the value.
Can you please, help me to consider multi-threading?
I couldn’t enable it.

mkelsey · April 7, 2021, 5:31pm

Did you build your Geant4 installation with multithreading enabled? Building and Installing from Source — Geant4 Installation Guide 10.7 documentation

Did you specify the multithreaded run manager in your main()?
https://geant4-userdoc.web.cern.ch/UsersGuides/ForApplicationDeveloper/html/GettingStarted/mainProgram.html#g4runmanager

msmaran · April 7, 2021, 5:35pm

I installed Geant4 from miniconda and run the exampleB1 from conda environment.

AI-Pranto · April 8, 2021, 5:13am

@msmaran

After digging some time,

github.com

conda-forge/geant4-feedstock/blob/63a9dd199c0962f3a533629bea6ed0fba20caa89/recipe/build_geant4.sh#L38


      
            $(find examples -name 'CMakeLists.txt')
          
          mkdir geant4-build
          cd geant4-build
          
          cmake                                                          \
                -DCMAKE_BUILD_TYPE="${CMAKE_BUILD_TYPE}"                 \
                -DCMAKE_INSTALL_PREFIX="${PREFIX}"                       \
                -DBUILD_SHARED_LIBS=ON                                   \
                -DGEANT4_BUILD_CXXSTD=17                                 \
                -DGEANT4_BUILD_MULTITHREADED=ON                          \
                -DGEANT4_BUILD_TLS_MODEL=global-dynamic                  \
                -DGEANT4_INSTALL_DATA=OFF                                \
                -DGEANT4_INSTALL_DATADIR="${PREFIX}/share/Geant4/data"   \
                -DGEANT4_INSTALL_EXAMPLES=ON                             \
                -DGEANT4_INSTALL_PACKAGE_CACHE=OFF                       \
                -DGEANT4_USE_FREETYPE=ON                                 \
                -DGEANT4_USE_GDML=ON                                     \
                -DGEANT4_USE_QT=ON                                       \
                -DGEANT4_USE_SYSTEM_CLHEP=ON                             \
                -DGEANT4_USE_SYSTEM_EXPAT=ON                             \

Yes, It’s built with MULTITHREADED=ON support for conda install. To reduce run time you have to specify multithreaded run manager in your main() for event-level parallelism via multi-threading.