GEANT4 Resource Profiling

Getting Started
1

Howdy,

My research group has started running GEANT4 on our cluster which is due for some upgrades. We’d like to take a look at what we’ve been running to optimize the new cluster hardware towards GEANT4 as it’s currently optimized for MCNP and reactor codes and creating a corner for GEANT4 would be great for this growing area of usage. Are there any inbuilt tools to profile something like a 1E5 primary particle run for CPU and Memory over time to see what is the current bottleneck and select hardware the is more balanced for these applications?

I think we’re running into raw clock speed limitations as some of us have done some crude comparison testing on consumer grade desktop and laptop systems and have seen greater raw performance but the endurance needed isn’t there on these systems for the length of runs required to meet statistics.

Thanks

2

Not as such - some alpha/beta development of builtin profiling was started, but we no longer have effort available to support this.

Assuming this is a Linux cluster, then standard tools like perf are probably your best bet. An exemplar use of this can be found here: GitHub - amadio/g4run: Simple Generic Program to Run Geant4 Simulations

3

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.