[MAP] G4beamline release 2.16 is now available on Hopper
Tom Roberts
tjrob at muonsinc.com
Tue Jan 14 12:58:41 EST 2014
With assistance from the NERSC consultants, the compiler issue at NERSC has been
resolved without any changes to the G4beamline source. G4beamline 2.16 is now
available on Hopper; it passes all regression tests on a worker node, including
MPI. To use it (bash):
source /project/projectdirs/map/bin/g4bl_latest.sh
... use g4blmpi as usual
There are also setup scripts for specific versions of G4beamline; older versions
remain usable by simply source-ing the appropriate script.
NOTE: these setup scripts work only for bash. If you want to use some other
shell, you'll have to re-implement the scripts for it (the scripts in G4beamline
itself are OK, just the MAP setup scripts need to be re-implemented). Possibly
you could run bash, source the script, and then run your shell; I haven't tried
that.
I'm working on Edison now, and will send out another email when it is ready....
This new release has many new features including boolean operations among solids
and tessellated solids; it also has a new 'detector' command that outputs the
energy deposited in the detector into the NTuple.
You can download it from:
http://g4beamline.muonsinc.com
There are still compiler issues at NERSC that prevent it from running on either
Hopper or Edison. I am working on them. I do not expect these to require changes
to G4beamline itself, just to the scripts that set it up at NERSC -- three weeks
ago it built and ran fine on both supercomputers.
It has not yet been installed on FermiGrid.
Tom Roberts
More information about the MAP-l
mailing list