sphenix-tracking-l AT lists.bnl.gov
Subject: sPHENIX tracking discussion
List archive
[Sphenix-tracking-l] Good news regarding magnetic field and memory usage
- From: Christof Roland <christof.roland AT cern.ch>
- To: Hugo Pereira Da Costa via sPHENIX-tracking-l <sphenix-tracking-l AT lists.bnl.gov>
- Subject: [Sphenix-tracking-l] Good news regarding magnetic field and memory usage
- Date: Wed, 4 May 2022 09:35:09 +0200
Hi Everybody,
The 1 hour field prining exercise turned into a two day project, as usual.
the good news is the memory profile using the pruned field is consistently
below 4GB for 1000 events.
Some observations:
ACTS insiste on a cube field map. Cutting is to a cylinder shape gives an
exceptions throw and job termination.
/phenix/u/bogui/data/Field/sphenix3dtrackingmapxyz.root is now the field in
|x,y,z| <110cm i.e 6% of the original field volume.
We have setterrs for the field name and path in the code, but they are
largely ignored and thee flied used is hardwired in the code.
We may want to fix this.
The SimpleKFProp and CASeeder still load the big map from the node tree using
PHFieldUtility.
I dont see an easy interface in PHFieldUtility to read a map from file. We
may want to change this.
CASeeding and KFProp land a separate fieldmap. If we switch this to the ACTS
interface we could save
another 0.5 - 0.75 GB of memory.
Cheers
Christof
- [Sphenix-tracking-l] Good news regarding magnetic field and memory usage, Christof Roland, 05/04/2022
Archive powered by MHonArc 2.6.24.