sherpa is hosted by Hepforge, IPPP Durham
close Warning: Can't synchronize with repository "(default)" (/hepforge/svn/sherpa does not appear to be a Subversion repository.). Look in the Trac log for more information.
Modify ↓

Opened 9 years ago

Closed 8 years ago

Last modified 8 years ago

#349 closed defect (fixed)

TTbar mpi integration stalls for N_CPU >5 or so

Reported by: holsch Owned by: Stefan Hoeche
Priority: major Milestone: rel-2.2.0
Component: ME Generator Version: 2.1.1
Keywords: Cc:

Description

With the attached runcard, the mpirun or mpiexec command gets stuck and never starts the integration when using -n 6, seems to depend on the machine as well. Anyway -n 20 defninitely does not work. Workaround that was used for the integration of the validation plots was (as suggested by Frank S) to use: mpirun -n 30 Sherpa -f Run.dat MI_HANDLER=None -e 1 STABLE[6]=1 STABLE[24]=1 STABLE[23]=1 STABLE[25]=1

Attachments (1)

Run.dat​ (1.2 KB) - added by holsch 9 years ago.
Run card ttbar leptonic decays

Download all attachments as: .zip

Change History (3)

Changed 9 years ago by holsch

Attachment: Run.dat​ added

Run card ttbar leptonic decays

comment:1 Changed 9 years ago by Frank Siegert

Note (to myself): Peter Onyisi mentioned that this happens for him only when the top width is set to 1.5 GeV as in the ATLAS defaults. With higher widths he doesn't seem to have this problem. We should test whether this is just by accident or whether that gives us a hint to a solution of this problem.

comment:2 Changed 8 years ago by Frank Siegert

Resolution: → fixed
Status: new → closed

This has been fixed by Stefan in r28511 by making the Decay_Channel class MPI aware. The bugfix will be included in release 2.2.1.

Last edited 8 years ago by Frank Siegert (previous) (diff)

Modify Ticket

Change Properties
Action
as closed The owner will remain Stefan Hoeche.

Add Comment

E-mail address and name can be saved in the Preferences.

Attachments ↑ Description ↑
 
Note: See TracTickets for help on using tickets.