Project

General

Profile

Bug #14085

Memory leak running LArG4 with NEST

Added by Gianluca Petrillo about 3 years ago. Updated about 3 years ago.

Status:
Closed
Priority:
Normal
Category:
Simulation
Target version:
-
Start date:
10/05/2016
Due date:
% Done:

100%

Estimated time:
Spent time:
Occurs In:
Experiment:
LArIAT, LArSoft
Co-Assignees:
Duration:

Description

Pawel Kryczynski reports that a job running generation and simulation with some custom simulation code is able to run only up to 50 events before exhausting GPVM memory.

20161004-pawel-massif.out.bz2 (11.8 MB) 20161004-pawel-massif.out.bz2 Massif profiling of the job Gianluca Petrillo, 10/05/2016 09:37 AM

Associated revisions

Revision eb2cd944 (diff)
Added by Gianluca Petrillo about 3 years ago

Fixing memory leak in NEST (issue #14085)

Our crippled NEST code does still a lot of useless stuff.

Revision 665ec811 (diff)
Added by Gianluca Petrillo about 3 years ago

Fixing memory leak in NEST (issue #14085)

Our crippled NEST code does still a lot of useless stuff.

Revision aac247a4 (diff)
Added by Gianluca Petrillo about 3 years ago

Fixing memory leak in NEST (issue #14085)

Our crippled NEST code does still a lot of useless stuff.

History

#1 Updated by Gianluca Petrillo about 3 years ago

  • Status changed from New to Assigned
  • Assignee set to Gianluca Petrillo

#2 Updated by Gianluca Petrillo about 3 years ago

Pawel has run valgrind's massif tool on the job.

#3 Updated by Gianluca Petrillo about 3 years ago

Massif points to a problem in NEST.

NEST code in LArG4 is a mutilated version of some original NEST code taken somewhere, sometimes.
Among the mutilations it had to suffer, there is likely the deletion of a particle object, that is now never deleted and finally leaked.

#4 Updated by Gianluca Petrillo about 3 years ago

  • Status changed from Assigned to Resolved
  • % Done changed from 0 to 100

The free pointer has been replaced with a std::unique_ptr. No other action has been taken on the code.
The commit has been tested by Pawel, confirming the physics is unchanged and memory usage is stable even after O(1000) processed events.
The fix is merged in larsim develop branch.

#5 Updated by Gianluca Petrillo about 3 years ago

  • Status changed from Resolved to Closed


Also available in: Atom PDF