[ET Trac] [Einstein Toolkit] #434: Keep track of masked-out volume in CarpetMask

Einstein Toolkit trac-noreply at einsteintoolkit.org
Mon Nov 7 18:13:27 CST 2011


#434: Keep track of masked-out volume in CarpetMask
--------------------------+-------------------------------------------------
  Reporter:  eschnett     |       Owner:  eschnett
      Type:  enhancement  |      Status:  review  
  Priority:  minor        |   Milestone:          
 Component:  Carpet       |     Version:          
Resolution:               |    Keywords:          
--------------------------+-------------------------------------------------

Comment (by rhaas):

 Doesn't seem to fix it yet though it's a step in the right direction.

 {{{
 INFO (CarpetReduce): Finalise the weight on level 0
 INFO (CarpetReduce): Testing weight
 INFO (CarpetReduce): Reduction weight sum: 2009.5
 INFO (CarpetReduce): Volume of map #0: 2156
 INFO (CarpetReduce): Simulation domain volume:  2068
 INFO (CarpetReduce): Additional excised volume: 88
         2     2.000
 WARNING level 1 in thorn CarpetReduce processor 0 host
 horizon.tapir.caltech.edu
   (line 127 of
 /home/rhaas/ET/arrangements/Carpet/CarpetReduce/src/mask_test.c):
   -> Simulation domain volume and reduction weight sum differ
 WARNING level 1 in thorn CarpetReduce processor 0 host
 horizon.tapir.caltech.edu
   (line 127 of
 /home/rhaas/ET/arrangements/Carpet/CarpetReduce/src/mask_test.c):
   -> Simulation domain volume and reduction weight sum differ
 LoopControl timing statistics:
 Loop #0 "CarpetSurfaceSetup_all":
    total count: 3   total setup: 9.53674e-07   total calc: 0.00165701
    avg calc: 0.000552336   avg first calc: 0.000618935
    avg improvement: 11%   saved: 0.000199795 seconds
 Loop #1 "MaskBase_SetMask_all":
    total count: 4   total setup: 0   total calc: 0.000102043
    avg calc: 2.55108e-05   avg first calc: 3.00407e-05
    avg improvement: 15%   saved: 1.81198e-05 seconds
 Total calculation time: 0.00182819 seconds; total saved time: 0.000217915
 seconds
 --------------------------------------------------------------------------------
 Done.
 }}}

 If I make the spherical surface large enough so that everything is
 excluded (radius=500) I get:
 {{{
 INFO (CarpetReduce): Finalise the weight on level 0
 INFO (CarpetReduce): Testing weight
 INFO (CarpetReduce): Reduction weight sum: 0
 INFO (CarpetReduce): Volume of map #0: 2156
 INFO (CarpetReduce): Simulation domain volume:  1001
 INFO (CarpetReduce): Additional excised volume: 1155
         2     2.000
 LoopControl timing statistics:
 Loop #0 "CarpetSurfaceSetup_all":
    total count: 3   total setup: 0   total calc: 0.00162721
    avg calc: 0.000542402   avg first calc: 0.000565052
    avg improvement: 4%   saved: 6.79493e-05 seconds
 Loop #1 "MaskBase_SetMask_all":
    total count: 4   total setup: 0   total calc: 9.67979e-05
    avg calc: 2.41995e-05   avg first calc: 2.5034e-05
    avg improvement: 3%   saved: 3.33786e-06 seconds
 Total calculation time: 0.00177097 seconds; total saved time: 7.12872e-05
 seconds
 --------------------------------------------------------------------------------
 Done.
 WARNING level 1 in thorn CarpetReduce processor 0 host
 horizon.tapir.caltech.edu
   (line 127 of
 /home/rhaas/ET/arrangements/Carpet/CarpetReduce/src/mask_test.c):
   -> Simulation domain volume and reduction weight sum differ
 WARNING level 1 in thorn CarpetReduce processor 0 host
 horizon.tapir.caltech.edu
   (line 127 of
 /home/rhaas/ET/arrangements/Carpet/CarpetReduce/src/mask_test.c):
   -> Simulation domain volume and reduction weight sum differ
 WARNING level 1 in thorn CarpetReduce processor 0 host
 horizon.tapir.caltech.edu
   (line 127 of
 /home/rhaas/ET/arrangements/Carpet/CarpetReduce/src/mask_test.c):
   -> Simulation domain volume and reduction weight sum differ
 }}}
 This works on single MPI process (which seems to point towards the ghost
 zones).

-- 
Ticket URL: <https://trac.einsteintoolkit.org/ticket/434#comment:6>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit


More information about the Trac mailing list