[cig-commits] r18248 - short/3D/PyLith/trunk
brad at geodynamics.org
brad at geodynamics.org
Tue Apr 19 14:19:33 PDT 2011
Author: brad
Date: 2011-04-19 14:19:33 -0700 (Tue, 19 Apr 2011)
New Revision: 18248
Modified:
short/3D/PyLith/trunk/TODO
Log:
Updated notes on current issues.
Modified: short/3D/PyLith/trunk/TODO
===================================================================
--- short/3D/PyLith/trunk/TODO 2011-04-19 20:50:41 UTC (rev 18247)
+++ short/3D/PyLith/trunk/TODO 2011-04-19 21:19:33 UTC (rev 18248)
@@ -4,37 +4,11 @@
Strip whitespace from UniformDB values and names.
-* Memory model [MATT and BRAD]
-
- (1) The memory logging for distribution logs Creation and Stratification
- as Creation. Because we set the logging stages in PyLith and these
- are lumped together within the same Sieve routine, we cannot
- separate them.
-
- (2) When we deallocate the distributed mesh, we do not deallocate all of
- the memory. This suggests some object is leaking memory.
-
- (3) The memory used by the distributed mesh is much greater than
- that predicted by the memory model. MATT- Could this be related to
- each processor having lots of gaps in the range of cells and
- vertices it has?
-
- - Can Fusion be leaking memory?
-
- - How do we separate parallel stratification?
-
- IntSections - Matt improved distribution to use allow IntSections
- over vertices or cells rather than vertices + cells. This cuts
- differences down to 75%.
-
- Need to check sum, because we may not be tracking deallocation.
-
* Field split with nonlinear solver [MATT]
- The custom preconditioner is not being used in the linear or
- nonlinear solvers. Matt added code to Solver.cc::initialize(). The
- switch for detecting whether the preconditioner is to be used is not
- correct.
+ Matt created MyMatGetSubMatrix() in Solver.cc to retrieve the
+ preconditioner. Currently this generates a segfault when using the
+ nonlinear solver with field split.
PREVIOUS NOTES
@@ -64,14 +38,43 @@
end up in the x-dof with zeros for the y-dof, the opposite of the
correct behavior.
+ cd examples/bar_shearwave/tri3/tmp
+ pylith
+
+ ERROR: Nonconforming object sizes. Invalid number of local indices
+ 256, should not be greater than 140!
+
(2) Add time dataset to vertex_fields and cell_fields. [BRAD]
(3) Add creation of .xmf metadata file for ParaView/Visit. [BRAD]
- Need a flag to use s->GetIndicesRaw() in scatter creation
- - examples/bar_shearwave/tri3
+* Memory model [MATT and BRAD]
+ (1) The memory logging for distribution logs Creation and Stratification
+ as Creation. Because we set the logging stages in PyLith and these
+ are lumped together within the same Sieve routine, we cannot
+ separate them.
+
+ (2) When we deallocate the distributed mesh, we do not deallocate all of
+ the memory. This suggests some object is leaking memory.
+
+ (3) The memory used by the distributed mesh is much greater than
+ that predicted by the memory model. MATT- Could this be related to
+ each processor having lots of gaps in the range of cells and
+ vertices it has?
+
+ - Can Fusion be leaking memory?
+
+ - How do we separate parallel stratification?
+
+ IntSections - Matt improved distribution to use allow IntSections
+ over vertices or cells rather than vertices + cells. This cuts
+ differences down to 75%.
+
+ Need to check sum, because we may not be tracking deallocation.
+
* Fault preconditioner [BRAD]
FaultCohesiveLagrange::calcPreconditioner() [need unit test]
More information about the CIG-COMMITS
mailing list