January 2009 Commit Log

Number of Commits:
106
Number of Active Developers:
11
guilkey 2009-01-30 22:36 Rev.: 43027

Holy cow this TestSuite stuff is CRAP! This couldn't possibly be more
contrived. I'm forced to kind of use it because I'm too stupid to get
make to build an executable and an object file from the same code. I'll
likely create a new TestMatrix3 directory next week in the image of
IteratorTest which is what John suggested in the first place.

Blech! Who hired Wayne anyway?

0 lines of code changed in 4 files:

jas 2009-01-30 22:20 Rev.: 43026

Add mpmice_amr to buildbot script.

0 lines of code changed in 2 files:

dav 2009-01-30 19:45 Rev.: 43025

Turn off debug message.

1 lines of code changed in 1 file:

guilkey 2009-01-30 16:13 Rev.: 43024

Fixed bug in the new polarDecompositionRMB method.

3 lines of code changed in 1 file:

guilkey 2009-01-30 16:12 Rev.: 43023

Added a main() to testmatrix3 so I could add some tests without having to
deal with the entire confusing TestSuite business. Hey Wayne, ever hear of
a README?

65 lines of code changed in 2 files:

harman 2009-01-30 15:55 Rev.: 43022

plotScript:
- tightened the curve fit tolerance

almgrenMMS.ups:
- removed extra scalar transport section
- lowered the viscosity
- tighened the residual tolerance on the pressure solver

This shows second order accuracy for both u & v.

11 lines of code changed in 3 files:

guilkey 2009-01-29 16:37 Rev.: 43014

Added a new iterative polarDecomposition method that will hopefully be more
robust than the current version. As yet untested.

170 lines of code changed in 2 files:

jas 2009-01-28 18:55 Rev.: 43008

Back out revision 43001. Add cstdio so that remove() is found.

0 lines of code changed in 2 files:

guilkey 2009-01-28 17:54 Rev.: 43007

Add in unrotating and rotating of stress before and after update, respectively.
Load up array of internal state variables to pass to fortran and unpack on the
other end.

47 lines of code changed in 1 file:

dav 2009-01-28 16:38 Rev.: 43006

M udaReaderMTMD/udaReaderMTMDPluginInfo.h
M udaReaderMTMD/udaReaderMTMDPluginInfo.C

Fix compilation warning.

M udaReaderMTMD/testavtudaReaderMTMDFileFormat.C.in

std::memcpy needs #include <cstring> to compile.


7 lines of code changed in 5 files:

jas 2009-01-27 20:07 Rev.: 43002

Remove return code testing, since the function didn't have a return code.
gcc-4.4 picked up a compiler error with static function call that was missing
an object for the call to remove().

2 lines of code changed in 1 file:

harman 2009-01-27 16:48 Rev.: 42996

Now computing order of accuracy for both u & v components of velocity.

47 lines of code changed in 4 files:

guilkey 2009-01-27 15:50 Rev.: 42994

Additional work on the model. Doesn't run yet.

119 lines of code changed in 2 files:

csafe-tester 2009-01-27 15:09 Rev.: 42993

Only send mail when an uda has been replaced

0 lines of code changed in 2 files:

csafe-tester 2009-01-27 15:06 Rev.: 42992

Copy sus.log.txt along with the uda to the goldstandard directory

-Email John when there is a change to the GS.

13 lines of code changed in 1 file:

luitjens 2009-01-27 11:30 Rev.: 42990

Some more modifications to the output interval logic

10 lines of code changed in 1 file:

luitjens 2009-01-27 10:24 Rev.: 42989

Set compare_uda's default exit mode to exit and not wait.

2 lines of code changed in 1 file:

luitjens 2009-01-27 10:09 Rev.: 42988

Fixed an off by 1 error in the data output frequency.

2 lines of code changed in 1 file:

sshankar 2009-01-26 18:48 Rev.: 42983

Added fix for AMR visualization

483 lines of code changed in 1 file:

jthornoc 2009-01-26 17:59 Rev.: 42982

Starting a simple interface for Paula's RMCRTing code.

There are a lot of comments in here to help Paula get going.

R stands for Reverse (Monte Carlo Ray Tracing)


0 lines of code changed in 4 files:

luitjens 2009-01-26 14:26 Rev.: 42977

Changed the logic for dumping out timesteps/checkpoints when using the timestepInterval parameters to work off the global timestep. This will cause restarts to produce output as if the job never terminated. Previously timesteps would not line up.

2 lines of code changed in 1 file:

jas 2009-01-26 13:49 Rev.: 42976

Add the MIT License to every *.cc, *.h,*.F, and sub.mk file.

70832 lines of code changed in 4518 files:

luitjens 2009-01-24 14:03 Rev.: 42973

Add svnStat and svnDiff to the RT tests

2 lines of code changed in 1 file:

luitjens 2009-01-24 00:19 Rev.: 42972

don't compare uda on uknown types in the RT.

0 lines of code changed in 2 files:

luitjens 2009-01-23 22:09 Rev.: 42971

Use the variable basis to determine the testing ranger for a variable. This prevents a bug where compare_uda would use a combination of node/cell index ranges which could pull data from the wrong patch.

19 lines of code changed in 1 file:

harman 2009-01-23 19:30 Rev.: 42970

components.xml
- Arches tests are now turned on

dx.tst
- changed the resolutions
- eliminated the delt_init replace strings

arches_mms.m
- LnError = sqrt(error)/sqrt(exact)

16 lines of code changed in 4 files:

jthornoc 2009-01-23 19:12 Rev.: 42969

Further corrections to the mms input file.

2 lines of code changed in 1 file:

jas 2009-01-23 16:26 Rev.: 42968

If either sus or compare_uda fails, print out the errors to the console.

2 lines of code changed in 1 file:

jas 2009-01-23 16:18 Rev.: 42967

Revert regression_tester to a previously working version.

10 lines of code changed in 1 file:

jas 2009-01-23 14:33 Rev.: 42965

Comment out the explode2D_amr for now.

2 lines of code changed in 1 file:

luitjens 2009-01-23 13:37 Rev.: 42964

Fixed a bug in the logic for containsPoint type functions. Previously the function would compare the point position to the position of the high and low index of the patch. However, the high index is actually larger than the highest point in the patch by 1 in each dimension. I also converted the new versions of these functions to use integer arithmatic instead of floating point in order to hopefully avoid problems with floating point error in the future.

14 lines of code changed in 1 file:

harman 2009-01-23 13:19 Rev.: 42963

empty xml tags are returned as ''.

Only analyze the result if the comparison util is something the user has specified.

14 lines of code changed in 5 files:

jas 2009-01-23 11:15 Rev.: 42962

Add the integrated simulation switcher test and the explode2D_amr test to
mpmice_amr.

34 lines of code changed in 1 file:

guilkey 2009-01-22 19:59 Rev.: 42961

Adding storage stuff for internal state variables. Far from done,
but too much work to risk losing.

129 lines of code changed in 2 files:

jthornoc 2009-01-22 19:22 Rev.: 42956

Changed some parameters for the almgren mms for the regression tester.

2 lines of code changed in 1 file:

harman 2009-01-22 18:04 Rev.: 42955

Added comparison utility and plotting script for Arches test.

110 lines of code changed in 4 files:

jthornoc 2009-01-22 17:49 Rev.: 42954

Replaced input file with updated one.

61 lines of code changed in 1 file:

amorris 2009-01-22 17:42 Rev.: 42951


Fixing up the change in r42788

MomentumSolver.cc
buildLinearMatrixVelHat()
- removed Tau timing statements.

the stop call was remaining.


0 lines of code changed in 1 file:

jthornoc 2009-01-22 17:36 Rev.: 42950

Fixed a bug in the error calculation for the v-component of velocity and cleaned things up.

15 lines of code changed in 1 file:

harman 2009-01-22 17:27 Rev.: 42949

Initial commit:
nightly order-of-accuracy test using almgrenMMS.ups.

234 lines of code changed in 4 files:

harman 2009-01-22 16:28 Rev.: 42948

problemSetup_BulletProofing()
- remove min_boundary_cells & cell_deletion_dilation test from bulletproofing

3 lines of code changed in 1 file:

guilkey 2009-01-22 13:42

Change the ordering of how elements of the Matrix3s are loaded and unloaded
into 1D arrays.

24 lines of code changed in 2 files:

luitjens 2009-01-22 10:55 Rev.: 42943

Only finalize MPI if it has been initialized and not already finalized.

8 lines of code changed in 1 file:

luitjens 2009-01-22 10:43 Rev.: 42942

Fully qualify the namespace to avoid a compile error.

1 lines of code changed in 1 file:

jas 2009-01-22 00:52 Rev.: 42941

Add the csafe group to the basic MIT license.

29 lines of code changed in 1 file:

dav 2009-01-21 20:45 Rev.: 42940

Cleanup for fix for exit_handler

1 lines of code changed in 1 file:

dav 2009-01-21 20:44 Rev.: 42939

Fix for exit_handler

9 lines of code changed in 1 file:

dav 2009-01-21 16:26 Rev.: 42930


Cleaning up the "thread cleanup" routines. Added "handleCleanup()". Note, this has
been tested with Uintah (which only (currently) uses one thread) and not with SCIRun
or any (truly) multi-threaded code... though it 'should work'. The issue is involved
is when (if) thread::exit is called for all threads. As best I can tell, if exitAll()
is called, then individual threads never call "Thread::exit()". Therefore I had to
place the handleCleanup() in the exitAll() routine. (Again, this should work, but hasn't
been tested on multiple threads.) It is also called in the Thread::exit() in case only
a single thread wants to exit.

(If you plan to use this facility for non-Uintah, see previous commit for more info
on this facility (or read comments in the code).)

M Thread.cc
M Thread.h

M Thread_pthreads.cc

* Update of exitAll to call handleCleanup()
* Cosmetics:
* moved "{" into standard location.
* renamed 't' to 'thread_priv' so I could search for it.

75 lines of code changed in 3 files:

luitjens 2009-01-21 14:10 Rev.: 42926

When there are overlapping patches (possible with extracells) compare uda gives priority to the patch that has that cell/node in the interior. This patch updates the tiebreaker to give priority to the patch that has the cell/node index within extra cells instead of extra nodes.

With extra nodes there is a corner case where two patches own the same node. In this case both nodes should have the same value. However, compare uda uses the same patch map for all variables (based on nodes). This causes problems when looking at a cell which is on this corner case. The current code treated the cell as a node and believed they would have the same value. This is incorrect because with cells one patch owns the cell and the other doesn't and thus only the patch that owns the cell is gaurenteed to have the correct value.

5 lines of code changed in 1 file:

ffjhl 2009-01-20 21:57 Rev.: 42923

Fixed a bug in radialDirection. tt is not supposed
to be of unit length. Note that axis/height2 is
the axial unit vector; projOnAxis is the projection
of pbot onto axis.

2 lines of code changed in 1 file:

guilkey 2009-01-20 19:50 Rev.: 42922

Added the beginnings of a model driver for the Sandia GeoModel, aka Kayenta.
Also generating a number of essentially empty fortran files that would be
replaced by a user who had licensed access to this model.

730 lines of code changed in 7 files:

guilkey 2009-01-20 18:07 Rev.: 42921

Fix outputProblemSpec error.

1 lines of code changed in 1 file:

csafe-tester 2009-01-20 17:21 Rev.: 42920

added explode2D_amr test.
-exploding 2D container with AMR.
-this test starts from a previously generated checkpoint (112K timesteps), runs for 50 timesteps and regrids 5 times.

1 lines of code changed in 1 file:

dav 2009-01-20 16:53 Rev.: 42919

M configure.ac
M configure
M include/sci_defs/uintah_testdefs.h.in

Record the fortran underscore policy for use in .cc files. (Note, this is
somewhat of a hack... if possible, you should use the fspec facility.) To
get access to the (#define) FORTRAN_UNDERSCORE, #include sci_defs/uintah_defs.h.

M Packages/Uintah/CCA/Components/MPM/ConstitutiveModel/HypoElasticFortran.h

Cosmetics. Cleaned up #includes. Removed no longer used/supported SGI stuff.

M Packages/Uintah/CCA/Components/MPM/ConstitutiveModel/HypoElasticFortran.cc

- Fixed (and turned on) the fortran calls. Should work with any fortran compiler
now. Note, this is somewhat of a hack... Normally the use of the fspec utility
is recommended... however, due to the simplicitly of this need, and some problems
with geting fspec to work (it isn't fully implemented for all fortran types),
this will do for now.
- Cleaned up the #includes (alphabetized, removed duplicates). Removed old SGI stuff.
- Placed the 'return type' of functions on separate line.

M Packages/Uintah/tools/fspec.pl.in

Added some white space to the out file so that it is easier to follow the code
in the generated "<code>_fort.h" files.

158 lines of code changed in 7 files:

luitjens 2009-01-20 16:02 Rev.: 42918

Re-enabled load balancing at compile time.

1 lines of code changed in 1 file:

luitjens 2009-01-20 15:52 Rev.: 42917

Rolled back 42618. Particle sets do not exist prior to relocate in the newDW until after relocate is called.

2 lines of code changed in 1 file:

luitjens 2009-01-20 13:57 Rev.: 42916

Reverted a few patches from the last few months that were required due to an incorrect bug fix. The bug they were fixing still exists and i'm looking into a correct fix. Also disabled load balancing without regridding. The code should be working now.

117 lines of code changed in 8 files:

charlesreid 2009-01-20 10:26 Rev.: 42913

Adding header files for file-checking and proc0cout

16 lines of code changed in 1 file:

csafe-tester 2009-01-18 23:17 Rev.: 42910

Modified latex configure file

0 lines of code changed in 2 files:

csafe-tester 2009-01-18 23:17 Rev.: 42909

Fixed latex comments

1 lines of code changed in 1 file:

charlesreid 2009-01-18 18:40 Rev.: 42908

Fixing formatting in MixingRxnTable; modifying getState method and cleaning up comments in TabPropsTable class

140 lines of code changed in 3 files:

luitjens 2009-01-17 11:09 Rev.: 42907

don't do the grid checksums when running in serial.

4 lines of code changed in 1 file:

csafe-tester 2009-01-16 19:24 Rev.: 42906

Include source files in doxygen.

1 lines of code changed in 1 file:

guilkey 2009-01-16 18:53 Rev.: 42905


Add MIGUtilsF.F which has some subroutines that I'm evidently going to
need. Also, change the functionality of the routines in MIGUtils.cc a
little bit.

130 lines of code changed in 3 files:

guilkey 2009-01-16 18:22 Rev.: 42904

Changed HelperC.cc to MIGUtils.cc, and get rid of Helper.F. Clean up the
sub.mk with Dav's help.

41 lines of code changed in 4 files:

luitjens 2009-01-16 13:39 Rev.: 42903

Trimmed the output of a patch.

1 lines of code changed in 1 file:

luitjens 2009-01-16 13:20 Rev.: 42902

added a SCI_DEBUG for outputting the grid on each processor.

Fixed a bug in the verifyGrid functioon.

5 lines of code changed in 1 file:

luitjens 2009-01-16 12:53 Rev.: 42901

Bug in my last commit

1 lines of code changed in 1 file:

luitjens 2009-01-16 10:35 Rev.: 42900

Added a grid sanity check when asserts are enabled. This verifies everyone has the same grid at the end of the BNRRegridder.

78 lines of code changed in 2 files:

luitjens 2009-01-15 16:28 Rev.: 42899

move the wait for debugger command before the cleanup function

3 lines of code changed in 1 file:

csafe-tester 2009-01-15 15:47 Rev.: 42898

Removed this file (to avoid confusion) as we don't support CMake now... Also testing tester commits. -Dav

0 lines of code changed in 1 file:

harman 2009-01-15 15:42 Rev.: 42897

Don't need this old file... also testing tester email.

0 lines of code changed in 1 file:

harman 2009-01-15 15:20 Rev.: 42896

doxygen is now run on blaze instead of inferno2

0 lines of code changed in 2 files:

harman 2009-01-15 15:18 Rev.: 42895

Updated to configure file to be compatable with newer versions of doxygen

233 lines of code changed in 1 file:

harman 2009-01-14 12:25 Rev.: 42893

Code consolidation and new patch interfaces

Deleted the fortran routine explicit_vel()

RHSSolver.cc
- added explicitUpdate_stencilMatrix() this now contains the fortran code.
(this will eventually be combined with explicitUpdate() )

-CalculateHatVelocity()
using new patch iterators and explicitUpdate_stencilMatrix()

* This should not change the answers.

150 lines of code changed in 5 files:

jthornoc 2009-01-14 12:08 Rev.: 42892

Removed a print statement that someone committed.

0 lines of code changed in 1 file:

luitjens 2009-01-14 10:16 Rev.: 42891

Cleanup variables created in the testmodel.


8 lines of code changed in 1 file:

luitjens 2009-01-13 12:04 Rev.: 42881

Added LBwoRegrid test to RT

2 lines of code changed in 1 file:

guilkey 2009-01-12 20:39 Rev.: 42879

This now works. Call to fortran is still commented out until
the fortran interface stuff is included.

19 lines of code changed in 2 files:

dav 2009-01-12 17:53 Rev.: 42874

Added the ability to specify an "abort" handing/cleanup function for a
thread. If this routine is specified, then it will be called from the
segfault/segbus signal abort handling routine. This allows the main
program to tell its thread how to clean up. It was inspired by the
need for Uintah to tell MPI to abort when one process in a large MPI
run dies.

M Core/Thread/Thread.h

* Added (g/s)etCleanupFunction() which allows a program to give its
thread a clean up routine to be called if the program segfaults
(segbus's).
* Indentation. Cleanup of '{'s.

M Core/Thread/Thread.cc

* Added (g/s)etCleanupFunction() code.
* Cleanup/Indentation.

M Core/Thread/Thread_pthreads.cc

* Only real coding update is this...

> Thread::ptr2cleanupfunc funcPtr = Thread::self()->getCleanupFunction();
> if( funcPtr != NULL ) {
> (*funcPtr)();
> }

...which calls the cleanup function if it was specified.

* Indentation/Cleanup.

M Packages/Uintah/StandAlone/sus.cc

* Added abortCleanupFunc that will be called in the Uintah thread segfaults (segbus, etc).
* Cleanup #includes a little.

444 lines of code changed in 4 files:

luitjens 2009-01-12 16:01 Rev.: 42872

Include extra cells on coarse boundaries when migrating data to a new processor. This potentially fixes the loadbalance/regrid/particle send set/unknown variable bug that Todd keeps hitting. (Fixing this bug is starting to feel like whack-a-mole).


29 lines of code changed in 2 files:

harman 2009-01-12 15:47 Rev.: 42870

calculateVelocitySource()
- removed unused parameters
- now using new patch iterators.

This should not change the answers.

17 lines of code changed in 3 files:

jas 2009-01-12 14:46 Rev.: 42869

Check if an input file can be loaded. If it can't, see if the first
character in the d_upsFilename is a '/' and strip that out. This allows
files that are included via the <include> directive to be found.

15 lines of code changed in 1 file:

guilkey 2009-01-12 13:37 Rev.: 42867

Comment out call to hooke_incremental for now so
that updraft build will work.

1 lines of code changed in 1 file:

luitjens 2009-01-12 10:47 Rev.: 42866

Only initialize the profile weights if cost profiling is being used.

1 lines of code changed in 1 file:

harman 2009-01-09 19:08 Rev.: 42862

buildLinearMatrix()
scalarLinearSolve()
-replaced triple nested loops with a cell iterator

43 lines of code changed in 1 file:

harman 2009-01-09 18:01 Rev.: 42860

removed loop over dimensions surrounding
d_momSolver->Solve
this should have been eliminated on r42788

1 lines of code changed in 1 file:

harman 2009-01-09 17:52 Rev.: 42859

removed remnants of computeVel_fort.h

0 lines of code changed in 2 files:

guilkey 2009-01-09 17:15 Rev.: 42858

Try to use fortran code to get the stress update. Doesn't run yet.

53 lines of code changed in 2 files:

harman 2009-01-09 16:28 Rev.: 42857

code consolidation:
- removed fortran:computeVel.f no longer needed

RHSSolver.cc:
calculateVelocity()
- now passing in the pressure and density instead of constVelocityVars.

MomentumSolver.cc:
buildLinearMatrixVelHat()
- inside of conditional
if(d_pressure_correction){
}
replaced individual calls to fortran subroutines with equivalent
call to calculateVelocity()

*This should not change the answers

9 lines of code changed in 5 files:

guilkey 2009-01-09 15:13 Rev.: 42856

Write error handling code in C++ which will (eventually) allow me to use
Uintah exception handling and hang large clusters.

51 lines of code changed in 3 files:

guilkey 2009-01-09 13:20 Rev.: 42855

Added Helper.F, which contains the routines needed by the fortran models
for outputting error messages.

36 lines of code changed in 3 files:

harman 2009-01-09 11:08 Rev.: 42853

add " around Helium_RT

1 lines of code changed in 1 file:

ffjhl 2009-01-08 21:54 Rev.: 42852

fixed a bug on -matl,mat for -varsummary. minor change in usage for
puda.cc.

2 lines of code changed in 2 files:

harman 2009-01-08 16:33 Rev.: 42848

-Added helium_RT to the Arches regression tester.

-Turned on <EKTCorrection>

1 lines of code changed in 1 file:

harman 2009-01-08 16:29 Rev.: 42847

buildLinearMatrix()inside of

if(doing_EKT_now){

}
moved the copyOut statement so it is executed on every integration step. This
was a bug introduced on r 41391

*This _will_ change the answers when EKT is turned on.

1 lines of code changed in 1 file:

jas 2009-01-08 16:20 Rev.: 42846

Remove the friend designation for Mult.

8 lines of code changed in 2 files:

luitjens 2009-01-08 15:32 Rev.: 42844

Temperaroly disable load balancing without regridding until a bug is fixed.

7 lines of code changed in 1 file:

harman 2009-01-08 15:07 Rev.: 42843

Removed loop over dimensions for the scheduling d_momSolver->solve()
This should have been in a previous commit.

This should not change the answers.

10 lines of code changed in 4 files:

guilkey 2009-01-06 20:26 Rev.: 42834

My Debian machine (gcc 4.3.2) won't swallow having all of the parameters
for my fortran function having the same name. So, give them different
names.

1 lines of code changed in 1 file:

guilkey 2009-01-06 20:18 Rev.: 42833

First step towards adding fortran based constitutive models in Uintah.
This model is basically the same as HypoElastic, but will call fortran
code to do the numerics. Currently, only HOOKECHK is called, which
checks the validity of the input parameters.

2270 lines of code changed in 9 files:

harman 2009-01-05 17:35 Rev.: 42823

MomemtumSolvers.cc
BuildLinearMatrix()
- eliminated unused variables from calculateVelocity()
- consolidated boundary conditions
RHSSolver.cc
calculateVelocity()
- New patch iterators
- moved fortran code to here.

** This should not change the answers

53 lines of code changed in 3 files:

harman 2009-01-05 14:27 Rev.: 42822

computeMMSError()
- now using new patch interators
- consolidated "put" calls
- eliminated extra .initialize(0.0)

140 lines of code changed in 1 file:

ffjhl 2009-01-02 18:41 Rev.: 42815

Added pressureBC by adapting code from explicit.

254 lines of code changed in 2 files:

ffjhl 2009-01-01 19:05 Rev.: 42814

fixed the right count for argc.

2 lines of code changed in 1 file:

ffjhl 2009-01-01 16:31 Rev.: 42813

added -cyl option so a cylinder can be created inside the given
geometry object. This is done primarily for image data (cropping).

60 lines of code changed in 2 files:

Generated by StatSVN 0.4.0