Killing the job if kinetic E blows.
3 lines of code changed in 1 file:
Updates to conform to new ExprLib FieldManager interface.
3 lines of code changed in 1 file:
Allow GPU code generation for specific compute capabilities: 2.0, 2.1, 3.0, 3.5. Default is now 2.0
Anything else specified with "enable--gencode=<n>" aborts configure and reports a meaningful message.
Note: We only support code generation for Fermi and Kepler GPU architectures now.
31 lines of code changed in 3 files:
1. Add a method to BCData to access the boundary conditions (BoundCondBase) vector associated with a given BCData class.
2. Add an enum and its access method to BoundCondBase that allows us to infer the data type of the value set by a boundary condition (BoundCondBase). This will make infering the data for BoundCond much more robust when doing the dynamic_cast. Currently, components infer the type of the BoundCond by doing logical deductions on the variable name, which is prone to errors. This new method will allow you to deduce data type based on the input specification.
56 lines of code changed in 5 files:
No comment.
0 lines of code changed in 1 file:
1. Fix a subtle bug in Wasatch that showed up when trying to remove the Auxiliary variable yesterday.
2. Add a new function to BCUtils that allows one to access the boundary value specified for a given variable.
333 lines of code changed in 3 files:
Add a method to ComputeSet to access the vector of ComputeSubset managed by a given set. This should make it easy to iterate over subsets using iterators. I use BOOST_FOREACH which is awesome.
5 lines of code changed in 1 file:
Further improve ARCHES multi-threaded output
98 lines of code changed in 8 files:
Add some extra offerings for sensible multi-threaded output.
3 lines of code changed in 3 files:
revert even more code to stop the segfaults.
106 lines of code changed in 3 files:
revert my previous commit on the ComputeSet which caused the universe to throw segfaults.
1 lines of code changed in 1 file:
Make ARCHES multi-threaded table/property/model output more coherent.
13 lines of code changed in 2 files:
Fix race condition in infrastructure timing code (DataArchiver).
3 lines of code changed in 1 file:
rip out the obsolete Auxiliary boundary condition.
0 lines of code changed in 3 files:
add a method to ComputeSet that allows us to access the vector of ComputeSubset(s). This makes it easy for one to iterate through a compute set, and specifically with the conveninent BOOST_FOREACH macro.
4 lines of code changed in 1 file:
Make find_val() functions thread-safe (Interp base and derived classes) within ARCHES ClassicTableInterface.
505 lines of code changed in 1 file:
Only calling problem setup on source terms which are actually used.
This should reduce memory usage because sources that aren't used won't
be allocated.
Adding a function to check for allocated sources of a specific type.
Simplifying the Shaddix enthalpy model given the above changes.
70 lines of code changed in 9 files:
Add a regression test for the 1-equation scalar variance model.
2 lines of code changed in 1 file:
Add in the property models for the scalar variance, and the source term required for a 1-eqn model.
596 lines of code changed in 10 files:
Finish with ARCHES::IntrusionBC thread safety
Fix XML element in methane_rmcrt iput file (<solveBoundFlux> --> <solveBoundaryFlux>)
Add item to SVN ignore
126 lines of code changed in 2 files:
change BoundCondBase::getBCType__NEW to getBCType. I don't know which of the following is better though, appending __OLD or __NEW...
20 lines of code changed in 13 files:
make the print() method of BCDataArray const.
2 lines of code changed in 2 files:
- Fix a memory leak that I introduced yesterday.
- rename a variable to be more descriptive.
31 lines of code changed in 2 files:
Turned off ray tracing loop, which will simplify the overhaul process.
Now passing in VRFlux, boundFlux and radVolq into rayTraceKernel.
Does not run.
72 lines of code changed in 5 files:
remove some leftover comments.
1 lines of code changed in 1 file:
repair some misleading nomenclature when using allpatches and localpatches in Wasatch.
5 lines of code changed in 1 file:
sched_rayTrace()
- removed requires for carry forward variables. This should have been removed in the last
commit.
rayTrace()
- changed
old_dw->getModifiable() to new_dw->getModifiable()
( I'm not sure how or why you would ever want to modify a variable in the old_dw).
4 lines of code changed in 1 file:
Created a separate task to carryForward variables that are not being computed but
just shuffled from old_dw -> new_dw. The Unified scheduler has a problem with this and it's needed
for GPU tasks.
93 lines of code changed in 4 files:
Use a map rather than list to avoid excess mallocs for the patch set that is associated with operators.
6 lines of code changed in 3 files:
Adding a scalar balance calculator that computes R = In - Out + Accum.
527 lines of code changed in 2 files:
support C++11 style hash map
42 lines of code changed in 5 files:
Update some GPU example components and corresponding input files.
6 lines of code changed in 9 files:
Eliminate linking with libcuda when turning on GPU support, only need libcudart.
Generate new configure script.
2 lines of code changed in 2 files:
rayTrace_dataOnion()
- created new function updateSumI_ML() (still in rough draft form)
This will be usefull when the boundary fluxes and the virtual radiometer are added.
*This commit should not change the answers.
249 lines of code changed in 2 files:
Some minor cosmetic changes.
30 lines of code changed in 3 files:
updateSumI()
- no longer passing the unused variables domainLo & domainHi around
3 lines of code changed in 2 files:
*RT Update Needed*
Some corrections to:
1) Intrusion inlets: The wrong index pointer was being used in the
one-off direction from the wall.
2) Wall BC application to momentum: **This is a band-aid** Now applying
all wall bc application in BoundaryCondition::wallVelocityBC(), which
should cover domain walls and intrusions. This sets the A and ConvA's
appropriately for the walls and for the normal velocity cells one-off
from the wall cell. The implementation was simply to reuse Kumar's
mmbcvelocity fortran code for each coefficient set making this a little
cumbersome, but it does get the job done. We need a little cleanup to
now remove some of the defunct BC tasks that are being performed.
3) Removed scalar diffusion at inlets. This is nothing but trouble for
mass balances.
Also, while I was at it, some minor cleanup of some unused (mms) functions.
258 lines of code changed in 6 files:
rayTrace()
VirtualRadiometer:
- moved constant variable calc to problem setup.
- using a new global struct to pass around the constants
- moved ray_location calculation to a new function
- moved ray_direction calculaton to a new function
DivQ:
- moved ray_location calculation to a new function.
*These changes should not change the answers.
215 lines of code changed in 2 files:
Change a few settings in one of the failing Warches tests. We'll have to see if this fixes the Jenkins builds. Must update GS for that test.
1 lines of code changed in 1 file:
Fix nasty memory leak in Warches. Also add information about VarLabel address in the VarLabel debug stream.
7 lines of code changed in 3 files:
Now pulling vel_CC from dw and passing into the turbulence function
instead of creating a temporary variable, copying vel_CC into that
variable, setting the boundary condition.....etc. Old way avoided a template issue that was
easy to fix.
57 lines of code changed in 6 files:
to update Enthalpy coal model
2 lines of code changed in 1 file:
update the surface reaction rate for charoxidation
1 lines of code changed in 1 file:
remove some warches tests from dbg regression.
4 lines of code changed in 1 file:
fix task dependency error in Warches computation of hatted velocities.
3 lines of code changed in 1 file:
Add regression tests for Warches BCs.
10 lines of code changed in 1 file:
RayTrace()
global:
- now using enum to define cell face indicies, easier for code readability.
solveBoundaryFlux:
- moved dirIndexOrder, dirSignSwqp, locationIndexOrder and locationShift to constructor.
These global vars are only assigned once and therefore multi-thread safe.
* This commit should NOT change the answers.
60 lines of code changed in 2 files:
updateSumI()
- passing in the ray direction instead of inv_ray_direction. Less code clutter.
BoundaryFlux section:
- encapsulated code for finding the ray location and direction on a cell face.
This commit *will* change the answers. 1e-16 for divQ and 1e-15 for boundaryFlux.
92 lines of code changed in 2 files:
Use input spec to drive Wasatch momenthm RHS computation in Warches. This will allow us to do much needed regression testing on this new Warches code.
445 lines of code changed in 6 files:
Enable Wasatch to calculate the Arches' hatted-velocity RHS (plural). This is done seamlessly via automatic input-file spec creation. The current implementation works for the following:
1. All constant density periodic problems
2. All constant density problems with Inlets, Outlets, and Pressure BCs (as defined by Arches)
3. All aforementioned problems with FE, RK2, and RK3
The remaining challenges have to do with the way Arches/Wasatch handle walls and variable density problems. We understand why the variable density problems differ: using two ghost cells for density interpolations and algebraic differences when constructing convective fluxes. A quick performance showed speedups of 3-to-4.5 times (depending on what tasks we exclude). This performance test was based on the 3D periodic, Taylor-Green vortex problem. By default, the momentum calculation is EXCLUDED in repository Warches code. To enable it, search for #ifndef WASATCH_IN_ARCHES and uncomment those lines (along with their corresponding #endif). Also uncomment lines [402,409] in Arches.cc. Send me an email if you are having problems with this.
471 lines of code changed in 10 files:
Fixed scoping error with direction_vector inside of RAY_SCATTERING.
This *will* change the isoScattering results by 1e-16.
3 lines of code changed in 1 file:
Reduce some of the output, make other gnuplot friendly
0 lines of code changed in 2 files:
updateSumI()
- Using Vectors for tMax and tDelta.
- replaced / by *
- simplified conditional tree
* This should not change the answers. Minor speedup on RMCRT_1L_perf test.
47 lines of code changed in 2 files:
Get a reference of Properties into shaddix enthalpy model.
46 lines of code changed in 4 files:
Fixing the advect_2L_MI regression test by adding the pLocalizedMPM to the
refine task.
0 lines of code changed in 2 files:
Sped up rayTrace_dataOnion by 10X by commenting out SCIDebugs in the inner most loop.
Placed reflection code into a separate function.
* This should not change the answers.
58 lines of code changed in 2 files:
to remove useless outputs
0 lines of code changed in 1 file:
Removed extra ';' to quiet compiler warning.
1 lines of code changed in 1 file:
Need to use Uintah's mpi_defs include instead of hardcoding mpi.h.
0 lines of code changed in 2 files:
Fix spacing, alphabetize.
4 lines of code changed in 1 file:
Added an mpi_hang test. This test can be used to reproduce a bug we
have seen in a few mpi libraries where calling MPI_Abort on a single
processor does not cause the mpi library to kill all processes (as it
should). This can (sometimes) cause sus to hang until the job runs
out of time. Note, this bug usually (only?) showed up when running on
at least 2 nodes. If on 1 node, the library behaved correctly.
(Note, there was a bug in sus that caused MPI_Abort to only be called
on a single processor in some situations, and this bug has been fixed.
However, it is very likely that there will be cases in sus when only
one process will detect an abort condition and thus only one processor
will call MPI_Abort... thus it is necessary for the MPI library to
actually handle this correctly...)
130 lines of code changed in 3 files:
Implement thread locks on flux limiter and upwind operators so that the use of the advective velocity is thread safe.
Also remove a const qualifier from the FluxLimiter apply_to_field() method since it is going to modify the mutex and reset the advective velocity member variable.
61 lines of code changed in 4 files:
M configure
M configure.ac
Fix previous commit of configure... When checking for hash map must be
in C++ mode. Also, somehow the initial "DEF_HASHMAP=None" was lost in
the last commit.
14 lines of code changed in 2 files:
Updates to allow for the use of the Boost hashmap (if Boost is enabled).
This is mostly necessary for the GPU build as we are forced to not use
the (default/normal) tr1 hashmap (as nvcc can't compile it) and this
causes us to default to the (deprecated) gnu hashmap which causes a
zillion warning messages to be displayed during compilation.
M configure
M configure.ac
- Fix help message for ray scatter flag (was reporting to use --ray-scatter
when it should (and now does) say --enable-ray-scatter.
- Move hash map check to after boost check.
- If boost is found, then use the boost hash map.
M include/sci_hash_map.h
- Add in USE_BOOST_HASHMAP mapping.
- Add some whitespace to make file a little more humanly parsable.
M include/sci_algorithm.h
- Fix title of file in file to be accurate (fixes a copy/paste error).
M CCA/Components/Schedulers/DWDatabase.h
- Changed the #else to #elif HAVE_TR1_HASHMAP. I think this is
correct... There are a number of possible hash map settings
(gnu/tr1/boost/etc, but the original #else lumped them all into
'else' section of the file. My guess is that none of the other
possible options were ever used and thus we didn't see a problem, but
if they had been, then a compilation error would have occurred as I
think the #else section is only applicable for the tr1 hashmap.
M StandAlone/tools/uda2nrrd/uda2nrrd.cc
- Doesn't use a hash map so remove the #include.
M Core/Grid/Variables/VarLabelMatl.h
- The VarLabelMatl is what we use for a hash key, and in the case of
using the boost hash table, we need to have the "hash_value()"
function as a friend of this class. Added it in.
- Removed the unnecessary ';' from the end of a couple of function
definitions.
M Core/Datatypes/sub.mk
- There are a whole bunch of (I believe?) deprecated files in this
directory (used in the old SCIRun). Most of them are not even
compiled (not in the sub.mk file). Removed them from the tree.
- Indented and alphabetized as necessary.
D Core/Datatypes/CurveMesh.h
D Core/Datatypes/cd_templates_fields_3.cc
D Core/Datatypes/TriSurfMesh.h
D Core/Datatypes/CCTensorField.cc
D Core/Datatypes/CMakeLists.txt.bak
D Core/Datatypes/NCVectorField.cc
D Core/Datatypes/NCVectorField.h
D Core/Datatypes/StructHexVolMesh.h
D Core/Datatypes/Mesh.h
D Core/Datatypes/Clipper.cc
D Core/Datatypes/StructCurveMesh.h
D Core/Datatypes/cd_templates_fields_2.cc
D Core/Datatypes/TensorField.cc
D Core/Datatypes/cd_templates_fields_6.cc
D Core/Datatypes/FieldInterfaceAux.cc
D Core/Datatypes/MultiLevelField.h
D Core/Datatypes/PrismVolMesh.h
D Core/Datatypes/NCScalarField.h
D Core/Datatypes/Mesh.cc
D Core/Datatypes/NCTensorField.cc
D Core/Datatypes/Clipper.h
D Core/Datatypes/NCTensorField.h
D Core/Datatypes/ImageMesh.h
D Core/Datatypes/cd_templates.cc
D Core/Datatypes/MaskedLatVolMesh.h
D Core/Datatypes/cd_templates_fields_1.cc
D Core/Datatypes/CCVectorField.h
D Core/Datatypes/cd_templates_fields_5.cc
D Core/Datatypes/TensorField.h
D Core/Datatypes/LatVolMesh.h
D Core/Datatypes/GenericField.h
D Core/Datatypes/Field.cc
D Core/Datatypes/HexVolMesh.h
D Core/Datatypes/FieldInterfaceAux.h
D Core/Datatypes/Field.h
D Core/Datatypes/FieldAlgo.h
D Core/Datatypes/TetVolMesh.h
D Core/Datatypes/FieldIterator.h
D Core/Datatypes/FieldInterface.h
D Core/Datatypes/QuadSurfMesh.h
D Core/Datatypes/cd_templates_fields_0.cc
D Core/Datatypes/CCVectorField.cc
D Core/Datatypes/cd_templates_fields_4.cc
D Core/Datatypes/StructQuadSurfMesh.h
D Core/Datatypes/CCScalarField.h
D Core/Datatypes/CCTensorField.h
- All these files were from the old SCIRun days and are not used now.
Cleaning them out of the repository.
283 lines of code changed in 56 files:
Add a computes for pLocalizedMPMLabel to scheduleInitialize
0 lines of code changed in 2 files:
rayTrace_dataOnion()
- adding capabilities present in 1L Raytrace()
- added computation of radiationVolqLabel
- removed redundant calculation for DyDx and DzDX
- eliminated one very expensive exp() call.
* This should not change the answers.
42 lines of code changed in 1 file:
added performance test ( RMCRT with data onion ) to the nightly RT
1 lines of code changed in 1 file:
Surgically construct graphs for Warches. In this commit, we do this for scalar transport equations and all force-on-graph expressions. This is an example of cherry-picking expressions from Wasatch. The pitfall to doing this is when you construct graphs that share one or more expressions - this leads to multiple computes errors. Note that this is different from graph cleaving in Wasatch which does this flawlessly. For the moment, we have to live with this in Warches and try to work around it.
37 lines of code changed in 3 files:
Use STATE_NONE for the dependent table variables when using Warches. please fix your input files appropriately.
1 lines of code changed in 1 file:
Added performance test for RMCRT 1L and added it to the nightly RT. Using radiometer, reflections, boundary flux and
scattering chunks of code.
9 lines of code changed in 2 files:
rip out support for Wasatch TurbulentViscosity calculation in Warches. The TurbulentViscosity will be calculated along with the Wasatch momentumPartialRHS in the new Warches framework.
106 lines of code changed in 7 files:
Delete the newly added pLocalizedMPMLabel_preReloc
1 lines of code changed in 1 file:
in Virtual Radiometer Section
- moved clamp of user input (orient) to problemSetup.
- moved variables that only need to be computed once outside of the ray loop.
- general code consolidation
in _solveDivQ section
- deleted unused variables and empty comments.
* This commit should NOT change the answers
64 lines of code changed in 1 file:
Add pLocalizedMPMLabel stuff here and there so the code will run.
10 lines of code changed in 1 file:
Backing out some of the changes associated with the misguided ExprLib API modifications yesterday.
28 lines of code changed in 9 files:
Boundary Flux section:
- clean out unused variables.
- minor formatting changes
- removed switch-case block, since you don't need it.
* This should not change answers.
34 lines of code changed in 1 file:
Previous fix to particle creator didn't fix the switcher. This does.
I just LOVE maintaining code that no one will ever use again!
12 lines of code changed in 2 files:
sorting that recognizes scientific notation.
0 lines of code changed in 2 files:
Cleaning up my earlier mess a bit. Don't add p.localizedMPM to the permanent
particle state for implicit MPM
4 lines of code changed in 1 file:
Optimization:
- only call the expensive exp() function once instead of twice in the inner most loop.
Small speed up of 10ish %.
* This commit should not change the answers.
11 lines of code changed in 1 file:
Fix the updated filename in the RT script.
0 lines of code changed in 2 files:
removed negative sign from tau_s calculation.
0 lines of code changed in 2 files:
Code Consolidation:
- created function findRayDirection()
Changed _mTwister -> mTwister. Prefix _ is for global variables
**This commit should not change the answers.
70 lines of code changed in 2 files:
Allow the deformation gradient to be modified for particles that have localized,
depending on the erosion algorithm chosen.
57 lines of code changed in 7 files:
Add in a small stability fix for mineralization.
5 lines of code changed in 1 file:
Rip out patch ID as an argument to most expression library interfaces per changes in ExprLib.
17 lines of code changed in 9 files:
code consolidation:
created function findStepSize().
moved DyDx and DzDx calculation to just inside of patch loop. No need to recompute it every ray.
** this commit should NOT change the answers.
65 lines of code changed in 2 files:
Fixed a threadsafety issue.
Moved a globally defined advection operator variable d_OFS to a locally defined struct
that it passed around.
129 lines of code changed in 8 files:
addExchangeToMomentumAndEnergy()
Fixed glaring performance issue:
- created a separate task for single material ICE problems. Reducing task execution time
from 8.4 -> 1.8 sec for Taylor Green.
113 lines of code changed in 2 files:
add_vel_FC_exchange:
performance improvements. Added if(numMatls ==1) section
Copy the StaticArray data into temporary arrays before accessing them. Static Arrays are
slow. Improved this code from 8.4 sec -> 3.6 sec for taylorGreen.
87 lines of code changed in 2 files:
Fixed some couts that were not commented out.
10 lines of code changed in 1 file:
putting lastWriteTimeLabel into DW
0 lines of code changed in 2 files:
Added rate dependence stuff for Hamid
193 lines of code changed in 1 file:
to update the DQMOM.cc
0 lines of code changed in 1 file:
to optimize the data access to source term, make code faster
55 lines of code changed in 1 file:
1) Fixed a bug with the density assignment when computing velocity for mass flow inlets.
2) Change density assignment in extra cells and intrusion inlets so that the interpolated value is consistent with the mixture fraction.
50 lines of code changed in 3 files:
One more time. Capture both stdout and stderr to a sus.log file.
2 lines of code changed in 1 file:
Moved delete defState downstream to the bottom of the particle loop.
Previously, defState was deleted and then passed into a function.
0 lines of code changed in 2 files:
Use unique strings to identify label names for lastWriteTime. If you don't
then there will be problems if multiple modules are used.
3 lines of code changed in 5 files:
send stdout to the screen and a file.
2 lines of code changed in 1 file:
to fix a bug for restart
6 lines of code changed in 1 file:
consistent use of d_ for all globally defined variables. This make is easy to find where they're
being assigned.
fixed one potential thread safety issue with min_mass_in_a_cell.
91 lines of code changed in 2 files:
to update the char oxidation model
1 lines of code changed in 1 file:
to update the coal particle heat transfer model
27 lines of code changed in 1 file:
Added the configure option
--enable-ray-scatter
to the nightly regression testing machine configuration.
Pruned out old config.
2 lines of code changed in 1 file:
New regression test. Testing DDT code
0 lines of code changed in 2 files:
Added the scattering test to the regression tester.
2 lines of code changed in 1 file:
Comment out line that only is used for verification testing of scattering
1 lines of code changed in 1 file:
Remove throw when non-scattering case is run
with --enable-ray-scatter in configure line.
0 lines of code changed in 1 file:
Apparently NVCC doesn't like BOOST_FOREACH. Replace with old-style iterator loop to make NVCC happy.
6 lines of code changed in 1 file:
M CCA/Components/ProblemSpecification/ProblemSpecReader.cc
Removed the comment that said "// FIXME fix error message..." that was
left in the code to remind the programmer to fix "the WORST error
message in the history of programming." :)
Also replaced the exact same "worst error message..." in another
section of the file to give the better error message.
4 lines of code changed in 1 file:
This is a commit: Fix the WORST error message in the history of programming.
1 lines of code changed in 1 file:
Fix a mistake from Isaac's commit last Friday. Thanks to Abhishek for finding this.
2 lines of code changed in 2 files:
Ensure wall emissivity doesn't exceed one.
Added warnings for low ray numbers.
Code cleanup.
Made variable for number of divQ rays more intuitive.
Removed redundant documentation.
27 lines of code changed in 8 files:
<Component>.py: Fixed typo.
GenerateGoldStandards.py & runSusTests.py
- added aurora to the hack that indicates which machines have gpus.
RMCRT_test_1L.ups
- this is now a symbolic link to the orginal input file.
17 lines of code changed in 11 files:
doAnalysis()
If the variable lastWriteTime is not in the dw then set it to 0.
The user may want to restart an uda that did not have the Module turned on.
This logic allows that.
Added missing logic to avoid calculation if time is outside of start & stop times.
73 lines of code changed in 11 files:
Use boost_foreach in a few places to clean up some code
Minor reformatting
21 lines of code changed in 1 file: