Use consistent device information on advecting velocity in UpwindInterpolant.cc
5 lines of code changed in 1 file:
Fix some const correctness issues in UpwindInterpolant.cc
5 lines of code changed in 1 file:
Fix some const correctness with the FluxLimiter stuff.
73 lines of code changed in 2 files:
Make a few expressions not GPU runnable since they require "time" which does not yet transfer properly.
4 lines of code changed in 1 file:
Some housekeeping on GPU code in Unified Scheduler.
56 lines of code changed in 2 files:
Slight modification to the source term as far as checking the
flammability limits. This will allow for the fuel stream to have
some non-fuel species.
10 lines of code changed in 3 files:
Fix multi-threaded deadlock scenario in Unified Scheduler.
2 lines of code changed in 1 file:
put the boundary conditions for density and mixture fraction back in the pressure projection test cases
223 lines of code changed in 3 files:
overload the multiplication operator for IntVector to allow multiplication by a constant.
6 lines of code changed in 1 file:
Tag a few more expressions as GPU-runnable. Thanks to Abhishek for pointing these out!
17 lines of code changed in 3 files:
Turnoff advect_2L_3D_slabs on DBG builds. It randomly crashes.
1 lines of code changed in 1 file:
Moving the rarely used CorrugEdgeGeomPiece to the newly created CodeAttic to
reduce maintenance of code that isn't used and has a questionable
implementation.
0 lines of code changed in 5 files:
Removing the test_corrugated problem since it tests a feature that hasn't been
used in 10 years and since Todd rocked the boat.
0 lines of code changed in 1 file:
formatting.
DDT1::computeBurnLogic()
The outputInterval and checkpointInterval variables must "put" something in the DW
or an exception is thrown on debug builds. Set those variables to a benign value as a default.
36 lines of code changed in 3 files:
Forgot to change Wasatch::NONE to Wasatch::NOTURBULENCE in Warches.
1 lines of code changed in 1 file:
1. Change the NONE enum for TurbulenceModelName to NOTURBULENCE.
2. Allow the TaskInterface to lockAllFields by passing a boolean flag. This may be needed in Warches to avoid cherry-picking locked fields.
16 lines of code changed in 8 files:
Bring Warches back to life after Jeremy's changes to the aeraFraction calculation. Also, use explicit instantiation of a Wasatch templated function that is used in Warches and that was causing the mac compilers to throw an compile error.
28 lines of code changed in 2 files:
Cleaning up.
50 lines of code changed in 6 files:
also save mDot through the faces.
0 lines of code changed in 2 files:
Cleaning up warnings.
13 lines of code changed in 8 files:
Missing some ";"
3 lines of code changed in 1 file:
Adding RMCRT capability into the wall heat transfer model.
118 lines of code changed in 4 files:
Update to the area fraction calculation to only occur if forced.
194 lines of code changed in 6 files:
Added comments that the smoothCylinder is not designed to be used with the union,difference,intersection
operator.
ParticleCreator::countAndCreateParticles()
Throw an exception if a smooth geometry piece tries to create a particle that lies outside the
computational domain.
25 lines of code changed in 3 files:
Made constitutive models consistent with changes made by Jim to VelGrad, DefGrad, and volume compuations.
0 lines of code changed in 14 files:
scheduleTimeAdvance()
- added scheduling of the particle scale factor.
4 lines of code changed in 1 file:
Add GPU kernel for DQMOM::constructLinearSystem. This can be further optimized by using partial products.
The idea here is based on the fact that the system of equations is normally constructed once per cell per time step, occuring on the order of several thousand times per time step. This type of high throughput computing is ideally parallelized on the GPU. In place of the normal cell-wise contruction, we can instead copy the necessary data to large arrays, indexed by cell number. We then construct all of the systems in a "single pass" on the GPU and push these results back to the host.
Also added MIT license and copyright header to DQMOM.h
296 lines of code changed in 5 files:
Fix for subtle logic error that was causing sus not to exit correctly
when an error occurred during UPS validation. The re-'throw'-ing of the
exception needs to occur on all processors, not just proc 0 where the
output is displayed.
Also, fixed catching of the case when a UPS_SPEC file is not found.
23 lines of code changed in 3 files:
output the total mass for MPM and ICE matls
10 lines of code changed in 1 file:
Split the Wasatch regression tests into meaningful and manageable chunks. Concatenate any combination to generate a more comprehensive test suite. This makes Wasatch among the first components to fully exploit Todd's changes to the RT.
112 lines of code changed in 1 file:
Carry forward value for pLocalized_new so that it is not allowed to be randomly
initialized.
7 lines of code changed in 1 file:
An "in development" version of cpdi that allows the particle domains to be
frozen at some compile time chosen factor, relative to their initial size.
This variable is lcrit and it must be set in multiple locations in the code
to have the same value.
0 lines of code changed in 1 file:
comment a few more tests that are not in the repository yet.
15 lines of code changed in 1 file:
remove some of the Warches tests that I inadvertently added in my previous commit.
4 lines of code changed in 1 file:
repair typo in the list of Warches tests in Warches.py
21 lines of code changed in 1 file:
Add support for PETSc v3.4.x (specifically minor version 4).
3 lines of code changed in 5 files:
Reinitialize the intrusions on restart for Arches and MPMArches.
3 lines of code changed in 1 file:
When generating gold standards the user can now select which subset of tests to run.
Throw an error if the subset of tests to run localRT or generate goldStandards
is not explicitly defined. Previously, the default was the nightly tests.
User can control which components to run via the env var:
TEST_COMPONENTS
and which subset of tests with the env var:
WHICH_TESTS
178 lines of code changed in 18 files:
add a method to wasatch to lock fields on the fly (needed by warches). fix a subtle logic bug when parsing the density expression that prevented warches from running properly.
7 lines of code changed in 2 files:
the following changes have been made in order to use RadPropertiesCalculation method to calculate abskg and abskp and to be used in DO and RMCRT radiation model: 1.to change the interface for computeRadProp(withparticles) and to add gas phase temperature as the passing variable. 2. to calculate the unscaled particle length and particle weights so that correct abskg/abskp canbe calculated. 3. to fix some bugs and to set abskg/abskp=1.0 for wall BC, in RMCRT model, abskg is actually wall emissivity. 4. to fix small bug for EnthalpyShaddix.cc. These changes may cause some goldenstandard cases changes. reported as: heptane_pipe, coal_table_pipe and rmcrt_bm1_1L
147 lines of code changed in 7 files:
Change environment variable.
1 lines of code changed in 1 file:
Update buildbot configuration file.
32 lines of code changed in 1 file:
The user can now specify a height. Below the height the powerLaw velocity profile is used and
above it U_infinity is used.
42 lines of code changed in 2 files:
added vol_frac_CC to the shear stress calculation.
4 lines of code changed in 1 file:
Fix typo in last commit
0 lines of code changed in 2 files:
Changed calculation of radiationVolq to match DO definition. This will change the answers and break the RT for all RMCRT tests.
2 lines of code changed in 1 file:
Add a conditional so that Ostwald Ripening does not give nan's in embedded geometry.
3 lines of code changed in 1 file:
Fix bug in Warches when parsing scalar transport equations.
1 lines of code changed in 1 file:
House cleaning
- moved localRT from scripts -> R_Tester/toplevel/localRT
27 lines of code changed in 3 files:
insert a missing "\"
1 lines of code changed in 1 file:
Trying out a fix for an arches linking error that comes up when turning radprops on.
3 lines of code changed in 1 file:
Added the ability for each component developer to add multiple
test subsets and select which subset to run via a dialog box. The subsets
can be comprised of any RT tests. For example, Arches now has
NIGHTLYTESTS = [....]
LOCALTESTS = [....]
DQMONTESTS = [....]
SCALARTESTS = [....]
In addition to defining the tests the user MUST MANUALLY update the list:
#LIST: LOCALTESTS DEBUGTESTS SCALARTESTS DQMONTESTS NIGHTLYTESTS
AND
in the getTestList() function.
270 lines of code changed in 13 files:
Turning these tests back on.
2 lines of code changed in 1 file:
- added initialization task to the Wall Shear Stress Models
- added sched_AddComputesRequires() so each model can schedule additional needed variables.
- Now using CCVariable<double> variable for the roughness instead of a double. This allows
for spatial variation. Still need to initialize it from either a file or geom_objects.
126 lines of code changed in 6 files:
Destroy varlabel for abskp.
1 lines of code changed in 1 file:
constructor
- fixed copy-paste error on z+ face
Turned off debugging output.
added comments and formatting.
** now gives consistent results on all faces ***
11 lines of code changed in 1 file:
added missing $
0 lines of code changed in 2 files:
octave script that plots the magnitude of the vector vs time for multiple dat files.
Good for comparing the total momentum of several udas.
usage:
plotVectorDatFile.m <dat1 dat2 dat3 dat4> <y/n for hardcopy>
93 lines of code changed in 1 file:
fix a bug that caused in 50679
13 lines of code changed in 1 file:
to fix some inconsitant specification for RMCRT and to remove bounding method for abskp and abskg
0 lines of code changed in 1 file:
added support for SFC(*)Variable<Vector>
15 lines of code changed in 1 file:
Fix a task dependency.
2 lines of code changed in 1 file:
Add interface function to access the Wasatch spec from the Wasatch class. Add a special case to the TaskInterface when Warches is used: STATE_N placeholder with an initialization taskgraph needs to use newdw.
41 lines of code changed in 3 files:
- The wallShearStressModel (WSS) is now applied to all ICE materials if the viscosity > 0;
- added hooks in viscousShearStress() task for calling the WSS model.
- instrumented code
** It's running but needs further testing **
119 lines of code changed in 10 files:
Some code went missing in IntrusionBC.cc...it has been put back. This might be the cause of the chronic Arches test failures (heptane, coal pipes.)
Fix a warning in HeatLoss.cc
26 lines of code changed in 2 files:
Give the user a hint at what might be wrong for strings.
1 lines of code changed in 1 file:
Turning off these two tests until the differences can be resolved.
2 lines of code changed in 1 file:
update radation property calculation model so that Lubima's code can be run in Arches
97 lines of code changed in 4 files:
save the turbulent viscosity to the DW for diagnostics.
8 lines of code changed in 1 file:
Changed names from NewtonianModel -> logLawModel
The main pieces are hacked in, except for spatially varying roughness.
*** This code has not been tested ***
263 lines of code changed in 8 files:
- Added a change to allow Uintah libraries that support CUDA to
successfully link.
Basically, we were receiving bizarre error messages (during library
linkage) like this:
.text.<internal compiler name> referenced in section ".text.<same
compiler name>[<same compiler name>]" of <object file in build
directory CCA/Components/Wasatch/Expressions/>: defined in discarded
section ".text.<same compiler name>[<same compiler name>]" of <same
object file>
The best we can determine is that this is caused when two separate .cc
files are compiled with a shared (common) template header file and
thus both instantiate it. The compiler throws away all but one of
these. (This issue apparently changed from a warning message to a
fatal linking error around binutils 2.16.) This issue apparently only
occurs if the .cc files were compiled using different compilation
arguments. The only two workarounds known are 1) to make sure that
everything is compiled exactly the same [unfortunately we can't use
this approach because some of the files are compiled by nvcc while the
rest are compiled by gcc], or 2) reorder the linking of the .o files.
We have chosen option 2 and thus moved the list of all .o files
produced from .cu files to the beginning of the link list. This seems
to have fixed the problem and hopefully will not cause more
complications down the road.
M CCA/Components/Wasatch/sub.mk
M CCA/Components/Wasatch/Expressions/sub.mk
- Alphabetize, line up, remove tabs.
M build_scripts/so_epilogue.mk
- Moved the list of .cu produced .o files to the beginning of the OBJS list.
42 lines of code changed in 4 files:
updated arenisca for optional constitutive model inputs
26 lines of code changed in 2 files:
added deformation gradient averaging
53 lines of code changed in 3 files:
Adding access to temperature in RMCRT property calculation.
11 lines of code changed in 2 files:
Silence warning by adding a newline at the end of the file.
1 lines of code changed in 1 file:
Adding a function for Warches to use RHS update from Wasatch
construction of the RHS.
82 lines of code changed in 3 files:
Adding particles to RMCRT.
131 lines of code changed in 3 files:
- removed debugging output
- cleaned up formatting on DebugStreams output
- taking into account possibility of a negative grid origin.
48 lines of code changed in 4 files:
Making wall model constant = 0.
Updating heptane_pipe to get RT passing.
1 lines of code changed in 1 file:
Deduce the sensible and adiabatic enthalpy names from the heat loss
model if the user hasn't saved them to the UDA.
23 lines of code changed in 2 files:
change coal model back to 50650
2 lines of code changed in 3 files:
First example at using Nebo DSL in Arches. See: BowmanNOx.cc. Also revert the BC reflection that was introduced to Warches recently to make the Warches DSL comparable to Arches.
144 lines of code changed in 4 files:
Set a few more things as gpu_runnable.
0 lines of code changed in 4 files:
update Enthalpy Shaddix.cc so that the arbitrary label can be read
38 lines of code changed in 3 files:
update char oxidation coal models
8 lines of code changed in 2 files:
cleanup wasatch::problemsetup.
29 lines of code changed in 1 file:
Fix mismatched CrowdMonitor write lock/unlock pair in OnDemandDataWarehouse.
1 lines of code changed in 1 file:
Adding a function to compute momentum (rho*U).
132 lines of code changed in 3 files:
Added equations to the inner loops. Cleaned out dead wood from copy & paste.
Next: verification on single and multi-patches.
Usage:
<Grid>
<BoundaryConditions>
<inletVelocity type="powerLawProfile">
<verticalDirection> 1 </verticalDirection>
<exponent> 1 </exponent>
</inletVelocity>
OR
<inletVelocity type="logWindProfile">
<verticalDirection> 1 </verticalDirection>
<roughness> 0.01 </roughness>
</inletVelocity>
< snip >
<Face side = "x-">
<BCType id = "1" label = "Velocity" var = "powerLawProfile">
<value> [1.,0,1.] </value>
</BCType>
OR
<BCType id = "1" label = "Velocity" var = "logWindProfile">
<value> [1.,0,1.] </value>
</BCType>
</Face>
snip
</Boundary Conditions>
147 lines of code changed in 5 files:
Fixed a multi-patch bug that resulted from using exclusive logic when we needed inclusive logic.
204 lines of code changed in 2 files:
zero out the velocity estimator. This makes the variable density tests behave better.
2 lines of code changed in 1 file:
Add regression tests for RK2SSP. NOTE: my previous commit will break four Wasatch tests due to a (beneficial) reordering in arithmetic operations. The differences are O(machine_precision).
28 lines of code changed in 1 file:
Generalize the time integration procedure to handle arbitrary Runge-Kutta, two-level-storage schemes (phi_old, and phi_new, e.g. Heun, RK3, RK3SSP...). Add the RK2SSP scheme as an example.
93 lines of code changed in 4 files:
Missing index++.
3 lines of code changed in 1 file:
start segregating the construction of the momentum partial RHS. These will be used when called from WARCHES. Simplify the interface for the Dilatation expression.
237 lines of code changed in 8 files:
BC cells are being touched, but the equations still need to be filled in.
125 lines of code changed in 6 files:
Update linker line for Wasatch.
1 lines of code changed in 1 file:
Fixing a couple issue with the coal variable transformation.
Force a table lookup for adiabatic enthalpy when inerts are used.
30 lines of code changed in 2 files:
More bug fixes related to heat loss:
1) Order HL last for cases where hl = f( other property models)
2) Don't compute heat loss in walls.
94 lines of code changed in 4 files:
3D driven cavity problem with comparison script -Greg Torkelson
446 lines of code changed in 1 file:
1. Allow Wasatch reduction variables to work with multistage integration. This required changing the reduction variable name at each stage so that Uintah can properly work with those. Note, however, that reduction per integrator stage is not recommended, hence the default behaviour is to reduce at the first integration stage only. This will be more than sufficient for all our reduction variable uses.
2. Add total kinetic energy decay calculations to the turbulence verification cases. These include ups input files, matlab, and shell scripts.
116 lines of code changed in 7 files:
fix for the RT.
1 lines of code changed in 1 file:
mvcc apparently doesn't like local paths with "" specification - use a full relative path.
1 lines of code changed in 1 file:
Update a sub.mk file to get some CUDA stuff fixed.
3 lines of code changed in 1 file:
fixed compile issue
3 lines of code changed in 1 file:
Initial commit for power law and log law inlet velocity profiles. This is just a
skeleton and needs to be filled in.
384 lines of code changed in 6 files: