Add RadProps library to the static build for sus.
0 lines of code changed in 2 files:
Allow users to specify moving geometry from input file. A few other cosmetic changes.
0 lines of code changed in 16 files:
sched_initialize()
- added bulletproofing. The number of levels must be > 1 if your using the
data onion algorithm.
0 lines of code changed in 2 files:
Remove timing calls that prevent the complete timing data for the solve to
print out.
0 lines of code changed in 2 files:
Add embedded geometry support to the momentum equations and the pressure solve. We currently handle wall boundary conditions on the embedded geometries. Regression tests will follow in the next commit.
389 lines of code changed in 24 files:
Fix bug when turning on hypre timing and specifying the setupFrequency to be 0.
2 lines of code changed in 1 file:
Add core framework for embedded geometries to Wasatch. In this commit, provide the basic expression wrapper to the Uintah GeometryPiece objects. Also provide appropriate mechanisms to create area fractions. Created an easy to use input specification for embedded geometries. Also provide support for creating embedded geometries through Wasatch expressions (see OscillatingCylinder in this commit). The next commit will add support of embedded geometries to the momentum equations and the pressure solve.
827 lines of code changed in 10 files:
Made adjustments to methane_RCCE.ups so the same number of output timesteps exist
on resarts when using sus -restart -t 0 -copy.
Turned off methane_RCCE. There are some numerical fuzz entering into the simulation on
restarts.
0 lines of code changed in 2 files:
Users are no longer required to provide an initial condition for the pressure from the input file. The initial condition is still needed though to make the solver work properly by using the previous pressure solution as an initial guess - therefore the initial condition on the pressure is now automatically setup when processing the momentum equations. The value is set at 0. Remove all pressure initialization from input files. Users also still have the flexibility of providing their own initial pressure via the input file.
11 lines of code changed in 3 files:
More information on thread execution counts.
6 lines of code changed in 1 file:
Report thread count information directly from the threadpools rather than the parsed counts.
2 lines of code changed in 1 file:
Adding a wall HT model that only executes on geometry within a specific
geometric region.
0 lines of code changed in 6 files:
Updates to the wall heat tranfer model framework, including putting it into
production mode.
Removing the <const_wall_temperature> spec from the new DORadiation
model. This was confusing and actually clashed with the
SolidWallTemperature spec in the <BoundaryConditions> section of the
input file. It is now on the user to specify BC's for the
SolidWallTemperature values if they want something other than the
resultant temperature from the indep. var space.
Updating some input files to be consistent with the new spec.
Local RT is reporting that methane_jet.ups will fail, but the
differences are in the noise.
115 lines of code changed in 14 files:
Adding the -pad option. Previously, for a multi-level situation, cell indices
that lay outside of the existing fine level were simply not populated with
data. This is a perfectly reasonable thing to do. However, to enable easier
post-processing, the -pad option fills those cells with zero values for
the variable being extracted. Default behavior is unchanged.
Note: There is a big opportunity for optimization here. Currently, for each
index for which data is wanted, every patch on the level is queried *even
after the correct patch is found*. A factor of 2, on average, can be expected
by just quitting the search after the patch is found. A bigger win would
come from trying that same patch that the last index was found in first.
Could be significant for petascale calculations with 10^5 patches.
0 lines of code changed in 2 files:
Changed name of debugstream from RAY-GPU -> RAY. It's just easier this way.....
0 lines of code changed in 2 files:
minor fixes and added some user feedback
0 lines of code changed in 2 files:
Adding a new test.
4 lines of code changed in 1 file:
Cleaning up the westbrook/dryer model.
0 lines of code changed in 6 files:
limit the output.
0 lines of code changed in 2 files:
Change version number from 1.4 to 1.5.
0 lines of code changed in 4 files:
Regression test should work now for Induction simulations
0 lines of code changed in 2 files:
Getting the Arches/RMCRT input file to work, which includes interacing to radProps.
Turning back on the constantMMS input file for local tests.
1 lines of code changed in 1 file:
Merging two lines.
0 lines of code changed in 6 files:
Git problems...recommitting code that was mistakenly removed.
6 lines of code changed in 1 file:
Revert "Fixing some faulty logic."
This reverts commit c4499c3016c5ef09e50c869f13d7b6f75bc46a01.
4 lines of code changed in 1 file:
Fix some faulty logic.
6 lines of code changed in 1 file:
Schedule the intrusion information to be printed. This allows for inlet
areas to be printed.
53 lines of code changed in 3 files:
Comments.
0 lines of code changed in 2 files:
Squash another warning.
Stop saving reduction variables in several UPS files.
14 lines of code changed in 1 file:
Initial commit for using radProps + RMCRT + Arches.
It's close to running.
3 lines of code changed in 3 files:
Added Locking tests demonstrating that shear locking is present in current MPM implementation. Also added some of the same Arenisca/general J2 plasticity benchmarks for Kayenta. -Dave A.
0 lines of code changed in 2 files:
All:
Now using < instead of <= in maxLevel loops
2 level approach
- Now setting the boundary conditions for abskg and sigmaT4 on the coarse levels
not just the fine level which is wrong.
This will change all ML results
18 lines of code changed in 3 files:
fix hardcoded lib path for udaReader (oops)
0 lines of code changed in 2 files:
Minor touchup and getting R_Tester ready for new tests. -Dave A.
65 lines of code changed in 1 file:
- Pulled the scinew Task() inside of if(on archesLevel) conditional statement to plug a memory leak.
3 lines of code changed in 1 file:
Regression tests for new induction code
6 lines of code changed in 1 file:
Squash a warning.
1 lines of code changed in 1 file:
Turns out that the reset on the Iterator didn't work. Trying another method for bulletproofing.
6 lines of code changed in 1 file:
Resetting the iterator out of paranoia.
1 lines of code changed in 1 file:
Removed an exception from ClassicTableInterface.
It appears that the patch is returning BC children that don't actually have
an iterator on the patch. This is causing problems for some of the coal runs.
Removed a debug print statement from CharOxidationShaddix.
Formatted and changed patch layout -- to a more realistic layout -- given the
current resolution. Also reformatted the file.
1 lines of code changed in 3 files:
formatting.
coarsenQ()
- set computeAve to true. This is a big change and it will
change all multi-level results.
16 lines of code changed in 1 file:
Added the "Data Onion" algorithm to Arches:RMCRT along with an input file.
It runs.
45 lines of code changed in 1 file:
Remove the Wasatch performance tests. We will have to figure out a more robust way to measure performance on the RT/buildbot.
0 lines of code changed in 2 files:
Added multi-level capabilities
constructor
defined the material set local to this model...yuck.
sched_initialize()
initialize cellType = 0 on non-arches levels This is a temporary hack
until we can initialize it in sched_cellTypeInit()
sched_computeSources()
-carryForward cellType on non-arches levels. All cellType data needs to be in the new_dw
- Only schedule on RK step 0 and on an archeslevel
96 lines of code changed in 3 files:
udaReader update: add stencil4/7 support, add particle id support, support for compiling in visit
149 lines of code changed in 7 files:
More work on ARCHES thread safety.
Working to get full boiler problem running with threaded scheduler.
49 lines of code changed in 5 files:
Each source terms is now responsible for initializing its own variables.
Motivation for this change is RMCRT needs to initialize vars on levels other
than the Arches level.
RMCRT: now using ArchesMaterials() as the matlSet
All RT tests should pass except rmcrt_1L.
312 lines of code changed in 64 files:
Moving a print statement so that mass flow rate actually has a real
value.
6 lines of code changed in 1 file:
Particles are now required for detonation and change was made to requirement to allow particles to be fully consumed
0 lines of code changed in 2 files:
Deleting the boundary condition object.
0 lines of code changed in 2 files:
Lock critical section in ClassicTableInterface. Another step toward ARCHES thread safety.
8 lines of code changed in 2 files:
Fix compiler error for non-CUDA builds.
0 lines of code changed in 2 files:
Merge changes in commit on branches/arches_alcc_dev r49696 to HEAD.
We now have GPU-RMCRT operational within Arches.
29 lines of code changed in 6 files:
increase the number of timesteps taken in the performance tests in an attempt to fix the failing wasatch perf tests. Hopefully this will work.
9 lines of code changed in 1 file:
Fix a bug when assigning the velocity component to the intrusion for
mass_flow.
Added some comments to the spec.
3 lines of code changed in 1 file:
Doxygen comments in PhysicalConstants.h.
Fixed a bug in intrusionbc where in the density wasnt being set
properly.
53 lines of code changed in 4 files:
forgot to remove performance tests from the local RT. Keep them on the nightly only.
0 lines of code changed in 1 file:
Add newline at end of file to remove compiler warning.
0 lines of code changed in 2 files:
John moved configure up to autoconf version 2.69, so specifying 2.69 as the required version. Also, on some computers PETSc needs to be tested under a C++ compiler.
17 lines of code changed in 2 files:
fix the wasatch performance tests. make them run longer to get proper stats. this will break the RT GS. will update tomorrow morning.
20 lines of code changed in 1 file:
Oops, committed one line of code that wasn't supposed to be there in my big
pile yesterday.
0 lines of code changed in 2 files:
Find the unified hypre libraries if the individual hypre libraries are
not found.
20 lines of code changed in 2 files:
There is no need to check the boundary condition type in the property calculation. Also, inerts were not being properly interpolated at boundaries when using the post_mix model. This will not change the answers for any RT test.
8 lines of code changed in 1 file:
Add various hypre libraries as required with Debian's hypre library
layout. This is needed for the static build for Debian Wheezy.
579 lines of code changed in 4 files:
Fix warning about redeclared __host__ __device__ function. Some UnifiedScheduler formatting.
3 lines of code changed in 3 files:
Remove the addMaterial and material conversion capability. Never has so much
code gotten so little use for so long.
6 lines of code changed in 151 files:
Bug fix. In the snippet of code that handles
the first step in a new level, 3 lines of code
got duplicated.
Now gives correct answer when all 3 levels are
the same.
This will change the answers, and will need a
new GS for the data onion test.
5 lines of code changed in 1 file:
Fix GPU-RMCRT problems on Kepler hardware.
This now runs on Kepler GPUs and is backward compatible with Fermi.
0 lines of code changed in 4 files:
Add better device debug support for debug builds using CUDA.
Generate new configure script.
21 lines of code changed in 4 files:
Now passing material id into new boundary condition interface rather
than the entire ArchesLabel set.
Applying BCs for heat loss.
28 lines of code changed in 13 files:
Put code back in from Rev 48281 that was removed
last spring. This code takes the coarsening ratios
between the levels of a data onion, and
handles the first step of ray tracing once a level
boundary has been breached.
This commit computes the coarsening ratios,
whereas the previous commit used hard-coded
values.
I thought this would change the answers to
RMCRT_bm1_DO, but somewhat surprisingly,
the results are unchanged. I will look into
why this is, but for now, based on my localRT,
the answers shouldn't change.
34 lines of code changed in 1 file:
Add nightly performance tests for Wasatch.
6 lines of code changed in 1 file:
added printScheduling() & printTasks().
39 lines of code changed in 7 files:
Arches CFD now runs on a single level in a multi-level grid.
delT must be computed on all levels.
*should not change the answers.
187 lines of code changed in 3 files:
- Passing bool do_AMR into ARCHES and MPMARCHES.
- Added stub functions for scheduleCoarsen scheduleRefineInterface.
- removed tabs.
53 lines of code changed in 8 files:
Fix the failing Wasatch test. This was due to a bug that I introduced last week when trying to activate the Vreman turbulence model for two-dimensional flows. The RT and buildbot may still fail today in case the old gold_standards for this test are still on those machines. If this is the case, I will update those tomorrow morning when the RT updates. \n
37 lines of code changed in 3 files:
remove the failing test from Wasatch until I figure out what is causing this.
0 lines of code changed in 1 file:
Fix a missing initialization call for the heat loss model.
2 lines of code changed in 1 file:
RMCRT_VR.ups
removed slice tag
SourceTermFactory
tabs and formatting
0 lines of code changed in 2 files:
Revert to CUDA runtime API from driver API. This gives easier programmability and better debug support.
<<< >>> kernel launch syntax now exists in an external function to avoid NVCC's (version 4 & 5) inability to handle GNU tr1/type_traits headers.
Generate new configure script after changes to build system to support the above.
470 lines of code changed in 31 files:
formatting
24 lines of code changed in 1 file:
Removed false comment statement.
Removed lines that never get executed.
0 lines of code changed in 1 file:
Cleaning up ABSKP model. Deleting the moment labels. Turning off property model for abskp for the coal_channel test. It appears that property models aren't being deleted correctly. Will look into this soon.
0 lines of code changed in 4 files:
Models.xml: added applyfilter tag, removed slices tag
Ray.cc: removed hardwiring of direction of ray in DataOnion
RMCRT_bm1_DO & Examples.py: added regression test
3 lines of code changed in 3 files:
Deleted zero order kinetics model.
4 lines of code changed in 3 files:
Add in additional libraries for ProblemSpec and libxml2 and zlib as
needed for the intel compiler on stampede.
0 lines of code changed in 20 files:
Add check for multi-threaded intel mpi library (mpi_mt).
446 lines of code changed in 2 files:
Moved filtering algorithm to its own task.
121 lines of code changed in 2 files:
Cleanup
0 lines of code changed in 1 file:
Turned wasatch back on. Radprops now compiles.
0 lines of code changed in 2 files:
Comment clean up.
Slight speed-up on has_a_boundary() by
minimizing re-assignment of adjacentCell.
37 lines of code changed in 1 file:
Turned off wasatch and --with-boost so other components developers
can continue.
4 lines of code changed in 1 file:
Removed boundFlux map. Cleaned up comments.
Added filtering capability for divQ.
79 lines of code changed in 2 files:
Set NGN=2 for any d_8or27 value other than 8.
0 lines of code changed in 2 files:
Update build_wasatch_3p.sh
0 lines of code changed in 2 files: