Only do the work of the SingleProcessorScheduler (within Unified Scheduler) if not using MPI or GPU, and also not using multiple threads. Previously, the specified number of threads were created, but task graph was executed serially... without using the created threads for task execution.
3 lines of code changed in 1 file:
Templated the DataOnion/AdaptiveMesh task and updateSum()
64 lines of code changed in 2 files:
Added prism and cyrus to the list of machines that have gpus.
0 lines of code changed in 4 files:
Added p.velocity as a <geom_object><file><var> option for pts files with spatially varying initial velocity
1 lines of code changed in 1 file:
Remove some headers, improve const correctness, and remove some BC operator registration now that we are doing this in SpatialOps.
14 lines of code changed in 6 files:
Remove a header file inclusion in preparation for a SpatialOps commit that will remove that header file.
4 lines of code changed in 2 files:
This fix correctly creates the <restarts> block and places the
<restart ...> into it. I tested this and restarts from old UDAs
continue to work. I think only restart_merger uses the <restarts>
block anyway. Also, this has the benefit of keeping the history
of restarts on an UDA, as before it only showed the very last one.
3 lines of code changed in 1 file:
Improve some error trapping in TabProps parsing.
0 lines of code changed in 2 files:
Started templating tasks/functions. This will allow for either double or float RMCRT
when complete. Hardwired for double.
* this won't change the answers.
90 lines of code changed in 6 files:
Removed redundant copies in simplest algorithm.
2 lines of code changed in 1 file:
Changed order of operations in Discrete Ordinance for efficiency. This should change DORadiation A-matrix in the 16 decimal place.
12 lines of code changed in 1 file:
rework portions of the varden algorithm:
1. Get rid of the rhof* business and use rhof (TimeAdvance expression) instead
2. Apply appropriate bc on rhof_rhs so that rhof_(n+1) reproduces the correct bc
41 lines of code changed in 4 files:
fix some namespace issues int Ray et al that were causing build failures on mac dbg builds
7 lines of code changed in 4 files:
momentumAnalysis.cc
- if a timestep restart has been requeste upstream don't do the analysis
MomentumAnalysis.m
- remove duplicate entries if they exist
0 lines of code changed in 4 files:
Fixed compile time warnings, as well as a compile time error.
6 lines of code changed in 4 files:
This is Stage 1 of a commit which adds reflections for DO-Radiation. The RT-Tests are expected to fail, due to a new VarLabel in the checkpoints. The uda files are expected to remain unchanged.
311 lines of code changed in 20 files:
RMCRT_Test: scheduleTimeAdvance()
- For 2 level approach changed looping limits for sched_Refine_Q()
Ray.cc: sched_Refine_Q()
- Now using ghosts:AroundCells 1 for requires of divQ from the coarse level. *** Key
Converted input file into a multipatch 2L problem with Refinement ratio = 4
Now using 8 cores for nightly RT tests.
5 lines of code changed in 3 files:
Fix a bug that could have left density unexposed to Uintah.
4 lines of code changed in 1 file:
Remove density predictor expression registration in a few places if we don't have momentum active.
34 lines of code changed in 2 files:
add varden 1D mms test using the Amagat-Leduc law
0 lines of code changed in 2 files:
Fix cuds compiler error due to inclusion path issue.
Clean up a warning.
3 lines of code changed in 4 files:
Fix for our goofy CUDA builds.
0 lines of code changed in 2 files:
Several new things:
- First rough cut at a DO implementation (not yet working - still needs BCs)
- Simple radiative emission model
- Input file (SimpleEmissionRadiation.ups) to drive a reaction-diffusion problem with radiative heat loss.
- misc. cleanup
1108 lines of code changed in 9 files:
- don't build predictor expressions when convection is not on.
- Minor formatting changes;
19 lines of code changed in 5 files:
# cat_OnTheFly:
# This script concatenates a series of "dat" files to a masterUda
#
# Usage:
# cat_OnTheFly < list of uda files >
#
# Notes:
# - Each dat file is sorted using the first column
# - It assumes that the directory "masterUda" has been created
80 lines of code changed in 1 file:
remove a cout statement
0 lines of code changed in 2 files:
Update the Re correlation for drag models. Replace the unneeded spherical coordinate calculation with only magnitude calculation
6 lines of code changed in 6 files:
Chaneg how cqmom sources are parsed into setup
40 lines of code changed in 5 files:
repair the Wasatch memory test failures.
19 lines of code changed in 1 file:
* In input files (<Scheduler> section), only accept Scheduler values that are in the UPS_Spec.
* Better error handling for 'command line/input file' scheduler parameters, specifically for the Unified Scheduler.
* Update environmentalFlags list
26 lines of code changed in 4 files:
M CCA/Components/Parent/ComponentFactory.cc
- Needed to #include Arches.h (if Arches is specified). Apparently it
was previously brought in indirectly through MPMArches. This fixes
builds where only the Arches component is specified to configure.
M Makefile.in
- Fix informational message on location of svn_info.h
0 lines of code changed in 4 files:
If MPM or Arches is turned off, don't include MPMArches.
4 lines of code changed in 1 file:
- Fixes for OP X Yosemite and a few minor other fixes. Tested on old OSX and 3 flavors of Linux including GPUs.
M configure.ac
M configure
- Fix excess spew from clang compiler during configure.
- Check for version 14 of OS x (Yosemite).
- Removed 'atlas' check under Darwin as it was actually a blas check.
- Put in checks for Blas and Lapack for old and new (yosemite) OS X.
M aclocal.m4
- Fix shell script output error when used in configure to show the results of `basename $1`.
- Clang compiler uses both stderr and stdout (sigh), and outputs in a different order then
other compilers (for --version). This commit handles this.
- Quote a few variables so that if they are blank, the shell doesn't give warning messages.
M include/sci_defs/blas_testdefs.h.in
- Fix to handle new/old OS X and other OSes in a cleaner manner.
M CCA/Components/Parent/sub.mk
- Arches now uses Boost (in Arches.h), so Parent needs to know to compile against boost.
M CCA/Components/Arches/Arches.h
- Move #if guards for multiple inclusion to the top of file so that the compiler has to do less work.
- Alphabetize #includes so I can see what is there.
- #include boost/shared_prt as it is used in this file.
M CCA/Components/Arches/Transport/URHS.h
- One of our compilers errored on the excessive use of 'typename', so trimmed back a little.
M CCA/Components/Arches/sub.mk
- Remove tabs, lined up, alphabetized, removed old crud.
M Core/Thread/ThreadLock.h
- Move #if guard for multiple inclusion to the top of file so that the compiler has to do less work.
- Fixed spelling in #if guard: Threadlock updated to ThreadLock (capital L).
- Lined things up a little.
M Core/DataArchive/DataArchive.h
- Cosmetic lineup of variables.
M Core/DataArchive/DataArchive.cc
- Cosmetic white space and coding standards update,
don't use "std::" if using "using namespace std".
1161 lines of code changed in 18 files:
Revert Jeremy's commit that changed the build system for Yosemite.
This caused build failures on several existing systems.
14 lines of code changed in 4 files:
* Move to shared pointers for the task factories.
* Get the momentum RHS building and updating. Still needs work on the SSP integration and finalize all the build steps.
* From a factory, allow groupings of tasks and provide a way to retrieve these groups.
* Fix a bug when retrieving the tasks: The wrong memory was being returned resulting in a bogus task and undefined behavior.
* Put the particle helper in the Task base class.
* For now: for the FE update, pass a boolean to avoid division by density^(t+dt) to the RHS upon update.
336 lines of code changed in 26 files:
** Dav **
Fixes to the build system to get Uintah building under Yosemite. Changed:
include/sci_defs/blas_testdefs.h.in
configure.ac
configure
** Dav **
Fixed a typo:
Core/Thread/ThreadLock.h
1385 lines of code changed in 4 files:
Better type inferences for the interface.
Adding a momentum RHS calculation. Still needs work.
595 lines of code changed in 27 files:
Use a counter to avoid scheduling tasks that dont do anything.
Also some minor cleanup.
66 lines of code changed in 5 files:
more minor cleanup
2 lines of code changed in 2 files:
Carry forward properties properly.
1 lines of code changed in 1 file:
More minor cleanup.
We really need to refactor the masks so that we can avoid repeated creation and so we can refactor all of these BC expressions to use straight nebo/masks rather than iterators in some cases.
23 lines of code changed in 1 file:
More minor cleanup
16 lines of code changed in 1 file:
Minor reformatting of misc. stuff.
14 lines of code changed in 6 files:
Fix bug with initialization of per-level maxGhostCells and maxLevelOffset.
6 lines of code changed in 1 file:
added blueprint for ICE_MM. It's outdated but is better than a blank piece of paper....maybe.
2584 lines of code changed in 9 files:
Fix PAPI support in trunk
14 lines of code changed in 3 files:
Removing a sleeping bug from the variable density 2D MMS imlpementation. Adding 3 new tests to the RT. This includes the variable density 2D MMS oscilating periodic tests in all three possible planes in the space (xy,xz,yz).
29 lines of code changed in 3 files:
Clean up Amir's commit:
- input files have .ups, not .xml extension
- don't use 16 procs for regression tests
- run 15 steps rather than 5
3 lines of code changed in 1 file:
Adding the variable density 3D annular jet case to the RT.
2 lines of code changed in 1 file:
RMCRT should be using the OLDDW values of abskg.
This was fixed by carrying forward the radiation properties from the
previous timestep rather than changing the DW in the RMCRT source term
due to a grid issue with the dataonion approach.
Also, fixed a missing call to commonProblemSetup for the radiation
properties. This was causing properties to have bogus values on the
first timestep that could lead to non-terminating rays.
This commit may change the answers.
6 lines of code changed in 2 files:
Remove the deformation gradient averaging code. It was never shown to work
and more bad code is not what Uintah-MPM needs.
0 lines of code changed in 4 files:
Code cleanup before another search for the pack-buffer meory leak
432 lines of code changed in 13 files:
cleanup some formatting
139 lines of code changed in 1 file:
kill any test that exceeds the maximum run time
debug: 30 minutes
optimized: 15 minutes
22 lines of code changed in 1 file:
Now producing a hard copy of the FFT plot
Fixed sign on viscous and pressure surface forces
12 lines of code changed in 1 file:
Rip out some unnecessary explicit template instantiation.
19 lines of code changed in 1 file:
More cosmetic/standards/white space/untabbifying/etc updates I missed in previous commit.
156 lines of code changed in 2 files:
Cosmetic/standards/white space/untabbifying/etc updates before digging into code for excess memory usage on restarts.
592 lines of code changed in 2 files:
Remove some unnecessary typedefs and a dead type inference struct.
0 lines of code changed in 1 file:
include viscous and pressure forces in the mean calculation.
19 lines of code changed in 1 file:
add regression test for particle-to-mesh interpolation
2 lines of code changed in 1 file:
add support for interpolating and postprocessing particle fields as Eulerian fields. This is driven through the input file. Tests will follow.
199 lines of code changed in 6 files:
Changed base directory name:
SCIRun -> Uintah
0 lines of code changed in 2 files: