January 2014 Commit Log

Number of Commits:
86
Number of Active Developers:
7
tsaad 2014-01-31 12:14 Rev.: 51521

add another 2D variable density MMS along with its boundary conditions. there is still an issue with one of the boundary conditions that I am working on figuring out.

1745 lines of code changed in 11 files:

jsutherland 2014-01-30 21:19 Rev.: 51520

This commit is on Abhishek's behalf.

Enable GPU for variable updates in Wasatch's time stepper.

116 lines of code changed in 3 files:

harman 2014-01-30 17:29 Rev.: 51519

Updated warning message and added a comment on how to add a hack that would
allow the user to process an uda that contains timesteps
that have a different number of variables per timestep.

29 lines of code changed in 1 file:

qymeng 2014-01-30 11:12 Rev.: 51518

fixed a mismatch of task computes and dw put.
task computes a reduction var into null(global) level, however puts it into level 0

1 lines of code changed in 1 file:

jas 2014-01-29 15:28 Rev.: 51517

Add std::cout for compilation on titan.

0 lines of code changed in 6 files:

harman 2014-01-29 13:11 Rev.: 51516

More housekeeping
renamed directory PatchComber -> ReduceUda

627 lines of code changed in 19 files:

harman 2014-01-29 12:46 Rev.: 51515

Housekeeping:
-Removed PatchCombiner code and cleaned out calls to it.

1 lines of code changed in 5 files:

harman 2014-01-29 11:20 Rev.: 51514

DataArchiver.cc
- reduceUdaSetup()
Just use a plain copy for coping the checkpoints.
Now copying the original ups and input.xml.org files

-beginOutputTimestep()
fixed compiler warning

- copy_outputProblemSpec()
cleaned up output statement

AMRSimulationController.cc
- run() inside reduceUda conditional block
setting the intital physical time from reduceUda

Dir.cc
- getFilenamesBySuffix()
comment out cout statement

UdaReducer.cc
- polished, added comments, cleaned up user output

- added user message

- initialize()
set delt = 0. You must have this if outputInitTimestep was used.

- computeLabels()
removed code that will never be used.

- deleted finalizeTimestep related code. It's not needed

- needRecompile()
recompile the taskgraph every timestep. The saved variable list
and the number of matls could change from timestep to timestep.

reduce_uda is in final testing phase. All initial tests show that it's working.

89 lines of code changed in 7 files:

harman 2014-01-28 13:18 Rev.: 51513

removed stray cout.

0 lines of code changed in 1 file:

harman 2014-01-28 12:48 Rev.: 51512

copySection()
- Slight generalization. Removed the hardcoded file name and use the name passed in.

copy_outputProblemSpec()
- added a method that copies the component's specifications to a new timestep.xml.

writeto_xml_files()
- Added call to copy_outputProblemSpec() if using reduceUda.

74 lines of code changed in 2 files:

ahumphrey 2014-01-27 20:33 Rev.: 51511

Temporarily disable cudaHostRegister() calls and associated machinery.

47 lines of code changed in 1 file:

harman 2014-01-27 16:36 Rev.: 51510

Stepping gently.

UdaReducer.cc
- updated debugstream messages.
- no longer setting the current top level timestep in the initialization task.

DataArchiever.cc/h
- created a map that is used only by reduceUda that holds what the directory timesteps are
and is keyed by the simulation timestep.

- added getTimestepTopLevel()
This returns the simulation timestep or if reduceUda is used a directory timestep.

- reduceUdaSetup()
created the global map that holds the directory timesteps

-makeTimestepDirs() & writeto_xml_files() & outputVariables()
use getTimestepTopLevel() for the timestep stamp.

81 lines of code changed in 3 files:

ahumphrey 2014-01-26 13:37 Rev.: 51509

Unregister all registered, page-locked host memory only after all tasks have completed (at the end of execute). This should fix the issue with CUDA reporting "part or all of the requested memory range is already mapped".

6 lines of code changed in 1 file:

ahumphrey 2014-01-25 22:13 Rev.: 51508

Fix error in device/host offset and size check (Unified Scheduler).

1 lines of code changed in 1 file:

jsutherland 2014-01-25 13:04 Rev.: 51507

Enable CUDA for another expression.

7 lines of code changed in 2 files:

jsutherland 2014-01-25 07:45 Rev.: 51506

Remove a superfluous ;

1 lines of code changed in 1 file:

jsutherland 2014-01-24 13:47 Rev.: 51505

Only include fields related to variable density in the pressure source term if we have a variable density calculation.

20 lines of code changed in 2 files:

jsutherland 2014-01-24 12:59 Rev.: 51504

Back out changes that converted the time step to a SingleValueField. It turns out that this wasn’t mandatory after all, since it only showed up on the RHS (not LHS) of Nebo statements.

5 lines of code changed in 1 file:

ahumphrey 2014-01-24 12:44 Rev.: 51503

Make the ARCHES 'Initializing all scalar equations and sources' output coherent when running with threads under the Unified Scheduler.

16 lines of code changed in 3 files:

ahumphrey 2014-01-24 12:20 Rev.: 51502

Make the ARCHES 'domain boundary condition summary' coherent when running with threads under the Unified Scheduler.

11 lines of code changed in 1 file:

harman 2014-01-24 11:51 Rev.: 51501

Overall:
Renamed functions and tasks to be more descriptive
Split big functions and tasks into smaller chunks of single purpose code where appropriate
Cleaned up the SCI_DEBUG output so you can now follow what is going on.
Note none of the logic was touched.

Specifics:

executedTimestep -> findNext_OutputCheckPoint_Timestep
-> writeto_xml_files


outputTimestep() -> makeTimestepDirs()

task:outputReduction -> outputReductionVars()

task:output -> task:outputVariables()

finalizeTimestep() -> finalizeTimestep + sched_allOutputTasks


203 lines of code changed in 4 files:

jsutherland 2014-01-24 10:01 Rev.: 51500

silence another warning.

1 lines of code changed in 1 file:

jsutherland 2014-01-24 10:01 Rev.: 51499

Missed a few “typename” decorators that the mac let me get away with, but broke linux.

1 lines of code changed in 1 file:

tsaad 2014-01-24 09:42 Rev.: 51498

add a missing typename to James' recent commit

1 lines of code changed in 1 file:

jsutherland 2014-01-24 09:11 Rev.: 51497

Convert the time step to a SingleValueField to make it play nice in nebo in preparation for moving the do_update method over to GPU.

125 lines of code changed in 2 files:

jsutherland 2014-01-24 09:11 Rev.: 51496

Clean up some compiler warnings & formatting.

146 lines of code changed in 2 files:

tsaad 2014-01-23 15:17 Rev.: 51495

add regression tests for the linear and parabolic BCs

4 lines of code changed in 1 file:

tsaad 2014-01-23 15:06 Rev.: 51493

remove an unused class

0 lines of code changed in 2 files:

tsaad 2014-01-23 14:42 Rev.: 51492

set staggered Dirichlet BCs on extra cells. This should resolve the confusion that came up when looking at the linear, parabolic, and powerlaw BCs. Also add a density expression to the initialization graph when the density is constant. This should resolve issues when setting velocity boundary conditions on constant density problems.

21 lines of code changed in 3 files:

jsutherland 2014-01-22 17:06 Rev.: 51490

Modify scalability test source term to be fully nebo-compliant so that it can go GPU.

34 lines of code changed in 1 file:

jsutherland 2014-01-22 16:39 Rev.: 51489

Fix a memory leak I just introduced. Apparently c++ doesn’t allow you to delete multiple pointers in the same statement.

1 lines of code changed in 1 file:

jsutherland 2014-01-22 16:14 Rev.: 51488

Mostly whitespace changes.
Move a few invariant variables outside of a loop.

21 lines of code changed in 2 files:

harman 2014-01-22 12:54 Rev.: 51487

Overall
Cleaning out combined_patches, reuse combinePatchSetup() & more heavy construction on reduceUda, details below.

sus.cc
- removed combine_patch option and related code.
- renamed doCombinePatches() -> setReduceUdaFlags()

Output.h
- alphabetize includes
- renamed doCombinePatches() -> setReduceUdaFlags()

SimulationController.h
- renamed doCombinePatches() -> setReduceUdaFlags()
- removed d_combinedPatches variable

SimulationController.cc
- renamed doCombinePatches() -> setReduceUdaFlags()
- remove d_combinedPatches variable
- cleaned up schizophrenic indentation
- using proc0cout instead of cout with conditional wrappers
- pulled reduceUda code from preGridSetup()

AMRSimulationController.cc
- removed d_combinePatches variable
- added d_timeinfo code from SimulationController to reduceUda conditional block


UdaReducer.cc/.h
- removed piles of code that failed for various reasons.

DataArchiver.h
- renamed doCombinePatches() -> setReduceUdaFlags()

DataArchiver.cc
-- renamed doCombinePatches() -> setReduceUdaFlags()
- cleaned up debugstream output so you can follow it.
- minor reformatting
- reduceUdaSetup()
- the entire checkpoints directory is now being copied to the new uda.


294 lines of code changed in 9 files:

ahumphrey 2014-01-22 09:53 Rev.: 51486

Add actual documentaion on GPU-related SCI_DEBUG vars.
Delete old GPU scheduler code. Keeping this around is only confusing to any new onlooker.

15 lines of code changed in 3 files:

harman 2014-01-22 08:35 Rev.: 51485

added notes to ExecTimes:+

1 lines of code changed in 1 file:

ahumphrey 2014-01-21 17:40 Rev.: 51484

Check for lib64 directory under zlib installation, accounts for /lib64, /usr/lib64, etc. Generate new configure script.

42 lines of code changed in 2 files:

qymeng 2014-01-21 15:30 Rev.: 51483

fix memory leak of postD2H and postH2D copies.

4 lines of code changed in 1 file:

jsutherland 2014-01-21 12:25 Rev.: 51482

Silence a compiler warning.

2 lines of code changed in 1 file:

jsutherland 2014-01-21 07:57 Rev.: 51481

Fix the failing wasatch test. I think that this was inadvertently broken last week when Tony made a seemingly innocuous commit.

14 lines of code changed in 1 file:

harman 2014-01-20 17:42 Rev.: 51480

Heavy construction zone with piles of scraps code laying around that may get reused. It
works for a simple 2 level, 2matl problem, no particles.


252 lines of code changed in 2 files:

ahumphrey 2014-01-18 22:14 Rev.: 51479

Correctly set PIC flag for GCC and XL compilers on BG/Q. -fPIC for GCC and -qpic for XL. Generate new configure script.

8 lines of code changed in 2 files:

ahumphrey 2014-01-17 14:35 Rev.: 51478

Add build system support for BG/Q XL compilers. Generate new configure script.

517 lines of code changed in 2 files:

tsaad 2014-01-17 13:36 Rev.: 51477

no need for the Builder class in the StrainTensorBase expression.

1 lines of code changed in 2 files:

tsaad 2014-01-17 13:26 Rev.: 51476

Fix logical bug when parsing the scalability test. Thanks to Abhishek for catching this.

11 lines of code changed in 1 file:

harman 2014-01-17 12:03 Rev.: 51475

Long overdue changes.
added qscale option to ffmpeg command. This fixed the poor movie quality.
- only resize images if needed
- only convert image formats if needed
- no longer ask the user to size images by a factor of 2.

For best results use jpegs and mpeg.

0 lines of code changed in 2 files:

tsaad 2014-01-17 10:50 Rev.: 51473

Rename the 1D varden MMS expressions to be more descriptive and concise. This will break the RT due to a change in one of the variable names that is being saved.

117 lines of code changed in 8 files:

tsaad 2014-01-17 10:07 Rev.: 51472

Use uniform naming for MMS source terms

28 lines of code changed in 3 files:

tsaad 2014-01-17 09:19 Rev.: 51471

Major simplification of the varden oscillating MMS spec. The entire MMS can now be specified within a single, concise block.

81 lines of code changed in 4 files:

tsaad 2014-01-16 22:02 Rev.: 51470

Add Taylor-Green-Vortex-like variable density MMS. Thanks to Amir for creating the original MMS expressions and plugging them into Wasatch. This commit is based on his initial implementation. The changes that I introduced include a correction to the mixture fraction source term, concise class names, and expansion and simplification of the input specification for the MMS. Additional work will follow by reducing the amount of input that the user has to specify. For a reference to the MMS committed here see: Shunn, L., Ham, F., & Moin, P. (2012). Verification of variable-density flow solvers using manufactured solutions. Journal of Computational Physics, 231(9), 3801–3827. doi:10.1016/j.jcp.2012.01.027.

1919 lines of code changed in 13 files:

harman 2014-01-16 14:46 Rev.: 51469

Use a "smart" absolute path to makeCombinedIndex.sh

4 lines of code changed in 1 file:

tsaad 2014-01-16 13:59 Rev.: 51468

simplify the TwoStreamMixing spec

2 lines of code changed in 1 file:

jas 2014-01-16 00:47 Rev.: 51467

Updated copyright information.

2483 lines of code changed in 3710 files:

jsutherland 2014-01-15 19:21 Rev.: 51466

Rework the diffusive flux and diffusive velocity expressions.

- Interpolate the product of density and diffusion coefficient
rather than separately interpolating them.

- Reduce the number of types provided and use more type inference

NOTE: this will diff a few tests due to changes in roundoff error associated with the interpolation change.

220 lines of code changed in 7 files:

tsaad 2014-01-15 10:32 Rev.: 51465

Initialize BoundFlux, VRFlux, and radiationVolq in an attempt to resolve the failing RMCRT test in Wasatch. Thanks Todd.

19 lines of code changed in 1 file:

harman 2014-01-14 17:20 Rev.: 51464

is_BC_specified() & BC_bulletproofing()
fixed flawed logic when circles/rectangles/annulus were
used in conjunction with a side.

69 lines of code changed in 2 files:

tsaad 2014-01-14 15:53 Rev.: 51463

put back the RMCRT Wasatch test to see if Todd's changes work for us.

1 lines of code changed in 1 file:

tsaad 2014-01-14 15:53 Rev.: 51462

remove the RMCRT test from Wasatch until Todd makes a change to the carryForward functionality in the Ray class.

0 lines of code changed in 1 file:

harman 2014-01-14 13:02 Rev.: 51460

Removed dead code from previous commit


0 lines of code changed in 2 files:

harman 2014-01-14 12:56 Rev.: 51459

Eliminated the carryforward task. That code has been moved to the rayTrace task

Added calc_frequency tag to ups files.

RMCRT_test_1L.ups & RMCRT_bm1_DO.ups:
- set calc_frequency to 2.

This should not change the answers.

23 lines of code changed in 5 files:

jas 2014-01-13 23:40 Rev.: 51458

Add boost library

3 lines of code changed in 1 file:

tsaad 2014-01-13 11:05 Rev.: 51457

Simplify the implementation of the Burns-Christon trilinear absorption coefficient. Add it to the regression tester. This will break the buildbot.

4 lines of code changed in 2 files:

harman 2014-01-10 12:58 Rev.: 51456

Removed
void sched_initProperties
void initProperties

3 lines of code changed in 1 file:

harman 2014-01-10 12:48 Rev.: 51455

More fallout from previous commit.
Removed _benchmark from RayGPU.
comment out code related to benchmarks 4 & 5.

2 lines of code changed in 3 files:

harman 2014-01-10 11:21 Rev.: 51454

Reorganization:
- moved the initialization of abskg from ray.cc -> RMCRT_Test. This is consistent
with how Arches & Wasatch interact with the ray class.

- updated the input files to reflect the changes

Removed _abskgBench4, this is redundant with _abskg.

Commented out code related to benchmark 4 & 5. It's not clear
which verification test these refer to and they're creating heartburn.

** This should not change the answers.


216 lines of code changed in 4 files:

tsaad 2014-01-10 10:45 Rev.: 51453

Simplify the world by improving the RMCRT benchmark interface in Wasatch.

26 lines of code changed in 3 files:

tsaad 2014-01-09 21:35 Rev.: 51452

Create an expression that computes the Burns-Christon abskg. This will allow us to depend less on the Ray class.

154 lines of code changed in 4 files:

tsaad 2014-01-09 12:57 Rev.: 51451

Switch back to constant expression for the RMCRT benchmark cases. When using the RMCRT benchmarks, the Ray class is responsible for computing the absorption coefficient. If the radiation calculation frequency is > 1, and in the absence of a carry_forward for the absorption coefficient, that field will be garbage and will therefore break the RT. This commit will fix this by setting the absorption coefficient to a constant value in the interim time steps. Alternatively, one could modify the Ray class to compute the absorption coefficient at EVERY time step regardless of the radiation calculation frequency. I will discuss this with Todd.

1 lines of code changed in 1 file:

tsaad 2014-01-09 12:13 Rev.: 51450

fix bug in RadiationSource that showed up on Linux.

2 lines of code changed in 1 file:

tsaad 2014-01-09 07:22 Rev.: 51449

remove the RMCRT test from the nightly RT due to a newly discovered bug.

0 lines of code changed in 1 file:

jas 2014-01-09 00:17 Rev.: 51448

Conflicts:
src/CCA/Components/Arches/ChemMix/ClassicTableInterface.cc
src/CCA/Components/Arches/ChemMix/ClassicTableInterface.h

439 lines of code changed in 4 files:

harman 2014-01-08 13:23 Rev.: 51443

outputTimestep()
removed commented out variables in the parameter list

- removed 80 columns wraps
- added vertical horizontal spaces
- add {}

67 lines of code changed in 2 files:

harman 2014-01-08 10:26 Rev.: 51442

No longer naming local saveItem class variables label_ or matlSet_

Now consistently using d_varname for global variables.

Cosmetics:
- removed some 80 column wraps
- added vertical and horizontal spaces
- added dividers between methods

87 lines of code changed in 2 files:

tsaad 2014-01-08 10:23 Rev.: 51441

Add Burns-Christon benchmark regression test to Wasatch

16 lines of code changed in 3 files:

tsaad 2014-01-08 10:02 Rev.: 51440

Add RMCRT interface to Wasatch. Create a RadiationSource expression to wrap divQ (computed by Ray). Add other expressions to provide a smooth integration of RMCRT into Wasatch. Additional work needs to be done on computing CellType with intrusions and storing the boundary fluxes.

897 lines of code changed in 18 files:

harman 2014-01-07 13:35 Rev.: 51438

Allow the component to set if it's using the lockStepAMR time integration scheme.

1 lines of code changed in 1 file:

harman 2014-01-07 13:33 Rev.: 51437

Set scrubbing to none if the load balancer is using getNthProc input
file option. Previously it was also set for reduceUda and it didn't work with
multi-level grids.


2 lines of code changed in 1 file:

harman 2014-01-06 17:13 Rev.: 51433

Pulled another weed, thanks Alan.
2 space indentation.

49 lines of code changed in 1 file:

harman 2014-01-06 16:45 Rev.: 51432

removed extra
lb = scinew ICELabel()
<Thanks Alan>

minor formatting

17 lines of code changed in 1 file:

qymeng 2014-01-06 13:52 Rev.: 51430

Loop through only neigbours instead of all processors when assign resources/nodes to ONCE_PER_PROC task.
This change speeded up the assign resource method by 6900X on Mira with 256K cores.

6 lines of code changed in 1 file:

harman 2014-01-06 11:32 Rev.: 51428

Cleanup just before the complete overhaul.

177 lines of code changed in 2 files:

qymeng 2014-01-05 23:18 Rev.: 51426

When more than one tasks compute a same reduction variable into same level,
only schedule one reduction task after the very last task that computes this reduction variable.

28 lines of code changed in 3 files:

qymeng 2014-01-05 16:51 Rev.: 51425

Turn off topologicalSort() of task graph compiling.
Move reduction task generation part of topo sort to ScheduerCommon.
This greatly speeded up the taskgraph compling on very large scale.

43 lines of code changed in 2 files:

ahumphrey 2014-01-02 20:45 Rev.: 51424

Report particle creation stats via std::cout

3 lines of code changed in 3 files:

harman 2014-01-02 17:01 Rev.: 51423

Each plane can now set when to start and stop writing out data.

Writing out the physical time to a file. The timestep number and physical time are different
by one timestep.

37 lines of code changed in 2 files:

dav 2014-01-02 15:57 Rev.: 51422

Remove Allen's 'fix' of my 'fix', neither of which worked... This should keep the buildbot from failing and we will figure out a real fix soon.

0 lines of code changed in 1 file:

jas 2014-01-02 15:42 Rev.: 51421

Squashed commit of the following:

commit 140a531e374ae16718f57324fdcc8c1c1c070de1
Author: John Schmidt <john.schmidt@utah.edu>
Date: Thu Jan 2 15:32:48 2014 -0700

Implement recursive copying of a directory.

commit e1093af35e37c88e63869ed1683a45be407f3f27
Author: John Schmidt <john.schmidt@utah.edu>
Date: Thu Dec 12 18:43:16 2013 -0700

Remove preprocessor directors for filesystem version 2 & 3 that were
found around the filesystem::copy().

commit 56f2c3f9944fdc9fe60bdbad336bd79c96207673
Author: John Schmidt <john.schmidt@utah.edu>
Date: Sun Nov 24 19:02:14 2013 -0700

Add boost filesystem to the configure check and the Dir_boost.cc to the sub.mk.

commit 0a5a9119fc3501ea8ce56112812296bf57fbe0b4
Author: John Schmidt <john.schmidt@utah.edu>
Date: Sun Nov 24 19:01:17 2013 -0700

Add the boost filesystem calls for doing directory manipulations.

341 lines of code changed in 4 files:

December 2013 »

Generated by StatSVN 0.7.0