April 2015 Commit Log

Number of Commits:
87
Number of Active Developers:
11
harman 2015-04-30 15:30 Rev.: 53421

Updated method names. Should have been in previous commit.


3 lines of code changed in 1 file:

harman 2015-04-30 13:43 Rev.: 53420

added
existsLevelDB() // checks for variable in levelDB
removeLevelDB() // flags variable for removal

clear()
- added loop over levelDB items and remove them.




89 lines of code changed in 2 files:

harman 2015-04-30 11:26 Rev.: 53419

Throw an exception if the user tries to use the -gpu option without the Unified scheduler.

7 lines of code changed in 1 file:

harman 2015-04-30 09:50 Rev.: 53418

postH2DCopies(), preallocateDeviceMemory(), postD2HCopies()

- light formatting,
- cleaned up the gpu_stats output so it's easier to follow.


89 lines of code changed in 1 file:

harman 2015-04-29 16:36 Rev.: 53417

added print() method, which dumps both varDB and levelDB.

36 lines of code changed in 2 files:

harman 2015-04-28 16:59 Rev.: 53416

Added level index and __LINE__ to DebugStream output.

12 lines of code changed in 2 files:

harman 2015-04-28 15:09 Rev.: 53415

Added unifiedMemory class and GPUString class which are used to pass the varLabel struct into
the kernel. The varLabel struct isn't plugged in yet.


133 lines of code changed in 3 files:

harman 2015-04-28 13:48 Rev.: 53414

DebugStream GPU related names are consistent with what's specified in environmentalFlags.txt

2 lines of code changed in 1 file:

harman 2015-04-27 17:27 Rev.: 53413

Output details on the GPU being used. On computers with multiple
GPUs this is useful. For example:

GPU Device 0: "GeForce GTX TITAN Black" with compute capability 3.5

8 lines of code changed in 1 file:

dav 2015-04-27 16:52 Rev.: 53412

Need to initialize parallel infrastructure (even if not running in parallel).

0 lines of code changed in 2 files:

harman 2015-04-27 16:36 Rev.: 53411

Fixed GPU bulletproofing to allow the 2Level RMCRT algorithm.

This would have been caught with GPU RT coverage.....

17 lines of code changed in 2 files:

harman 2015-04-27 14:16 Rev.: 53410

Ray.cc
- mimic the debugging output in the GPU kernel code for direct comparison
- simplified the jumping levels conditional statement.
- Don't jump levels at the edge of the computational domain.

RayGPU*
- mimic the debugging output in the CPU code for direct comparison
- simplified the jumping levels conditional statement.
- throw exception if number of levels > d_MAXLEVELS


141 lines of code changed in 4 files:

guilkey 2015-04-26 20:36 Rev.: 53409

Cleaned up some commented out trial balloons. Added some logic to keep
fine patches from disappearing, even if they don't have refined particles
on them.

37 lines of code changed in 1 file:

ahumphrey 2015-04-24 14:01 Rev.: 53406

Some minor refactoring and debug stream improvements while hunting down the multi-threaded Hypre bug.

13 lines of code changed in 6 files:

harman 2015-04-24 08:52 Rev.: 53405

Increased max variableDB and levelDB items. Needed for RMCRT DO 3Levels


2 lines of code changed in 1 file:

guilkey 2015-04-23 17:06 Rev.: 53404

Don't include extra cells when looking to see if a cell has a finer cell above
it.

1 lines of code changed in 1 file:

ahumphrey 2015-04-23 16:12 Rev.: 53403

Refactor mispelled method name.

4 lines of code changed in 2 files:

jthornoc 2015-04-23 14:34 Rev.: 53402

Swith order of execution. Final property models should have occurred
second.

4 lines of code changed in 1 file:

jthornoc 2015-04-23 11:30 Rev.: 53401

Avoid divide by zero.

5 lines of code changed in 1 file:

harman 2015-04-23 09:16 Rev.: 53400

Added ability to monitor a vector of dbgCells when debugging.

45 lines of code changed in 4 files:

jthornoc 2015-04-22 13:03 Rev.: 53399

More modification to the heat flux model. Now integrating total heat
into the surface then renomalizing by area to get a total flux into the
wall.

18 lines of code changed in 1 file:

dav 2015-04-21 17:37 Rev.: 53398

Update to newest BOOST flags. Add Mustang

20 lines of code changed in 2 files:

jthornoc 2015-04-21 17:32 Rev.: 53397

Compute the exposed surface area for the heat flux and store it.

32 lines of code changed in 2 files:

harman 2015-04-21 15:06 Rev.: 53395

added bulletproofing for GPU code:
- throw exception if halo > 0
- throw exception if ROI != patch_based

clean out some sanity checks debugging code.

21 lines of code changed in 3 files:

ahumphrey 2015-04-21 14:22 Rev.: 53393

Remove unused variable.

0 lines of code changed in 1 file:

ahumphrey 2015-04-21 14:20 Rev.: 53392

Abstract out GPUDataOnion functionality into device function (updateSumI_MLDevice<T>).

221 lines of code changed in 1 file:

harman 2015-04-21 13:50 Rev.: 53391

Report the level that detailed task is running on.

1 lines of code changed in 1 file:

harman 2015-04-20 12:47 Rev.: 53389

output run time in sec.milliseconds to the stats file

0 lines of code changed in 2 files:

derekhar 2015-04-20 12:43 Rev.: 53388

Added initialization model. This model makes radiation comparisons on the first time-step.

16 lines of code changed in 5 files:

harman 2015-04-20 09:41 Rev.: 53387

added -f option to /bin/csh command

This ignores the user's .cshrc file.

0 lines of code changed in 4 files:

ahumphrey 2015-04-17 18:24 Rev.: 53386

Delay deletion of Ray object.

There seems to be a race condition surrounding the deletion of the SimulationInterface object (UintahParallelComponent) in sus.cc. When this is done, it appears that for the RMCRT_Test destructor may evaluate d_RMCRT as non-null before it is deleted. This allows a second attempt to delete d_RMCRT, causing a double delete.

NOTE: credit goes to John Holmen for finding this one. Thanks John.

4 lines of code changed in 1 file:

jthornoc 2015-04-17 17:22 Rev.: 53385

Previous logic was broken for multi-patch

19 lines of code changed in 1 file:

jthornoc 2015-04-17 16:26 Rev.: 53384

Adding a total heat flux variable that is the sum over all components.
Placing this value in the wall cell adjacent to the flow.

25 lines of code changed in 1 file:

harman 2015-04-17 09:16 Rev.: 53383

Consolidated sci_debug stream code for "put" calls into printDebuggingPutInfo() function.
Added several calls to printDebuggingPutInfo()

This eases the swirling head symptoms.

39 lines of code changed in 2 files:

ahumphrey 2015-04-16 15:20 Rev.: 53382

Stop kernel execution once printError calls are made.
At this point we know it's fatal and the reason has been reported.

6 lines of code changed in 1 file:

ahumphrey 2015-04-16 15:10 Rev.: 53381

Fix bug in GPUDW error reporting. Was reporting mixture of varDB and LevelDB info in levelDB error reporting.
Modify/enhance other GPUDW reporting.

23 lines of code changed in 3 files:

jas 2015-04-16 12:52 Rev.: 53380

Added new DOSweep example component for testing/development purposes.

334 lines of code changed in 4 files:

ahumphrey 2015-04-15 21:58 Rev.: 53379

* Refactoring on var and level databases.
* Better reporting on capacity, etc.

55 lines of code changed in 2 files:

ahumphrey 2015-04-15 16:46 Rev.: 53378

Eliminate duplicate levelDB items in GPUDW.

8 lines of code changed in 1 file:

harman 2015-04-15 15:58 Rev.: 53377

printGetError() & printGetLevelError()
output matlIndx, patchID and levelIndex about all available labels


4 lines of code changed in 1 file:

harman 2015-04-15 15:30 Rev.: 53376

For clarity global variable name changes

levelID -> levelIndx
matlID -> matlIndx


94 lines of code changed in 2 files:

harman 2015-04-15 15:24 Rev.: 53375

added printGetLevelError() a method for printing out errors related to the levelDB .

27 lines of code changed in 2 files:

ahumphrey 2015-04-15 14:53 Rev.: 53374

* Further GPUDW thread safety.
* Check return values of CUDA calls.
* Change private methods to return bool - status of CUDA calls made within fucntion body.

214 lines of code changed in 2 files:

guilkey 2015-04-15 14:02 Rev.: 53373

Get the A in AMR working. Fixed a small bug in the calculation of the particle
size as particles move between levels.

Contains a fair bit of debugging output, but I'm the only one running it, so...


265 lines of code changed in 2 files:

guilkey 2015-04-15 13:53 Rev.: 53372

Adding a CC variable to mark cells for refinement.

4 lines of code changed in 2 files:

harman 2015-04-15 13:19 Rev.: 53371

Added matlIndx to the GPUDW level database. Calling methods need to include a matlIndx.


24 lines of code changed in 4 files:

harman 2015-04-15 12:24 Rev.: 53370

SCI_DEBUG output: more descriptive and improved formatting.

26 lines of code changed in 2 files:

harman 2015-04-14 15:29 Rev.: 53369

previous commit had a compile bug. Fixed.

18 lines of code changed in 2 files:

harman 2015-04-14 13:21 Rev.: 53368

1 patch with a ray in 1 cell now works GPU:DataOnion results == CPU:DataOnion

Mirroring debugging printf statements in GPU & CPU code.


31 lines of code changed in 2 files:

harman 2015-04-13 17:05 Rev.: 53367

changed DEBUG_MT -> FIXED_RANDOM_NUM to be consistent with GPU RMCRT code.

3 lines of code changed in 1 file:

harman 2015-04-13 13:19 Rev.: 53366

computeExtents()
- using >getExtraCellLowIndex() & getExtraCellHighIndex() instead of getCellLowIndex & getCellHighIndex()
This fixes the halo[0,0,0] bug.

Cleaned up the debugging output so it's easier to follow. No using d_dbgCell for
a tracking cell instead of hardwired IntVector.

37 lines of code changed in 3 files:

ahumphrey 2015-04-10 23:47 Rev.: 53365

Report when using GPUs

9 lines of code changed in 1 file:

ahumphrey 2015-04-10 23:19 Rev.: 53364

Fix race conditions in the GPU DataWarehouse (varDB and levelDB)
This was unpleasent to track down.

GPU DataOnion now runs fully distributed with arbitrary numbers of patches.

30 lines of code changed in 3 files:

harman 2015-04-10 18:06 Rev.: 53363

Comment out the Ray Tracing chunk of code for Alan.

Wrapped conditionals with multiple tests in (). The behavior of CUDA differs from C++

Added more debugging output.

77 lines of code changed in 3 files:

harman 2015-04-10 16:36 Rev.: 53362

Incremental progress. One z slice will run through all
the pieces of code. There are "GPU OUT_OF_BOUND" errors.


66 lines of code changed in 2 files:

harman 2015-04-10 13:10 Rev.: 53361

Moved the gut of updateSumI_ML into the main kernel. Hack to get it up and running.

updated ToDo list.

It compiles and doesn't crash.

238 lines of code changed in 2 files:

harman 2015-04-10 08:59 Rev.: 53360

Changed name of global variable levels -> d_levels,

updateSumI_MLDevice()
Now Using d_levels
Added debugging code
There are problems when accessing the grid vars.

85 lines of code changed in 2 files:

jas 2015-04-09 22:00 Rev.: 53359

Revert hypre configure changes.

18 lines of code changed in 2 files:

harman 2015-04-09 17:15 Rev.: 53358

More pieces filled in.
getCellPosition() and mapCellToCoarser() are now working.

Moved operators to the top of the .cuh file. Now they can be used downstream in that file.


219 lines of code changed in 3 files:

ahumphrey 2015-04-09 16:37 Rev.: 53357

#if an additional line until updateSumI_MLDevice< T > is working.
#if some debugging code.

8 lines of code changed in 1 file:

jsutherland 2015-04-09 16:24 Rev.: 53356

Revert "Add a simple memory allocation check for hypre for library determination."

This reverts commit e1bbba9fb08718314244a92876afce8eed923060.

1 lines of code changed in 2 files:

ahumphrey 2015-04-09 16:09 Rev.: 53355

Correct access to fine level patches in GPUDW. Add enhanced debug output.

43 lines of code changed in 1 file:

harman 2015-04-09 13:31 Rev.: 53354

GPUVector.h
- removed my previous commit. It's compilicates debugging in DDT.

Passing fineLevel_ROI into the kernel

Retreiving fine level data from GPU_dw is not working as it should....

74 lines of code changed in 5 files:

jas 2015-04-09 12:36 Rev.: 53353

Add a simple memory allocation check for hypre for library determination.

7 lines of code changed in 2 files:

jas 2015-04-09 10:49 Rev.: 53352

Reverse the order in which the hypre libraries are discovered.

17 lines of code changed in 2 files:

ahumphrey 2015-04-08 19:28 Rev.: 53351

Strip out 2-level hard-wiring, now using levelParams.
Clean up GPUDataWarehouse debugging info.

8 lines of code changed in 4 files:

ahumphrey 2015-04-08 18:57 Rev.: 53350

Copy levelParams array to device constant cache. levelParams for GPU:DataOnion is now working.
-- Todd is even happier.

This commit is a point of reference before yanking the 2-level hard-wiring.

68 lines of code changed in 2 files:

harman 2015-04-08 16:58 Rev.: 53349

added more pieces to the puzzle.
This compiles.

38 lines of code changed in 2 files:

harman 2015-04-08 15:29 Rev.: 53348

Added convinence methods for GPUVector. They should be moved upstream.

updateSumI_MLDevice()
- fill in the easy pieces.

minor formatting.


283 lines of code changed in 3 files:

harman 2015-04-08 15:24 Rev.: 53347

Added comment and convience methods for gpuVector & gpuIntVector

5 lines of code changed in 2 files:

harman 2015-04-08 09:59 Rev.: 53346

Incremental commit. Removed some of the #if 0s.

Most of the pieces are working. UpdateSumI_MLDevice still needs to be filled in.


22 lines of code changed in 1 file:

ahumphrey 2015-04-08 00:40 Rev.: 53345

Increase the limit on MAX_LVITEM.

1 lines of code changed in 1 file:

ahumphrey 2015-04-08 00:34 Rev.: 53344

Fix issues with GPUDataWarehouse levelDB.
GPU DataOnion coarse levels are now handled correctly.
- Todd is happy

14 lines of code changed in 2 files:

ahumphrey 2015-04-08 00:07 Rev.: 53343

Use levelDB in GPUDW for per level vars. Provide get, put, allocateAndPut interface for this via GPUDW.
Also remove code related to cudaHostRegister and pinned pointers in UnifiedScheduler. This won't be used and ultimately will be replaced with contiguous buffer copies.

174 lines of code changed in 4 files:

ahumphrey 2015-04-07 23:06 Rev.: 53342

Add correct multi-level GPUDW support.
This also fixes faulty level->findCellIndexRange() and subsequent dw->getRegion() calls on coarse levels.

17 lines of code changed in 1 file:

ahumphrey 2015-04-07 18:04 Rev.: 53341

Revert commit r53338

1 lines of code changed in 1 file:

derekhar 2015-04-07 16:40 Rev.: 53340

Removed clutter in code by Changing T*T*T*T to pow(T,4). Will break all radiation tests.

2 lines of code changed in 1 file:

harman 2015-04-07 16:30 Rev.: 53339

added updateSumI_MLDevice() which is just a copy ofupdateSumI_Device() at this point.
Heavy construction zone with little working.


275 lines of code changed in 2 files:

ahumphrey 2015-04-07 16:27 Rev.: 53338

Put GPUGridVariables into GPUDW under the correct levelID.

1 lines of code changed in 1 file:

awabboud 2015-04-07 15:17 Rev.: 53336

Fix CQMOM inversion for N-D internal coordinates, update to input files for these cases and decrease total quadrature nodes

3 lines of code changed in 1 file:

harman 2015-04-07 15:09 Rev.: 53335

Added the utilities
isThread0
printThread()
printBlock()

printGetError()
- added level to the print statement. Only output with thread0, this helps but
there are still issues.

48 lines of code changed in 4 files:

harman 2015-04-06 18:25 Rev.: 53331

The most shameful commit I've ever made.

- Pass each of the level variables, level by level into
the kernel. I need to move foreward and this grossness works. I spent 10X too much time
trying to figure out how to pass an array of struct into a kernel.

It crashes trying to get abskg from the dw.



264 lines of code changed in 3 files:

jthornoc 2015-04-06 16:03 Rev.: 53330

Pair the density label in the scheduler.

8 lines of code changed in 3 files:

jholmen 2015-04-04 13:03 Rev.: 53327

* Explicitly identify the number of processors

1 lines of code changed in 1 file:

jholmen 2015-04-04 12:57 Rev.: 53326

* Cleanup

9 lines of code changed in 2 files:

harman 2015-04-02 09:56 Rev.: 53324

Added parametric study to quantify divQ error vs fine patch layout

127 lines of code changed in 3 files:

tsaad 2015-04-01 11:24 Rev.: 53322

fix bug when gas-particle coupling is turned on. The bug affected the pressure projection and resulted in mass addition to the system. This will change RT answers.

3 lines of code changed in 1 file:

March 2015 »

Generated by StatSVN 0.7.0