Lines of Code
[root]/Packages/Uintah/CCA/Components/SimulationController
| Author | Changes | Lines of Code | Lines per Change |
|---|---|---|---|
| Totals | 52 (100.0%) | 458 (100.0%) | 8.8 |
| luitjens | 27 (51.9%) | 294 (64.2%) | 10.8 |
| worthen | 20 (38.5%) | 149 (32.5%) | 7.4 |
| harman | 1 (1.9%) | 7 (1.5%) | 7.0 |
| amorris | 1 (1.9%) | 4 (0.9%) | 4.0 |
| jas | 2 (3.8%) | 2 (0.4%) | 1.0 |
| dav | 1 (1.9%) | 2 (0.4%) | 2.0 |
move SCISHARE to UINTAHSHARE. Also disable the share flags if building static
15 lines of code changed in 6 files:
Make the toplevel ProblemSpec responsible for deleting its xml tree when deleted. Comment out calls to releaseDocument
1 lines of code changed in 2 files:
First iteration of reducing the "extra communication between patches" warnings. Handles Grid Variables, not particle variables.
Only runs if
<small_messages>true</small_messages>
is specified in the <Scheduler> section.
Requires the DW to know whether or not it's on the initialization timestep.
Also add a new version of VarTracker, which can be configured to print out before receiving communication is scheduled, before a task is run, and after a task is run.
3 lines of code changed in 1 file:
Means to leverage the line-number malloc tag to track iterations, and use it to match tags with the current timestep in sus
2 lines of code changed in 1 file:
Merge from r38647 and r38648
1 lines of code changed in 2 files:
Output the processor that is the max in addition to the max. This could be useful in tracking down our memory issues.
35 lines of code changed in 1 file:
new->scinew
2 lines of code changed in 1 file:
Fixed an array out of bounds bug.
5 lines of code changed in 1 file:
Removed sleep while TAU profiling (was used to help differentiate iterations when tracing)
4 lines of code changed in 1 file:
indenting
2 lines of code changed in 1 file:
Modified the way the SCIDebug MPIBarriers:+ works.
44 lines of code changed in 1 file:
Modified the way dynamic dilation is controlled.
Dilation is now controlled by specifiying the amount of overhead
associated with the grid that is acceptable. At run time the simulation
is profiled and dilation is adjusted in order to keep the amr overhead
in the acceptable range.
grid_reuse_target_low and grid_reuse_target_high have been removed. Now
the controls are amr_overhead_low and amr_overhead_high. These are
values between 0 and 1 that specify the fraction of of the runtime that
is allowed to be spent in amr overhead. The defaults are .05 and .15.
In addition, the parameter max_dilation now exists. It is important to
set this to be smaller than the smallest patch possible (the default is
4,4,4). This sets a ceiling on the amount of total dilation (stability
dilation + regrid dilation) that can
exist in the system.
To enable dynamic dilation add the following to the AMR section of the input file:
<dynamic_dilation>true</dynamic_dilation>
In addition you may want to specify the following in the same section:
amr_overhead_low
amr_overhead_low
max_dilation
40 lines of code changed in 1 file:
Add the percent time in overhead to the ComponentTimes:+ spew
9 lines of code changed in 1 file:
Calculate the percent time spent doing AMR grid related things. This
will be used to control dilation eventually.
59 lines of code changed in 1 file:
Fix seg-fault when printing once-per-proc tasks.
In Switcher::problemSetup, check to see if the regridder needs to be turned on (like on a restart)
6 lines of code changed in 1 file:
Include the time in MPI_Wait{some/all} in the SCI_DEBUG ComponentTimes:+
2 lines of code changed in 1 file:
Free the memory for sub-cycle timesteps when using W-Cycle. Previously this would not be deleted until the end of the coarsest timestep leading to excess memory usage.
Theoritically this could provide up to a factor of (refinement ratio)^(num levels-1) improvement in memory usage.
Bryan & Justin
8 lines of code changed in 1 file:
Reorganize recompile loop for multi taskgraphing to make sure finalizeTimestep gets called on all levels
14 lines of code changed in 1 file:
Implement a task queue based on whether or not MPI recvs have taken place. To do this, I also needed to implement Reduction Variable Modifies. Read closely if you plan to use this: Modification will take place before reduction. Also override the delt on global level in the sim controller (this is mainly if the delt changes as a result of a switch)
5 lines of code changed in 1 file:
A couple small tau changes.
1 lines of code changed in 1 file:
(23 more)