Commits

Author Commit Message Labels Comments Date
Nathan Goldbaum
Updating the NASA Pleiades makefile.
Michael Kuhlen
Adding documentation for the new H2-regulated star formation method.
Michael Kuhlen
Adding H2-regulated star formation (star_maker_h2reg.F).
Michael Kuhlen
Make performance_tools play nicely with single-cycle performance data (performance-single.out).
Brian O'Shea
Merged in MatthewTurk/enzo-dev (pull request #39)
Matthew Turk
PyArray_SimpleNewFromData does not need to have NPY_OWNDATA turned off. This also removes YT_Temperature from the grid object, makes it dynamically allocated, and ensures that it is memory-managed by NumPy.
Matthew Turk
Adding a PythonReloadScript parameter and adding calls to gc.collect() after clearing the dictionary of grid data and hierarchy information.
Matthew Turk
Fixes for getting inline python to work on Blue Waters
Matthew Turk
Merged in samskillman/enzo-dev-timing (pull request #38)
Sam Skillman
Some touchups.
Sam Skillman
Adding Documentation to reference/PerformanceMeasurement, updating README
Brian O'Shea
Merged in mqk/enzo-dev-mqk (pull request #36)
Michael Kuhlen
Adding rates k27 - k31 for RadiationFieldType == 13.
Michael Kuhlen
Fixed a bug in DetermineNumberOfNodes.C, which gave incorrect results in cases where one node name is a substring of another (e.g. r196i0n1 and r196i0n10). Use strcmp() instead of strstr().
Michael Kuhlen
Merged in changes from bb repo.
Matthew Turk
Merging
Matthew Turk
Correct a typo in the Blue Waters makefile
Britton Smith
Merged in samskillman/enzo-dev-timing (pull request #35)
Sam Skillman
Fixing a bug with use-mpi-false configuration.
Cameron Hummels
Updated the comments and width of lines of text so they don't wrap in a standard xterm window.
Cameron Hummels
Fixed a bug in plot_maxmin().
Sam Skillman
Moving performance_tools to be in src/ and not src/enzo/. Also changing default performance.out so that it reflects changes to using updates/s/processor.
Sam Skillman
Instead of ignoring the nan messages, modify the error call to still warn the user, but let them know it is probably okay, and just due to a 0.0 timer.
Sam Skillman
Updating both README's. The performance_tools to reflect the addition of two more plots by default, and the enzo one to add me to the author list.
Sam Skillman
Ignore divide by zero warning messages, and get rid of NaNs that pop up when things like WriteAllData are not called every cycle. Also set ylim to 0-1 for p4.png.
Cameron Hummels
Changing instances of "cells" to "updates", as in "cells/sec/processor" is now "updates/sec/processor". This reflects the change made to report the cell updates instead of just computing the number of cells existing per dt.
Cameron Hummels
Added plot_maxmin function to performance_tools.py. This function plots the maximum time taken on a processor minus the minimum time taken on a processor for a given task or level. You can also scale this against the mean time taken per processor for that task to give you a clue as to how well you're balancing load.
Sam Skillman
It turns out some compilers (gcc-4.6 and higher, maybe?), require both <string> and <cstring> to get the string class and the C string functions. This should fix continuing errors on kraken with gnu compilers.
Sam Skillman
<string> is now the accepted way as opposed to <string.h>, fixes compile on Kraken
Sam Skillman
Updating some of the description in README
  1. Prev
  2. Next