[Yt-dev] RMS mass overdensity method
Stephen Skory
stephenskory at yahoo.com
Wed Mar 4 18:52:13 PST 2009
Matt,
> In your lib/python2.6/site-packages/
> is there a 'numpy' directory, or is it an egg? What's numpy* in that
> directory?
It's a directory.
> As I recall, on Ranger you have to use the mpi4py
> executable to run the scripts; did you re-compile and re-install this
> when you upgraded to the newest mpi4py from google code?
Yes, I did.
> Can you
> ensure that you did by deleting the existing one and re-installing
> from the google code repo?
I'm confident that I already had the newest version of mpi4py. Also, see below.
> What happens if you run on *one* processor
> launched with python rather than mpi4py, with *no* MPI calls? Does it
> finish?
I did 1000 samples, no problem, with the new install of yt.
> And the request would be to try with a new, fresh install using the
> latest install script
I just installed a piping fresh new copy of yt, along with the google code mpi4py, and a parallel run is crashing in exactly the same place as before.
I think I need to try this stuff tomorrow on Kraken.
_______________________________________________________
sskory at physics.ucsd.edu o__ Stephen Skory
http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student
________________________________(_)_\(_)_______________
More information about the yt-dev
mailing list