[Octave-bug-tracker] [bug #63791] movfun - high intermediate memory use for moderate data input resulting in out of memory errors

[Nicholas Jankowski]

Follow-up Comment #31, bug #63791 (project octave):

not to keep off on tangents, but 
1) in delaunayn we came up with two schemes that vectorize nD code, but they
scale differently. once is much faster up to 4D, the other is much faster
after that but hits multiplicative memory limits. We had wondered about a
smart way to partition. seems like your try/catch might be good enough (try
the faster way first, if error do the slower but low memory use way.)  will
look into it. 

2) I just helped rework median and mean for statistics that include reshaping
vecdim calls (e.g., dim = [2 4]) to dim1 after the permutation. i did note
that it could hit index limits, but it makes the code simply compatible with
sum, etc that still can't do vecdim.  the error catch is less fruitful there i
think because it would require changing all following processes (for mean
recursively operating on each dim independently being the only easy
alternate.)

will look into these options.


    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?63791>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/