|
|
Ublas : |
From: Toon Knapen (toon.knapen_at_[hidden])
Date: 2005-06-01 15:19:32
> That would definitely explain it and would be consistent the differences
> are so great between my old Celeron and Stefan's Pentium 4 which have
> wildly different sized caches. However, the question then become why
> would the operators be implemented so naively?
Performance is of course of utmost performance to a vector-matrix
library but even more important is correctness. We have to assume that
there can be aliasing so therefore the operators are implemented so
'naively'.
If OTOH you know that you have no aliasing (and you know what that means
-> thus you have some ublas experience) we suppose you will find out
about the plus_assign and minus_assign etc. and thus be able to optimise
your routines.
toon