Loop performance disappearance

[François Pinard]

Mikael Johansson <mikael.johansson at helsinki.fet> writes:

> for i in range(loops):
>     pass

> If loops=500 000 (space added for clarity) the "program" executes in ~2
> secs on my machine. But if loops is set to 5 000 000, the execution time
> rises to substantially more than tenfold (I terminated it after one
> minute). However the CPU-load is quite small, most of the time is spent
> disk swapping like crazy!

> Any ideas of the reason for this, work-arounds?

Sure.  range(N) does allocate a list containing integers from 0 to N-1.
So, if you allocate a huge list, more than the physical memory can hold,
you are likely to need swapping space.

The work-around is to use xrange instead of range.  It yields a kind of
lazy-evaluated tuple, consuming almost no memory.  I would guess that
range is to be preferred over xrange for when N is not very big.

-- 
François Pinard   http://www.iro.umontreal.ca/~pinard