gc penalty of 30-40% when manipulating large data structures?

Istvan Albert istvan.albert at gmail.com
Fri Nov 16 11:43:56 EST 2007


On Nov 16, 10:59 am, "Chris Mellon" <arka... at gmail.com> wrote:

> The GC has a heuristic where it kicks in when (allocations -
> deallocations) exceeds a certain threshold,

As the available ram increases this threshold can be more easily
reached. Ever since I moved to 2Gb ram I stumbled upon issues that
were easily solved by turning the gc off (the truth is that more ram
made me lazier, I'm a little less keen to keep memory consumption down
for occasional jobs, being overly cavalier with generating lists of
1Gb in size...)

One example, when moving from a list size from 1 million to 10 million
I hit this threshold. Nowadays I disable the gc during data
initialization.

i.






More information about the Python-list mailing list