Re: Taking advantage of L1 and L2 cache in sort

[Jim Meyering]

Chen Guo wrote:
> Fascinating, Padraig. I wonder, did this email just get to my inbox really,
> really late, or what prompted you to dig up a 2 year old thread?
>
> Anyhow, it seems you had to specifically unset MALLOC_PERTURB_,
> so is it by default set?

MALLOC_PERTURB_ should be set for most people.
The idea is to detect problems earlier rather than later,
and essentially for free.  (except when profiling/timing
and with certain work loads or tight mem/CPU constraints)

I've written about this a few times.
Here's the most recent appeal:

    http://thread.gmane.org/gmane.linux.redhat.fedora.devel/165080