Re: [boost] intrusive container performance example

From: Michael Fawcett (michael.fawcett_at_[hidden])
Date: 2007-03-17 21:25:55

On 3/17/07, Gennadiy Rozental <gennadiy.rozental_at_[hidden]> wrote:
>
> "Michael Fawcett" <michael.fawcett_at_[hidden]> wrote in message
> news:bc5bffe80703161407ybbc7c71ja06b0c2230c5602c_at_mail.gmail.com...
> > On 3/16/07, Gennadiy Rozental <gennadiy.rozental_at_[hidden]> wrote:
> >> So? could you give an example of an algorithm? And/or other speific
> >> performace advantage example?
>
> Would you care to share with us a source code for your test?
>
> > I just constructed a quick test and used VTune and std::clock to
> > sample.  Here are my results:
> >
> > sorting across 1,000,000 random integers (averaged across 5 runs)
> > std::list::sort - 6.87s
> > boost::ilist::sort - 4.45s
> >
> > number of L2 cache requests (L1 cache misses)
> > std::list - 1,790
> > boost::ilist - 862
>
> How about std::vector?

Sure, I'm at home now, but I can post it from work on Monday.  I
didn't test vector, but I assume it would beat intrusive list in speed
and have fewer cache misses as well.

--Michael Fawcett