OONumerics User : |
From: Paul Beardsley (pab_at_[hidden])
Date: 2006-03-27 13:47:11
Hari,
Your first step should (always) be to use a profiler, if you haven't
already
done that. Identify explicitly where the cycles are going and then think
about
speed-ups for each case in turn.
If you are using standard library containers, be sure to use the appropriate
container according to the types of insert and deletions you are doing,
be sure
to use 'reserve' whenever possible.
Also try your compilation at optimization levels beyond O2,
Paul.
Hari Sundar wrote:
> I understand ...
>
> When I meant it is very slow .. it is slow because it is doing a lot
> of number crunching .. and I want to speed it up as much as possible.
> I have already manually unrolled the loops within the function being
> called and now was trying to get rid of the additional cost of the
> virtual function call ... The Barton-Nackman trick seems to work ...
> as Simon suggested. If there is something more I can do .. please let
> me know.
>
> thanks for all the suggestions,
> ~Hari
>
>
> On 3/27/06, *Paul Beardsley* <pab_at_[hidden] <mailto:pab_at_[hidden]>> wrote:
>
> Also have you verified that the slow performance is due to the virtual
> function call? In general, this is not an area that you can
> target and
> expect to get big gains. You said that your program is 'very slow'.
> If you want to speed it up by an order of magnitude, you probably
> won't
> be in luck just attacking the virtual function call,
>
> Paul.
>
> Simon Perreault wrote:
> > On Sunday 26 March 2006 22:53, Hari Sundar wrote:
> >
> >> Do people have any experience with this kind of stuff ? what
> would be the
> >> best way to do something like this ... where the data is stored
> within the
> >> parent class and the derived classes are mainly providing
> different
> >> functions that process this data differently.
> >>
> >
> > Can't you just use the Barton-Nackman trick? Make the data
> protected in the
> > parent so that the children can access it. Or am I missing
> something?
> > _______________________________________________
> > oon-list mailing list
> > oon-list_at_[hidden] <mailto:oon-list_at_[hidden]>
> > http://www.oonumerics.org/mailman/listinfo.cgi/oon-list
> >
>
>
> --
> Paul Beardsley, Senior Research Scientist
> MERL, 201 Broadway, Cambridge, MA 02139, USA
> tel: +(617) 621 7569
> fax: +(617) 621 7550
>
>
>
>
>
> --
> इयं विसृषà¥à¤Ÿà¤¿à¤°à¥à¤¯à¤¤à¤½à¤†à¤¬à¤à¥‚व यदि वा दधे यदि वा न ।
> योऽअयसà¥à¤¯à¤¾à¤§à¥à¤¯à¤•à¥à¤·à¤ƒ परमे वà¥à¤¯à¥‹à¤®à¤¨à¥à¤¤à¥à¤¸à¥‹à¤½à¤…ङà¥à¤— वेद यदि वा न वेद ।।
>
> Whence all creation had its origin,
> He, whether he fashioned it or whether He did not,
> He, who surveys it all from the highest heaven,
> He knows - or maybe even He does not know.
>
> ~Rg veda