develooper Front page | perl.perl5.porters | Postings from January 2001

Re: perls surprising areas of slowness

Thread Previous | Thread Next
From:
Nicholas Clark
Date:
January 5, 2001 03:44
Subject:
Re: perls surprising areas of slowness
Message ID:
20010105114451.B26914@plum.flirble.org
On Sun, Dec 24, 2000 at 12:44:08AM +0000, Simon Cozens wrote:
> On Wed, Dec 20, 2000 at 10:12:39PM +0000, Nicholas Clark wrote:
> > pp_subtract and pp_gt featured more highly than pp_add
> > (IIRC) the latter is in pp_hot.c, the other two are not.
> 
> For information, here are the top forty PP functions in the profile. Those
> marked with an asterisk are not in pp_hot.c. Some of them (like rv2sv, seq and
> not) seem bizarre omissions.

Thanks for generating this
 
[17 pp functions]
> Perl_pp_modulo      *

These timings are for the regression test suite, aren't they?
Is the regression test suite considered a representative enough cross
section of perl code to consider optimising for (I realise parts are
written in an odd style, but does it do typical things in the correct sort
of proportions, or is something like the perlbench suite a better guide?)

Because (IIRC) pp_modulo is called very few times and takes a really really
long time for each in the regression tests, but doesn't feature in the
list Alan Burlinson sent.

Would it benefit from "work"? (actually, currently, it an pp_divide haven't
had the 64 bit integer treatment so some work is needed)

Nicholas Clark

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About