Account Options

  1. Sign in
The old Google Groups will be going away soon, but your browser is incompatible with the new version.
Google Groups Home
« Groups Home
comp.lang.perl.misc
+ new post
About this group
Subscribe to this group
This is a Usenet group - learn more
Message from discussion Benchmark: ST vs OM
The group you are posting to is a Usenet group. Messages posted to this group will make your email address visible to anyone on the Internet.
Your reply message has not been sent.
Your post was successful
 
From:
To:
Cc:
Followup To:
Add Cc | Add Followup-to | Edit Subject
Subject:
Validation:
For verification purposes please type the characters you see in the picture below or the numbers you hear by clicking the accessibility icon. Listen and type the numbers you hear
 
John Porter  
View profile  
 More options Oct 2 1998, 3:00 am
Newsgroups: comp.lang.perl.misc
From: John Porter <jdpor...@min.net>
Date: 1998/10/02
Subject: Benchmark: ST vs OM
Print | View thread | Show original | Report this message | Find messages by this author
I've recently been benchmarking the relative performance of
the Schwartzian Transform (ST) (aka. map-sort-map) vs.
the Orcish Maneuver (say "OR-cache").

I recently posted some results that indicate that ST is
more than an order of magnitude slower than OM.
However, those tests were not quite realistic, in that they
discarded the results of the operation.  Apparently Perl can
optimize for this, under the right conditions.  What follows
is a benchmark wherein the results are not discarded.

#!/usr/local/bin/perl -w
use Benchmark;

# read 3100 7-letter words.  shuffle them.
my $file = 'words.7';
open F, $file or die "open $file: $!\n";
my @in = shuffle( <F> ); chomp @in; close F;
my @out;

sub control {  # explicit comparison
  @out = sort { $a cmp $b } @in

}

sub control0 { # default comparison
  @out = sort @in

}

sub ST {
  @out =
    map { $_->[0] }
      sort { $a->[1] cmp $b->[1] }
        map  { [ $_, $_ ] }
          @in;

}

sub OM {
  my %c = (); # the cache
  @out = sort { ($c{$a} ||= $a) cmp ($c{$b} ||= $b) } @in;

}

timethese( 10, {
  'ST' => \&ST,
  'OM' => \&OM,
  'ct' => \&control,
  'c0' => \&control0,

} );

sub shuffle {  # the usual Fisher-Yates shuffle.
  my @i = @_;
  for my $i ( reverse ( 0 .. $#i ) ) {
    my $j = int( rand( $i ));
    @i[$i,$j] = @i[$j,$i];
  }
  @i;

}

Benchmark: timing 10 iterations of OM, ST, c0, ct...
        OM: 33 secs (32.51 usr  0.05 sys = 32.56 cpu)
        ST: 30 secs (29.70 usr  0.09 sys = 29.79 cpu)
        c0:  2 secs ( 2.52 usr  0.00 sys =  2.52 cpu)
        ct: 12 secs (11.48 usr  0.00 sys = 11.48 cpu)

In these tests, I am trying to compare the underlying
algorithms of ST and OM.  Therefore, the compare keys
are identical to the original strings, rather than
adding the overhead of computing some other comparison
keys.
Note how using an explicit {$a cmp $b} adds an order
of magnitude overhead.  Perl clearly optimizes for
the default sort (the implicit {$a cmp $b}).

--
John "Many Jars" Porter


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Create a group - Google Groups - Google Home - Terms of Service - Privacy Policy
©2013 Google