Re: Request for further enlightenment

[yodabjorn]

Heres a good way to test:

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

my $count = shift || "-3" ;  # set the test counter 
my @array = ();

#load up an array of crap
for ( my $i = 0; $i < 1000; $i++)
{ push @array, rand(256); }


open NULL, "+/dev/null" or die "$!";
timethese (
    $count,
    {
        'join'      => '&print_join',
        'map'       => '&print_map',
        'field sep' => '&print_field'
    }
);

# use join to print each array elemnet. 
sub print_join
{
    print NULL join("\n", @array) ;
}

# use map to print each array elemnt, 
sub print_map 
{
    print NULL map{ $_ .= "\n"}  @array ;
}

# use map to print each array elemnt, 
sub print_field
{
    local $,="\n";
    print NULL  @array ;
}

Ran for default 3 CPU seconds:

./time_print 
Benchmark: running field sep, join, map, each for at least 3 CPU seconds...
 field sep:  3 wallclock secs ( 3.15 usr +  0.00 sys =  3.15 CPU) @ 53645.71/s (n=168984)
      join:  3 wallclock secs ( 3.05 usr +  0.01 sys =  3.06 CPU) @ 1256.86/s (n=3846)
       map:  5 wallclock secs ( 4.56 usr +  0.04 sys =  4.60 CPU) @ 119.78/s (n=551)

looks like the field sepperator is the fastest with 53,645.71 /sec
hope this helps

[Juerd]

You are being unfair to map.

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

my $count = shift || -3 ;  # set the test counter
my @array;

#load up an array of crap
push @array, rand(256) for 1 .. 1000;

open NULL, ">/dev/null" or die $!;
timethese (
    $count,
    {
        'join'      => \&print_join,
        'map'       => \&print_map,
        'list map'  => \&print_map_list,
        'field sep' => \&print_field,
        'for'       => \&print_for
    }
);

# use join to print each array elemnet.
sub print_join {
    print NULL join("\n", @array);
}

# use map to print each array elemnt,
sub print_map {
    print NULL map{ $_ .= "\n"}  @array;
}

# same but a no concat
sub print_map_list {
    print NULL map { ($_, "\n") } @array;
}

# use map to print each array elemnt,
sub print_field {
    local $, = "\n";
    print NULL @array;
}

# use a loop.
sub print_for {
    print NULL $_, "\n" for @array;
}

Yields:

Benchmark: running field sep, for, join, list map, map, each for at least 9 CPU seconds...
 field sep: 11 wallclock secs ( 9.60 usr +  0.05 sys =  9.65 CPU) @ 1121.04/s (n=10818)

       for: 11 wallclock secs ( 9.28 usr +  0.02 sys =  9.30 CPU) @ 621.40/s (n=5779)
      join: 11 wallclock secs ( 9.20 usr +  0.02 sys =  9.22 CPU) @ 2888.07/s (n=26628)
  list map: 11 wallclock secs ( 9.07 usr +  0.04 sys =  9.11 CPU) @ 268.94/s (n=2450)
       map: 32 wallclock secs (26.02 usr +  0.26 sys = 26.28 CPU) @ 57.95/s (n=1523)

What happened with your field sep test? I can't get it as fast as you claim it to be. You should also note that join does not put a newline after the final item.

- Yes, I reinvent wheels.
- Spam: Visit eurotraQ.

[MeowChow]

Still unfair...

sub print_map_interp { print NULL map "$_\n", @array; } ### RESULTS ### Benchmark: running field sep, for, interp map, join, list map, map, ea +ch for at least 3 CPU seconds... field sep: 3 wallclock secs ( 3.09 usr + 0.05 sys = 3.14 CPU) @ 11 +76.75/s (n=3695) for: 3 wallclock secs ( 3.14 usr + 0.03 sys = 3.17 CPU) @ 81 +4.63/s (n=2584) interp map: 3 wallclock secs ( 3.09 usr + 0.02 sys = 3.11 CPU) @ 74 +5.90/s (n=2319) join: 4 wallclock secs ( 3.05 usr + 0.17 sys = 3.22 CPU) @ 36 +80.02/s (n=11846) list map: 3 wallclock secs ( 3.00 usr + 0.02 sys = 3.02 CPU) @ 44 +6.62/s (n=1347) map: 6 wallclock secs ( 6.19 usr + 0.17 sys = 6.36 CPU) @ 61 +.17/s (n=389) Tool completed successfully

   MeowChow                                   
               s aamecha.s a..a\u$&owag.print

[yodabjorn]

I lookoed at the code for using the field sep and it tests fine. I ran the program a few more times:

 ./time_print 
Benchmark: running field sep, join, map, each for at least 3 CPU seconds...
 field sep:  4 wallclock secs ( 3.15 usr +  0.01 sys =  3.16 CPU) @ 53475.95/s (n=168984)
      join:  4 wallclock secs ( 3.15 usr +  0.00 sys =  3.15 CPU) @ 1300.00/s (n=4095)
       map:  3 wallclock secs ( 3.12 usr +  0.00 sys =  3.12 CPU) @ 302.88/s (n=945)

each time on my maching the field sep is really fast.
I changed the map to map {"$_\n"} @array for these runs