Re: Read the first character of a string

[blakem]

Interesting.... and its nearly as fast as inlining it in C.

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

my $str = "some string here";

die unless (pop_chord($str) . inlined($str) . _substr($str)   .
            _unpack($str)   . _match($str)   
            eq substr($str,0,1)x5);

cmpthese( 500000, {
           'pop_chord' => sub { pop_chord($str) },
                   'inlined'   => sub { inlined($str)   },
           '_substr'   => sub { _substr($str)   },
           '_unpack'   => sub { _unpack($str)   },
           '_match'    => sub { _match($str)    },
          }
        );

sub pop_chord {
  chr ord pop
}

sub _substr {
  substr pop,0,1
}

sub _unpack {
  unpack"a",pop
}

sub _match {
  (pop=~/(.)/s)[0]
}

use Inline C => <<'END_OF_C_CODE';
char inlined(char* str) {
  return str[0];
}
END_OF_C_CODE

__END__

  Benchmark: timing 500000 iterations of _match, _substr, _unpack, inl
+ined, pop_chord...
  _match:    3 wallclock secs ( 3.98 usr +  0.00 sys =  3.98 CPU) @ 12
+5628.14/s (n=500000)
  _substr:   2 wallclock secs ( 1.45 usr +  0.00 sys =  1.45 CPU) @ 34
+4827.59/s (n=500000)
  _unpack:   4 wallclock secs ( 2.68 usr +  0.00 sys =  2.68 CPU) @ 18
+6567.16/s (n=500000)
  inlined:   1 wallclock secs ( 0.58 usr +  0.00 sys =  0.58 CPU) @ 86
+2068.97/s (n=500000)
  pop_chord: 1 wallclock secs ( 1.30 usr +  0.00 sys =  1.30 CPU) @ 38
+4615.38/s (n=500000)
              Rate    _match   _unpack   _substr pop_chord   inlined
_match    125628/s        --      -33%      -64%      -67%      -85%
_unpack   186567/s       49%        --      -46%      -51%      -78%
_substr   344828/s      174%       85%        --      -10%      -60%
pop_chord 384615/s      206%      106%       12%        --      -55%
inlined   862069/s      586%      362%      150%      124%        --

-Blake