Accessor performance mop(Jul 21, 2013) vs Class::Accessor::Fast vs direct hash

[yuki-kimoto]

use mop;
use Data::Dumper;

class mopClass {
  has $foo is rw;
}

package ClassAccessorFastClass;
use strict;
use warnings;
use base 'Class::Accessor::Fast';

__PACKAGE__->mk_accessors('foo');

package main;

use Benchmark qw/timethese cmpthese/;

my $mop_obj = mopClass->new;
my $class_accessor_fast_obj = ClassAccessorFastClass->new;
my $hash = {};

my $result = timethese(100000, {
  # mop accessor
  mop_accessor => sub {
    $mop_obj->foo('Hello');
    my $foo = $mop_obj->foo;
  },
  # Class::Accessor::Fast accessor
  accessor_fast_accessor => sub {
    $class_accessor_fast_obj->foo('Hello');
    my $foo = $class_accessor_fast_obj->foo;
  },
  # Direct hash
  direct_hash => sub {
    $hash->{foo} = 'Hello';
    my $foo = $hash->{foo};
  }
});

cmpthese $result;

Result.

Benchmark: timing 100000 iterations of accessor_fast_accessor, direct_hash, mop_accessor...
accessor_fast_accessor:  1 wallclock secs ( 0.17 usr +  0.00 sys =  0.17 CPU) @ 588235.29/s (n=100000)
            (warning: too few iterations for a reliable count)
direct_hash:  0 wallclock secs ( 0.04 usr +  0.00 sys =  0.04 CPU) @ 2500000.00/s (n=100000)
            (warning: too few iterations for a reliable count)
mop_accessor: 17 wallclock secs (17.22 usr +  0.06 sys = 17.28 CPU) @ 5787.04/s (n=100000)
                            Rate mop_accessor accessor_fast_accessor direct_hash
mop_accessor              5787/s           --                   -99%       -100%
accessor_fast_accessor  588235/s       10065%                     --        -76%
direct_hash            2500000/s       43100%                   325%          --

mop accessor is very very slow.

[stevan]

Examining performance at this stage is very premature, it is just a prototype.

[stevan]

I am going to close this because performance is irrelevant at the prototype stage, we can revisit this later.