in reply to Re: Count number of occurrence of word
in thread Count number of occurrence of word

Hi, the strange thing is that, replicating your benchmark (together with a third solution), I get a much higher advantage for the 'a' solution than what you got: 
$ perl -e 'use Benchmark(cmpthese);
cmpthese(1000000, {
     "a" => sub {
        $_ = "this that this those this these.";
        $a = () = /this/g;
    },
     "b" => sub {
        $_ = "this that this those this these.";
        $a++ while(s/this//);
    },
     "c"  => sub {
         $_ = "this that this those this these.";
         $a = s/this/this/g;
    }
});'
      Rate    a    b    c
a 282406/s   -- -28% -28%
b 390778/s  38%   --  -1%
c 393391/s  39%   1%   --

[download]
Note that I had to use one million iterations, with 100_000 I got the "too few iterations" warning.

Replies are listed 'Best First'.
Re^3: Count number of occurrence of word
by Lennotoecom (Pilgrim) on Mar 09, 2014 at 19:22 UTC
    ummm you mean a disadvantage? 
    use Benchmark(cmpthese);
cmpthese(1000000, {
    'a' => sub {
        $_ = 'this that this those this these.';
        $a = () = /this/g;
    },
    'b' => sub {
        $_ = 'this that this those this these.';
        $a++ while(s/this//);
    },
    'c' => sub {
        $_ = 'this that this those this these.';        
        $a = s/this/this/g;
    }
});

    [download]
    my machine shows that your 'c' option is the fastest of these
          Rate    a    b    c
a 478240/s   --  -4% -19%
b 497018/s   4%   -- -16%
c 588235/s  23%  18%   --

    [download]
[reply]
[d/l]
[select]
      Gosh, you are right, how silly on my part. I looked at the results too quickly and interpreted them the wrong way.
[reply]
        happens to everybody,
        and still your option is the fastest one here ^)
[reply]
[reply]

In Section Seekers of Perl Wisdom