Make it 13x faster by making the code unreadable

[AlexDaniel]

Also, potentially up to 39x faster using the recommendation below.

Every single change in this commit actually had a positive impact on
the performance. It is sad that one has to use nqp for everything to
get things working reasonably fast.

By all means do not use .min method in your user code if you want to
get every last drop of performance. At this point everything you do
outside of Sift4 module will have a much bigger impact on the
performance.

For example, this is insanely slow:

say @a.min({ sift4($_, $str, 5) });

And this is twice faster:

my $ans = ‘default value’;
my $ans-min = 6;

for @a {
    my $dist = sift4($_, $str, 5, 5);
    if $dist < $ans-min {
        $ans = $_;
        $ans-min = $dist;
    }
}

One would think that in the code above sift4($_, $str, 5, 5) can be
changed to sift4($_, $str, 5, $ans-min), but no, for some reason it
does not work. Whether it is an issue in this implementation or in the
algorithm itself I don't know. Should be investigated I think.

Anyway, to make it easier for the users I think this module should
provide a function for processing an array (so that users are less
likely to screw it up).

Random testing suggests that these changes do not affect
functionality.

[MasterDuke17]

Thanks. Merged and version bumped.

…

On Mon, Jan 9, 2017 at 11:15 PM, Aleks-Daniel Jakimenko-Aleksejev < ***@***.***> wrote: Also, potentially up to 39x faster using the recommendation below. Every single change in this commit actually had a positive impact on the performance. It is sad that one has to use nqp for everything to get things working reasonably fast. By all means *do not use .min method* in your user code if you want to get every last drop of performance. At this point everything you do outside of Sift4 module will have a much bigger impact on the performance. For example, this is insanely slow: say @a.min({ sift4($_, $str, 5) }); And this is twice faster: my $ans = ‘default value’; my $ans-min = 6; for @A <https://github.com/a> { my $dist = sift4($ *, $str, 5, 5); if $dist < $ans-min { $ans = $*; $ans-min = $dist; } } One would think that in the code above ｢sift4($ *, $str, 5, 5)｣ can be changed to ｢sift4($*, $str, 5, $ans-min)｣, but no, for some reason it does not work. Whether it is an issue in this implementation or in the algorithm itself I don't know. Should be investigated I think. Anyway, to make it easier for the users I think this module should provide a function for processing an array (so that users are less likely to screw it up). Random testing suggests that these changes do not affect functionality. ------------------------------ You can view, comment on, or merge this pull request online at: #4 Commit Summary - Make it 13x faster by making the code unreadable File Changes - *M* lib/Text/Diff/Sift4.pm6 <https://github.com/MasterDuke17/Text-Diff-Sift4/pull/4/files#diff-0> (146) Patch Links: - https://github.com/MasterDuke17/Text-Diff-Sift4/pull/4.patch - https://github.com/MasterDuke17/Text-Diff-Sift4/pull/4.diff — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AJx8UQpiZSro_0GF_ASo46Z0U7OH0X43ks5rQyIagaJpZM4LfENx> .