Re: Benchmark for Scala, Ruby and Perl

Hello Paul

Do you mean by undef $/ and with <$fh> we can read the file into memory 
at one time?
Yes that would be faster b/c we don't need to read file by each line, 
which increases the disk IO.

Another questions:
1. what's the "truss" command?
2. what's the syntax "<:mmap"?

Thank you.


On 15.01.2022 15:45, Paul Procacci wrote:
> Hey Jon,
> 
> The most glaringly obvious thing I could recommend is that at least in
> your perl routine (and probably the other languages) most of your time
> is context switching reading from the disk.
> 
> Now, my perl version is indeed faster, but one has to ask themselves,
> was .015193256 seconds really worth the effort?  /shrug   -- If this
> is for a financial industry perhaps, but then they'd just have written
> it in C.  Otherwise, probably not.
> Also note, there's other ways to speed this up even further, but at
> that point it isn't really worth the time.  We're talking a couple of
> microseconds at best.  I've included my version for your reference.
> 
> Before closing, I happen to like micro benchmarks whether or not you
> think 'I know this benchmark is maybe meaningless' as your site says.
> 
> If anything, it can absolutely be useful.  I personally think
> sometimes they are and others not so much.  Just depends on the
> context.
> 
> Your perl source (doit) .. my perl source (doit2):
> # ./doit2.pl [1] | md5
> 786be54356a5832dcd1148c18de71fc8
> root@nas:~ # ./doit.pl [2] | md5
> 786be54356a5832dcd1148c18de71fc8
> 
> # truss -c ./doit.pl [2]
> <!--snip-->
> 
> syscall                     seconds   calls  errors
> read                    0.036828813    4140       0
> 
> <!--snip-->
>                       ------------- ------- -------
>                         0.037821821    5227     284
> 
> # truss -c ./doit2.pl [1]
> <!--snip-->
> 
> syscall                     seconds   calls  errors
> read                    0.000245121      19       0
> <!--snip-->
>                       ------------- ------- -------
>                         0.022628565     804      59
> 
> -------------------------------------
> use strict;
> 
> $/ = undef;
> my %stopwords = do {
>         open my $fh, '<:mmap', 'stopwords.txt' or die $!;
>         map { $_ => 1; } split /\n/, <$fh>;
> };
> 
> my %count = do {
>         my %res;
>         open my $fh, '<:mmap', 'words.txt' or die $!;
>         map { $res{$_}++ unless $stopwords{$_}; } split /\n/, <$fh>;
>         %res;
> };
> 
> my $i=0;
> for (sort {$count{$b} <=> $count{$a}} keys %count) {
>     if ($i < 20) {
>         print "$_ -> $count{$_}\n"
>     } else {
>        last;
>     }
>     $i ++;
> }
> 
> On Sat, Jan 15, 2022 at 12:37 AM Jon Smart <[email protected]> wrote:
> 
>> Hello,
>> 
>> May I show the result of my benchmark for perl5, ruby, and scala?
>> https://blog.cloudcache.net/benchmark-for-scala-ruby-and-perl/
>> 
>> Welcome you to give any suggestion to me for improving this.
>> 
>> Thanks.
> 
> --
> __________________
> 
> :(){ :|:& };:
> 
> Links:
> ------
> [1] http://doit2.pl
> [2] http://doit.pl