advice on improving performance of this code

Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

Hello again,

Everyone was so helpful the other day in regards to my thread regarding performance revisions, I thought I would ask another.

I have the following code, and I'm curious if it can be made more efficient.

BLOCK 1
sub CONVERT 
{   
    # record->[1] is the key for the hash of arrays
    foreach my $record (@Data) {
    $record->[1] =~ s/\s+//g;
    $record->[1] = join ("_", $Map{$record->[1]}[1], $Map{$record->[1]
+}[0]);
    }
}

BLOCK 2
foreach my $record (@Data ) {
    for my $i (0 .. 4) {
        print DATA $record->[$i] . " ";
    }
    print DATA "\n";
    }
[download]

These are consuming the most time according to -d:SmallProf

Thanks again for the help

2006-03-28 Retitled by planetscape, as per Monastery guidelines
Original title: 'performance'

Comment on advice on improving performance of this code Download Code

Replies are listed 'Best First'.
Re: advice on improving performance of this code by gjb (Vicar) on Jan 03, 2003 at 21:43 UTC
Hi again ;-) In the first block I'd replace `$record->[1] = join ("_", $Map{$record->[1]}[1], $Map{$record->[1]}[0] +);` [download] with `$record->[1] = $Map{$record->[1]}[1] . '_' . $Map{$record->[1]}[1];` [download] In the second block I'd avoid the iteration by using join: `print DATA join(" ", @$record[0..4]), "\n";` [download] rather than `for my $i (0 .. 4) { print DATA $record->[$i] . " "; } print DATA "\n";` [download] or even `print DATA join(" ", @$_[0..4]), "\n" for (@Data);` [download] that does the whole block in one line. You can find the code I used to do the benchmarks on my scrachpad. Hope this helps, -gjb-	[reply] [d/l] [select]
Re: Re: advice on improving performance of this code by tachyon (Chancellor) on Jan 03, 2003 at 23:09 UTC
Update Upon testing $, tests 25% faster that join in the best case to 50% SLOWER in many cases! Presumably it relates to being faster to pass one argument to print (after the join) rather than many. Suprised me. Credit gjb for this pickup. I will be using join in future You can set $, to ' ' and ditch the join as well which will be faster again. Manipulating $, and $" can often useful in controlling output. local ensures no suprises elsewhere in the code. `{ local $, = ' '; print @{$_}[0..4], "\n" for @Data; }` [download] Note that setting $, and doing it in one line like this has the side effect of adding a ' ' after the last element and before the "\n". This can be rectified with a separate call to print for the "\n" if required. print() is actually quite slow so it may be faster to do this: `{ local $, = ' '; print @$_[0..3], $$_[4] ."\n" for @Data; }` [download] cheers tachyon s&&rsenoyhcatreve&&&s&n.+t&"$'$`$\"$\&"&ee&&y&srve&&d&&print	[reply] [d/l] [select]
Re: Re: advice on improving performance of this code by The_Rev (Acolyte) on Jan 04, 2003 at 02:14 UTC
I appreciate all the advice, and especially found gjb's suggestions helpful. I took a look at your scratchpad, and noticed a significant improvement over my old code. Thanks again!	[reply]
Re: advice on improving performance of this code by jdporter (Paladin) on Jan 03, 2003 at 21:35 UTC
That one line in block 1 could be written as: `$record->[1] = join '_', @{ $Map{ $record->[1] } }[1,0];` Block 2 could be written as: `local $, = ' '; local $\ = "\n"; foreach my $record ( @Data ) { print OUT @{ $record->[$i] }[0..4]; }` [download] But I'm not sure how much those will impact performance... (PS - I'm not so sure using `DATA` as an output filehandle is such a great idea...) jdporter The 6th Rule of Perl Club is -- There is no Rule #6.	[reply] [d/l] [select]
Re: advice on improving performance of this code by webengr (Pilgrim) on Jan 03, 2003 at 22:17 UTC
This is more of a question than a recommendation: would using "tr" instead of the regex substitution in the first block would be more efficient? PCS	[reply]
Re^2: advice on improving performance of this code by diotalevi (Canon) on Jan 03, 2003 at 22:50 UTC
Yes, but then you have to replace the '\s' with the equivalent escapes something like `tr/\t\r\f\n //d` to match all of what \s means. (assuming no locales or unicode) Fun Fun Fun in the Fluffy Chair	[reply] [d/l]