Comments on: Efficient looping in Javascript

By: Solutoire.com › Javascripts Loop Benchmarks

Solutoire.com › Javascripts Loop Benchmarks — Tue, 29 Jul 2008 18:04:30 +0000

[...] Possibly related Efficient looping in Javascript → read this1 year, 5 months ago [...]

By: juraj

juraj — Thu, 24 May 2007 13:09:40 +0000

nice site thanks for doing the effort, I have a question for anyone I guess, why is the descending loop (the improved) more efficient?

I suspect it is at bit level manipulation, is decrementing a number faster than incrementing it?

bye

By: Bas

Bas — Thu, 29 Mar 2007 16:38:15 +0000

When looping over 100 records most tests ‘ll tell you it took 0ms…

By: tc

tc — Thu, 29 Mar 2007 16:34:06 +0000

Well I’ll take this into consideration the next time I’m looping over 100,000 records, but how much of a difference does it make for list of more realistic size. I mean on the web, I have never really looped over anything more than probably 100 records. Anyway… still interesting article.

By: Bas

Bas — Thu, 29 Mar 2007 14:46:50 +0000

I don’t think you read the article very well. It’s not a proof of concept, it’s a comparison between different ways to loop over an Array. You’re right about the optimizations, but I wanted to show the difference between normal for loops and ‘improved’ loops.
And about the live preview: yeah I know, otherwise this place would be heaven for Mr Evil…

By: SillyCone

SillyCone — Thu, 22 Mar 2007 11:41:43 +0000

Of course, the while loops can be replaced as easily by an optimized version. It’s just basic algorithmics : as the tests are repeated for every loop, make them as light as you can.
So it would be like :
var max = data.length; var i = 0; while (i < max) { tmp = data[i++]; }
Djee, your preview box doesn’t work as the posted version… We have to use ‘& lt ;’ instead of ‘<’.

By: SillyCone

SillyCone — Wed, 21 Mar 2007 15:45:47 +0000

Sorry about those 2 post trials (can delete them if you wish).

About your 2 native for loops, the second one (reverse iteration) is not an optimization at all, it’s actually the proof of poor algorithm (sorry to be so rude :).
I bet the only thing that slows down the first loop compared to the second is the fact that the second for argument (i < data.length) is run at every cycle of the loop, thus evaluating the length of your array each and everytime.
To “optimize” the first loop, just replace this :
for(var i = 0; i < data.length; i++){tmp = data[i];};
by this :
for(var max = data.length, i = 0 ; i < max; i++) { tmp = data[i]; }
It should do the trick, be as fast as the first one and not reversed. Haven’t tested it though, I must admit (not enough time in a day :)

Apart from that little point, it was a very usefull and instructive test, thanks a lot !

By: Spiros Denaxas

Spiros Denaxas — Wed, 14 Mar 2007 17:15:23 +0000

It is interesting to see that the reverse iteration loop is faster than the normal iteration loop. Out of curiosity, I did the same with Perl:

idaru:/tmp spiros$ ./loop.pl total elements: 500000 Normal loop : 0.269156 Reverse loop : 0.253984

500000 elements took 0.26ms to get traversed normally and 0.25ms to get traversed backwards.

The difference seems to get larger as the size of the array grows:
idaru:/tmp spiros$ ./loop.pl total elements: 700000 Normal loop : 0.379325 Reverse loop : 0.353184

Spiros

By: Stathis

Stathis — Wed, 14 Mar 2007 16:46:27 +0000

It seems that your blogging software has rendered “i minus minus” as a “i ndash” in your code (an ndash is a slightly longer version of a dash). In my browser it looks like you’re just writing “i minus” which can’t be right. Consider fixing this to avoid confusion :-)

By: Matt Kruse

Matt Kruse — Mon, 12 Feb 2007 18:37:31 +0000

I did some interesting tests on loop speeds here:
http://www.mattkruse.com/temp/gb_speed.html

The difference is most noticeable in IE. I use the lessons learned in these tests in my table sorting script, which I believe outperforms any other script performing the same functionality:
http://www.JavascriptToolbox.com/lib/table/