Comments on: Sorting Algorithms: The Comb Sort http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/ What would OJ do? Wed, 17 Mar 2010 08:58:04 +0000 http://wordpress.org/?v=2.9.2 hourly 1 By: OJ http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1579 OJ Fri, 11 Dec 2009 19:05:35 +0000 http://buffered.io/?p=534#comment-1579 Hi Annie,<br><br>Thanks for your feedback! I'm glad you found the article helpful in your learning process.<br><br>Good luck with your work!<br>Cheers<br>OJ Hi Annie,

Thanks for your feedback! I'm glad you found the article helpful in your learning process.

Good luck with your work!
Cheers
OJ

]]> By: Annie http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1577 Annie Thu, 10 Dec 2009 15:42:10 +0000 http://buffered.io/?p=534#comment-1577 Hi OJ!<br>Thx a lot!!! I was required to do a assignment about this sort and I didn't even heard of this before. I did lots of research and I have to say your work helped me! Thx again. Hi OJ!
Thx a lot!!! I was required to do a assignment about this sort and I didn't even heard of this before. I did lots of research and I have to say your work helped me! Thx again.

]]> By: OJ http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1122 OJ Thu, 02 Oct 2008 00:04:09 +0000 http://buffered.io/?p=534#comment-1122 I'll definitely be covering esoteric and ridiculous sorts like the Bogosort, but they won't appear until the end of the series ;) I’ll definitely be covering esoteric and ridiculous sorts like the Bogosort, but they won’t appear until the end of the series ;)

]]> By: Keef http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1123 Keef Wed, 01 Oct 2008 16:53:12 +0000 http://buffered.io/?p=534#comment-1123 Would you (in this series) find it interesting to cover impractical sorting algorithms (such as the Bogosort) to further explain big O notation perhaps? <a href="http://en.wikipedia.org/wiki/Bogosort" rel="nofollow">http://en.wikipedia.org/wiki/Bogosort</a> It has an interesting thought experiment in that a quantum bogosort would be able to sort any sequence in O(n) time, but result in the destruction of n-1 universes... Would you (in this series) find it interesting to cover impractical sorting algorithms (such as the Bogosort) to further explain big O notation perhaps?

http://en.wikipedia.org/wiki/Bogosort

It has an interesting thought experiment in that a quantum bogosort would be able to sort any sequence in O(n) time, but result in the destruction of n-1 universes…

]]> By: OJ http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1121 OJ Mon, 29 Sep 2008 00:12:50 +0000 http://buffered.io/?p=534#comment-1121 @Vorlath: Thanks for the comment mate. Haven't seen you around for a while. I see you've had a bit of fun in the past playing around with sorting. By the end of this little series I do plan on getting a collection of datasets together and doing a few benchmarks myself. I'm not sure if I'll get to the point where I'll be writing ASM versions of something the algos, it depends on the time. Plus, most people won't see a great deal of value in an ASM implementation (most.. but not all ;) ). I was pleasantly surprised at the results of this particular algo. Crazy/scary to think that I didn't really know much about it until recently. Must continue reading! Thanks for the link! Cheers. @Vorlath: Thanks for the comment mate. Haven’t seen you around for a while. I see you’ve had a bit of fun in the past playing around with sorting. By the end of this little series I do plan on getting a collection of datasets together and doing a few benchmarks myself. I’m not sure if I’ll get to the point where I’ll be writing ASM versions of something the algos, it depends on the time. Plus, most people won’t see a great deal of value in an ASM implementation (most.. but not all ;) ).

I was pleasantly surprised at the results of this particular algo. Crazy/scary to think that I didn’t really know much about it until recently. Must continue reading!

Thanks for the link! Cheers.

]]> By: Vorlath http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1120 Vorlath Sat, 27 Sep 2008 00:04:08 +0000 http://buffered.io/?p=534#comment-1120 This is my favourite sorting algorithm.  If you use SSE on the PC, you can use min and max instructions which automatically compare and exchange values.  With each register holding 4 values, you can compare 16 values at a time.  With this, you use a factor of 1.6 for the gap.  I found it through trial and error.  But just a little off and it really degrades performance. The really cool thing is that if you use this with insertion sort, it's actually faster than quicksort when you have lots of random elements.  What isn't cool is that it takes the same amount of time no matter what, so it sucks for lists that are not random. Here's an article I wrote quite a while back trying to write a custom routine that was faster than anything out there.  Comb sort did the job. http://my.opera.com/Vorlath/blog/show.dml/94793 This is my favourite sorting algorithm.  If you use SSE on the PC, you can use min and max instructions which automatically compare and exchange values.  With each register holding 4 values, you can compare 16 values at a time.  With this, you use a factor of 1.6 for the gap.  I found it through trial and error.  But just a little off and it really degrades performance.
The really cool thing is that if you use this with insertion sort, it’s actually faster than quicksort when you have lots of random elements.  What isn’t cool is that it takes the same amount of time no matter what, so it sucks for lists that are not random.
Here’s an article I wrote quite a while back trying to write a custom routine that was faster than anything out there.  Comb sort did the job.
http://my.opera.com/Vorlath/blog/show.dml/94793

]]> By: OJ http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1113 OJ Mon, 15 Sep 2008 20:13:45 +0000 http://buffered.io/?p=534#comment-1113 Well, as I said, I don't know the reasons for the number, it's something that I'm trying to determine. I can understand where you're coming from, but using 1.25 is obviously going to result in different numbers compared to 1.3 when you start dealing with large datasets. I don't think that it was as simple as "the author rounded up". I reckon it'd be more a case of "the author tried rounding in both directions, but found that 1.3 yielded better results than 1.2". Of course, I am speculating wildly. Cheers :) Well, as I said, I don’t know the reasons for the number, it’s something that I’m trying to determine. I can understand where you’re coming from, but using 1.25 is obviously going to result in different numbers compared to 1.3 when you start dealing with large datasets.

I don’t think that it was as simple as “the author rounded up”. I reckon it’d be more a case of “the author tried rounding in both directions, but found that 1.3 yielded better results than 1.2″. Of course, I am speculating wildly.

Cheers :)

]]> By: gav http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1114 gav Mon, 15 Sep 2008 12:58:35 +0000 http://buffered.io/?p=534#comment-1114 on the subject of rounding, you (and the authors) state that 1.3 is the best value to use and wikipedia states that the acutal value is 1 / (1 - (1 / e^phi)), approx = 1.247330950103979.  So in the case of the magic value you round up, but the gap you round down. Seems a little odd to be rounding up the magic value, especially when you could just use 1.25. on the subject of rounding, you (and the authors) state that 1.3 is the best value to use and wikipedia states that the acutal value is 1 / (1 – (1 / e^phi)), approx = 1.247330950103979.  So in the case of the magic value you round up, but the gap you round down.

Seems a little odd to be rounding up the magic value, especially when you could just use 1.25.

]]> By: OJ http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1116 OJ Mon, 15 Sep 2008 11:02:48 +0000 http://buffered.io/?p=534#comment-1116 @Keef: That's a question I'm trying to answer myself mate. On the surface the whole idea makes sense to me, but I'm still unsure where the magic number comes from. I am trawling for some theory but not really turning up much. I'm kinda hoping that a math's wonk will come on here and tell us why :) If I find something I'll be sure to post it. If you happen to stumble on something please let me know :) Cheers! @Keef: That’s a question I’m trying to answer myself mate. On the surface the whole idea makes sense to me, but I’m still unsure where the magic number comes from. I am trawling for some theory but not really turning up much. I’m kinda hoping that a math’s wonk will come on here and tell us why :)
If I find something I’ll be sure to post it. If you happen to stumble on something please let me know :) Cheers!

]]> By: Keef http://buffered.io/2008/09/14/sorting-algorithms-the-comb-sort/comment-page-1/#comment-1115 Keef Mon, 15 Sep 2008 09:29:22 +0000 http://buffered.io/?p=534#comment-1115 Nice post, and not a bad algorithm either!  However, I find myself a little wary of where the 1.3 magic number comes from.  Is it just a case of "this seems to work well" or have they actually proved it's particularly efficient for a wide selection of datasets. Nice post, and not a bad algorithm either!  However, I find myself a little wary of where the 1.3 magic number comes from.  Is it just a case of “this seems to work well” or have they actually proved it’s particularly efficient for a wide selection of datasets.

]]>