From:

Date:

July 7, 2004 07:39Subject:

Re: optimizing Short circuiting sort method based on [M]Message ID:

200407071439.KAA27006@raptor.research.att.comdavid nicol asked for comments thus: > On Tue, 2004-06-29 at 18:24, Glenn Linderman wrote: > > > I'm not sure offhand how many result elements are at the crossover > > between where bubble sort is best and heapsort becomes best, but I > > believe both of those will generate entries in sorted order faster than > > sorting the whole array to throw away most of the results. > > That is what I was trying to figure out, and I would like peer review > on my result of bubble for five or less and heap for six or more. M=5 > is the judgement call, giving always two (after the unavoidable initial > comparison) in insertion mode whereas bubble gives an expected two with > random data. I meant to reply to the porters at large earlier, but succeeded in only contacting a couple people. Heap sort is going to double the number of comparisons realtive to a binary search, since it has to do two compares at each level of the log M tree. It saves moves, but moves are cheap, and comparisons are dear, so I don't see heap sort ever being the winner. You can get N log M comparisons from the binary search you have at http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2004-06/msg00592.html and N * M moves, which will beat the wholesale sorts for small enough M. But the wholesale merge sort does a nice job on orderly data, and will beat the specialized sort if there are fewer than log M runs, for example. I can imagine keeping track of where insertions happen, and biasing the initial guess-point for the binary search when it appears the data are non-random. That could make comparisons look more like N than like N log M for orderly data. -- jplThread Previous | Thread Next

- RE: optimizing Short circuiting sort method based on [M] by Orton, Yves
- Re: optimizing Short circuiting sort method based on [M] by Glenn Linderman
- Re: optimizing Short circuiting sort method based on [M] by John P. Linderman
- Re: optimizing Short circuiting sort method based on [M] by John P. Linderman
- Re: optimizing Short circuiting sort method based on [M] by Nicholas Clark
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- more 5.9 sort tests by david nicol
- Re: more 5.9 sort tests by Rafael Garcia-Suarez
- Re: more 5.9 sort tests by david nicol
- Re: more 5.9 sort tests by Nicholas Clark
- Re: more 5.9 sort tests by Rafael Garcia-Suarez
- Re: more 5.9 sort tests by Ronald J Kimball
- Re: more 5.9 sort tests (second draft) by david nicol
- Re: more 5.9 sort tests (second draft) by Rafael Garcia-Suarez
- Re: optimizing Short circuiting sort method based on [M] by Nicholas Clark
- Re: optimizing Short circuiting sort method based on [M] by Nicholas Clark
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Nicholas Clark
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Dave Mitchell
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Paul Fenwick
- Re: optimizing Short circuiting sort method based on [M] by Nicholas Clark
**Re: optimizing Short circuiting sort method based on [M]**by John P. Linderman- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Paul Fenwick
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- sort in scalar context, but broken due to inexperience. by David Nicol
- Re: sort in scalar context, but broken due to inexperience. by Dave Mitchell
- Re: sort in scalar context, but broken due to inexperience. by hv
- Re: sort in scalar context, but broken due to inexperience. by david nicol
- Re: sort in scalar context, but broken due to inexperience. by merlyn
- Re: sort in scalar context, but broken due to inexperience. by Rafael Garcia-Suarez
- [PATCH] sort plays nethack in a scalar context (was: sort in scalarcontext, but broken due to inexperience.) by Paul Fenwick
- Re: [PATCH] sort plays nethack in a scalar context (was: sort inscalar context, but broken due to inexperience.) by Rafael Garcia-Suarez
- Re: [PATCH] sort plays nethack in a scalar context by Paul Fenwick
- Re: [PATCH] sort plays nethack in a scalar context by Nicholas Clark
- Re: [PATCH] sort plays nethack in a scalar context by H.Merijn Brand
- Re: [PATCH] sort plays nethack in a scalar context by Nicholas Clark
- Re: [PATCH] sort plays nethack in a scalar context (was: sort in scalar context, but broken due to inexperience.) by merlyn
- Re: sort in scalar context, but broken due to inexperience. by Graham Barr
- Re: sort in scalar context, but broken due to inexperience. by Dave Mitchell
- Re: sort in scalar context, but broken due to inexperience. by Nick Ing-Simmons
- Short circuiting sort by David L. Nicol
- Re: Short circuiting sort by Glenn Linderman
- Re: Short circuiting sort by david nicol
- Re: sort in scalar context, but broken due to inexperience. by david nicol
- Re: sort in scalar context, but broken due to inexperience. by Graham Barr
- RE: [PATCH] sort plays nethack in a scalar context by Orton, Yves
- Re: [PATCH] sort plays nethack in a scalar context by Abigail
- Re: [PATCH] sort plays nethack in a scalar context by Rafael Garcia-Suarez
- Re: [PATCH] sort plays nethack in a scalar context by Ovid
- Re: [PATCH] sort plays nethack in a scalar context by Nicholas Clark
- Re: [PATCH] sort plays nethack in a scalar context by david nicol
- Re: Short circuiting sort by John P. Linderman
- Re: Short circuiting sort by Glenn Linderman
- Re: Short circuiting sort by david nicol
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Glenn Linderman
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Paul Fenwick
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Paul Fenwick
- Re: optimizing Short circuiting sort method based on [M] by Glenn Linderman
- Re: optimizing Short circuiting sort method based on [M] by Glenn Linderman
- Re: optimizing Short circuiting sort method based on [M] by perl5-porters
- Re: optimizing Short circuiting sort method based on [M] by Glenn Linderman
- Re: optimizing Short circuiting sort method based on [M] by Nicholas Clark
- Re: optimizing Short circuiting sort method based on [M] by Glenn Linderman
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Dave Mitchell
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by Dave Mitchell
- Re: optimizing Short circuiting sort method based on [M] by Paul Fenwick
- Re: optimizing Short circuiting sort method based on [M] by david nicol
- Re: optimizing Short circuiting sort method based on [M] by perl5-porters
- Re: optimizing Short circuiting sort method based on [M] by Glenn Linderman

nntp.perl.org: Perl Programming lists via nntp and http.

Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About