Talk:Merge algorithm: Difference between revisions

Revision as of 19:55, 29 November 2015

Revision as of 19:56, 29 November 2015

Line 102:

This section includes the statement: ''A k-way in-memory merge sort can also be constructed from a k-way merge algorithm, but his provides no benefits over the ordinary binary merge sort.'' On a processor with 16 registers, such as a PC in 64 bit mode, a 4-way merge sort using 10 or so pointers that a compiler optimizes by using registers for all those pointers will result in a 4-way merge sort being somewhat faster than a binary merge sort, and about as fast as quick sort. PC based benchmarks showed the 4-way merge sort was about 15% faster than binary merge sort. [[User:Rcgldr|Rcgldr]] ([[User talk:Rcgldr|talk]]) 19:19, 29 November 2015 (UTC)

:The comparison is between a {{mvar|k}}-way merge ''where {{mvar|k}} is variable'' and a binary merge. I made the statement more precise by specifying that the number of comparisons is being counted, but maybe the variable nature of {{mvar|k}} should be stressed more?. [[User:Qwertyus|Q<small>VVERTYVS</small>]] <small>([[User talk:Qwertyus|hm?]])</small> 19:55, 29 November 2015 (UTC)