Howard Kaikow's Sort Performance Comparison program

You are visitor number since 29 February 2004

Date: 6 March 2004(Revision of 29 February 2004 version of this document)

Author: Howard Kaikow <kaikow@standards.com>

Subject: Sort Performance Comparison program - Data

Some of the sorting algorithms are very efficient, but a few, such as Bubble Sort, are relatively inefficient.

For example, on a Pentium II 400 with 768MB of memory, using randomly ordered Long data, using version 1.0.0 of the program, I've seen:

The program includes 4 implementations of the bubble sort algorithm. Using 3 000 data items, the time used was:

KB: 585 milliseconds

S6.4: 584 milliseconds

S6.4#: 309 milliseconds

RS: 287 milliseconds

The program includes 17 implementations of the quicksort algorithm. Using 40 000 000 data items, the time used was:

(HK): 92475 millisecond

(HK#): 101754 milliseconds

(HK##): 94883 milliseconds

(S7.1): 100256 milliseconds

(S7.1=): 99097 milliseconds

(S7.1#): 100499 milliseconds

(S7.1#=): 103899 milliseconds

(S7.4): 94073 milliseconds

(S7.4=): 97041 milliseconds

(S7.4#): 94023 milliseconds

(S7.4#=): 94004 milliseconds

(S7.5): 57879 milliseconds

(S7.5#): 57140 milliseconds

(S7.5#M): 60242 milliseconds

(S7.5#M9): 56843 milliseconds

(S7.5#M9=): 56453 milliseconds

(RS): (OUT_OF_STACK_SPACE)

There can be a significant difference in the time used by differing implementations of the same algorithm.