Re: [tablix-list] Scope for small (5%) speed improvement in Tablix

From: Tomaz Solc <tomaz.solcREMOVE@THISsiol.net>
Date: Sun Aug 08 2004 - 23:31:17 CEST

Hi

Just a small correction. The modified version of Tablix I benchmarked
yesterday included a change I didn't mention in my post. Besides "*map"
structures I also simplified the "chromo" structure:

struct chromo {
        int *time;
        int *room;
        int grade;
        int *subtotals;
        int possible;
};

I removed the "gen" structure and placed "time" and "room" fields
directly in "chromo" (this simplified mate_chromo() function and I
thought it made it faster)

Today I changed "chromo" structure back the way it was in 0.0.8 and
performed another test. The results were:

Unmodified Tablix 0.0.8: 974.0, 968.1, 973.0, 972.6
average: 971.9

Modified: 1012.0, 1012.0, 1015.1
average: 1013.0

This is around 5% faster than unmodified 0.0.8.

Best regards
Tomaz

Received on Sun Aug 08 23:52:35 2004

This archive was generated by hypermail 2.1.8 : Tue Aug 16 2005 - 20:42:18 CEST