Author: Marc Lacrosse
Date: 06:41:31 11/22/05
I just ran a series of matches between Shredder 9 and Spike 1.0a on one hand and
various fruit flavors on the other hand.
Each match was a series of 100 games at 1 min + 1sec from 50 standard
startpositions (played as white and as black) without opening book.
The standard start positions that I used are a mix of Nunn and Noomen positions.
All GambitFruit flavors played with futility pruning and delta pruning enabled.
Tablebase were not activated for engines able to use them.
The games were played either on a Intel Centrino 2 Ghz with 2GB Ram or on a PIV
3.0Ghz with 1GB ram.
128 MB for each engine, ponder off. Winboard+Polyglot.
Here are the results of the various Fruits:
against Shredder9 against Spike1.0 TOTAL
Fruit 2.2.1 48.5 70.5 119.0
GambitFruit 1.0b2 46 69 115.0
GambitFruit 1.0b2 RR+ 43.5 67.5 111.0
Fruit 2.1 41.5 63 104.5
TogaII 1.1 43.5 60.5 104.0
GF 1.0b2.5 40 61.5 101.5
GF 1.0b3 (with .per file) 39 46 85.0
GF 1.0b3 (without .per file) 32 48.5 80.5
RR+ : = experimental Rebel Reduction feature turned on
Program Elo + - Games Score Draws
1 Fruit 2.2.1 : 2762 43 43 200 59.5 % 22.0 %
2 Shredder 9 UCI : 2762 21 21 800 58.2 % 23.0 %
3 GFruit 1.0beta2 : 2748 42 42 200 57.5 % 25.0 %
4 GFruit 1.0beta2 RR+ : 2734 43 43 200 55.5 % 22.0 %
5 Fruit 2.1 : 2711 41 41 200 52.2 % 28.5 %
6 TogaII 1.1 : 2711 44 44 200 52.3 % 18.1 %
7 GFruit10b25 : 2701 42 42 200 50.7 % 23.5 %
8 GFruit 1.0beta3 : 2643 41 41 200 42.5 % 28.0 %
9 Spike 1.0a : 2628 21 21 800 39.2 % 24.4 %
10 GFruit10b3noper : 2627 43 43 200 40.2 % 22.5 %
Fr S9 GF GF To Fr GF GF Sp GF
Fruit 2.2.1 50 65 84 96 96 98 99 99 99
Shredder 9 UCI 49 70 90 98 99 99 99100 99
GFruit 1.0beta2 34 29 72 91 92 95 99 99 99
GFruit 1.0beta2 RR+ 15 9 27 78 79 86 99 99 99
TogaII 1.1 3 1 8 21 51 62 98 99 99
Fruit 2.1 3 0 7 20 48 61 98 99 99
GFruit10b25 1 0 4 13 37 38 97 99 99
GFruit 1.0beta3 0 0 0 0 1 1 2 75 70
Spike 1.0a 0 0 0 0 0 0 0 24 52
GFruit10b3noper 0 0 0 0 0 0 0 29 47
So there is
- a 95% or higher probability that GambitFruit 1.0beta2 is stronger than 1.0b25
and 1.0b3.
- a 91% probability that 1.0b2 is stronger than Toga 1.1 in these test
conditions!
Marc Lacrosse
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.