Hi there,
Friday I am ready with my tournament.
56 engines played each one against each other 50 games = 2.750 for each engine = 1.540 pairings = 77.000 games.
I think the biggest computer chess tournament ever played. And all this with 45-minute games on strong and fast i7 4.0 Ghz hardware. It need a very long way to present such an exactly database with an over optimated opening book all 500 ECO codes were played. All games are checked for blunders ... and most populare opening Systems have of course the most games in the database. The database will be very interesting for further statistics and my playing style selection. Also I have interest to make some error bar experiments I discuss with Bob Hyatt for a while. For such experiments I need a strong database, better I need such a big tournament.
As next one I will start the contempt experiment with the realease version of Stockfish 7 KP x64. This experiment should give us a very clear answere to the contempt question. Later we can compare SF 7 with or without contempt.
After some test-runs I started on other PCs the last weeks I think that contempt 7-9 will give Stockfish a small booster from around 8.5 - 12.5 Elo. Sure that I produce such a result.
Biggest problem with Contempt:
The move average without resign mode will be higher. An to high contempt and the move average can go 15-20 moves higher with a lot of boring endgames. The secret of contempt is the combination / mix between small move average and more points with lesser fast draw games after opening books moves. Komodo is indeed very strong here because Komodo prodcued with Contempt 10 the same move average as Stockfish without contempt with 82 moves. The Komodo programmers do here a great job.
Later I will test again dev. versions of Stockfish with Contempt 7 if I can see that this setting will give me 8.5 - 12.5 Elo more I await ... again I am very sure here.
But at first (before I test a newer Stockfish dev. version) all the interesting updates I have on my ToDo list.
Statistics and playing styles I try to update on weekend. Database and opening book will be available Saturday in the night.
Have fun ... I am working and I am very proud that I can very soon Display a very exactly Rating List ... in my opinion with max. possible exactly results because more strong engines are not available. Only with more games I can make this Rating list more exactly ... but more games aren't needed because I added 56 different engines.
Best
Frank
FCP: SF 7 KP BMI2 x64 Contempt=7 ... announcement
Moderators: hgm, Rebel, chrisw
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: FCP: SF 24Apr2016 BMI2 x64 Contempt=10 ... changed
Hi there,
I will make the test later, to many engine updates to test for me at the moment.
Tomorrow in the night ... around 02:00 in 14 1/2 hours ...
Stockfish 24Apr2016 BMI2 x64 Contempt = 10 will be start at first.
Same Contempt setting Komodo 9.42 have as standard setting.
Later ...
Best
Frank
I will make the test later, to many engine updates to test for me at the moment.
Tomorrow in the night ... around 02:00 in 14 1/2 hours ...
Stockfish 24Apr2016 BMI2 x64 Contempt = 10 will be start at first.
Same Contempt setting Komodo 9.42 have as standard setting.
Later ...
Best
Frank
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: FCP: SF 24Apr2016 BMI2 x64 C10 ... still running!
Hi there,
as next Stockfish 24Apr2016 BMI2 x64 C10 (Contempt=10) is still running for the next 6 days.
If you like to follow:
http://www.amateurschach.de/ftptrigger/ ... 4-c10.html
Website will be updated in the night ... database, opening book and so one. Need time ...
But have fun with the still running test-run.
Maybe the new Number 1 with Contempt = 10 and the current dev. Version? We will see ...
Best
Frank
as next Stockfish 24Apr2016 BMI2 x64 C10 (Contempt=10) is still running for the next 6 days.
If you like to follow:
http://www.amateurschach.de/ftptrigger/ ... 4-c10.html
Website will be updated in the night ... database, opening book and so one. Need time ...
But have fun with the still running test-run.
Maybe the new Number 1 with Contempt = 10 and the current dev. Version? We will see ...
Best
Frank
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: FCP: Contempt works fine ... after 219 games!
Hi there,
only two games with draw undo move 50.
vs. Octochess after 40 and vs. Rhetoric after 45 moves.
And move average is only 81!!
Could be an interesting figth for place 1.
With my table each one against each other ...
All is very simple to follow.
Have a look on Score%
87.9% to Komodo 9.42
86.3% to the release Version of SF 7.
At the moment ...
85.2 after 223 of 2750 games!
Best
Frank
only two games with draw undo move 50.
vs. Octochess after 40 and vs. Rhetoric after 45 moves.
And move average is only 81!!
Could be an interesting figth for place 1.
With my table each one against each other ...
All is very simple to follow.
Have a look on Score%
87.9% to Komodo 9.42
86.3% to the release Version of SF 7.
At the moment ...
85.2 after 223 of 2750 games!
Code: Select all
# Player : Elo Games Score% won draw lost Points Draw% Error OppAvg OppE OppD
----------------------------------------------------------------------------------------------------------------------------
1 Komodo 9.42 x64 : 3192.91 2750 87.9 2122 593 35 2418.5 21.6 15.47 2807.86 10.25 55.0
2 Stockfish 7 KP BMI2 x64 : 3166.47 2750 86.3 2017 714 19 2374.0 26.0 14.29 2808.34 10.27 55.0
Frank
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: FCP: Great stats / games ... after 444 games!
Hi there,
with more games Elo goes higher ...
Now 87.19%
1. Three draw games undo move 50 only. Contempt = 10 works absolutely perfect.
2. A lot of very fast won games, much more as the older SF Version. Also vs. GullChess or other stronger opponents.
Fact = This one is the best Stockfish Version I ever test and after the first 444 games no criticism from my site on a high niveau. I see nothing what can be bad. So many wonderful games ...
For the first Elo calculation I have to wait ... to many x:0 results and Elo will be unclear. I think I can create the first calculation after 40% ... after round 20 of 50. Made no sense to do it before.
This is indeed a nice fight for place 1.
After all I can see ... I think SF have a big chance for place 1 in my Rating List.
Can't believe that after all I saw now SF will have a bad seria and Elo go down.
Best
Frank
PS: 9 bad opening lines I found in my book, games are replayed.
444 x 9 : 100 = 2% my FCP Live Book works more and more fantastic.
with more games Elo goes higher ...
Now 87.19%
1. Three draw games undo move 50 only. Contempt = 10 works absolutely perfect.
2. A lot of very fast won games, much more as the older SF Version. Also vs. GullChess or other stronger opponents.
Fact = This one is the best Stockfish Version I ever test and after the first 444 games no criticism from my site on a high niveau. I see nothing what can be bad. So many wonderful games ...
For the first Elo calculation I have to wait ... to many x:0 results and Elo will be unclear. I think I can create the first calculation after 40% ... after round 20 of 50. Made no sense to do it before.
This is indeed a nice fight for place 1.
After all I can see ... I think SF have a big chance for place 1 in my Rating List.
Can't believe that after all I saw now SF will have a bad seria and Elo go down.
Best
Frank
PS: 9 bad opening lines I found in my book, games are replayed.
444 x 9 : 100 = 2% my FCP Live Book works more and more fantastic.
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: After round 13/50 ... 3200.86 = +34,39 Elo *Nr 1*
And the Elo go higher ... seven to 0 results after 13 rounds!
I believe it can be clear over 89.0% ...
I believe it can be clear over 89.0% ...
Code: Select all
FCP Rating List (during a LIVE Test-Run)
----------------------------------------
Date : May 01st, 2016 (13:15)
Games : 74.965
Version : 4.09 Beta 1
Conditions : http://www.amateurschach.de/main/_conditions.htm
Generated with : Ordo 1.2.6 x64 (Ratings for chess and other games)
https://sites.google.com/site/gaviotachessengine/ordo
by Miguel A. Ballicora (Argentina)
**********************************************
Still running after round 13/50 = 26%
Stockfish 24Apr2016 BMI2 x64 C10 = +34.39 Elo
# Player : Elo Games Score% won draw lost Points Draw% Error OppAvg OppE OppD
----------------------------------------------------------------------------------------------------------------------------
1 Stockfish 24Apr2016 BMI2 x64 : 3200.86 715 88.3 552 158 5 631.0 22.1 28.77 2809.34 10.37 55.0
2 Stockfish 7 KP BMI2 x64 : 3166.47 2750 86.3 2017 714 19 2374.0 26.0 14.29 2808.34 10.27 55.0
****************************************************************************************************************************
# Player : Elo Games Score% won draw lost Points Draw% Error OppAvg OppE OppD
----------------------------------------------------------------------------------------------------------------------------
1 Stockfish 24Apr2016 BMI2 x64 : 3200.86 715 88.3 552 158 5 631.0 22.1 28.77 2809.34 10.37 55.0
2 Komodo 9.42 x64 : 3196.43 2713 88.5 2119 565 29 2401.5 20.8 14.95 2804.08 10.37 54.6
3 Houdini 4 STD B x64 : 3099.69 2713 82.0 1846 757 110 2224.5 27.9 13.01 2805.86 10.41 54.6
4 GullChess 3.0 BMI2 x64 : 3045.40 2713 77.4 1633 932 148 2099.0 34.4 11.68 2806.86 10.43 54.6
5 Fire 4 x64 : 3043.56 2713 77.2 1624 941 148 2094.5 34.7 11.80 2806.90 10.43 54.6
6 Equinox 3.30 x64 : 3001.95 2713 73.2 1456 1058 199 1985.0 39.0 10.55 2807.66 10.45 54.6
7 Fritz 15 x64 : 2997.59 2713 72.7 1455 1036 222 1973.0 38.2 11.01 2807.74 10.44 54.6
8 Critter 1.6a x64 : 2992.95 2713 72.2 1427 1066 220 1960.0 39.3 10.88 2807.83 10.45 54.6
9 Protector 1.9.0 x64 : 2954.64 2713 68.1 1283 1130 300 1848.0 41.7 10.66 2808.53 10.45 54.6
10 Andscacs 0.85 POP x64 : 2950.06 2713 67.6 1297 1074 342 1834.0 39.6 10.47 2808.62 10.45 54.6
11 Nirvanachess 2.2 POP x64 : 2939.17 2713 66.4 1232 1137 344 1800.5 41.9 10.57 2808.82 10.45 54.6
12 iCE 3.0 v658 POP x64 : 2927.88 2713 65.1 1218 1094 401 1765.0 40.3 10.58 2809.03 10.45 54.6
13 Chiron 2.0 x64 : 2915.18 2713 63.6 1186 1077 450 1724.5 39.7 10.29 2809.26 10.46 54.6
14 Hannibal 1.5 x64 : 2912.41 2713 63.2 1149 1133 431 1715.5 41.8 10.42 2809.31 10.46 54.6
15 Texel 1.05 x64 : 2909.31 2713 62.9 1141 1129 443 1705.5 41.6 9.98 2809.37 10.46 54.6
16 Naum 4.6 x64 : 2890.00 2713 60.5 1043 1198 472 1642.0 44.2 10.33 2809.73 10.46 54.6
17 SmarThink 1.80 AVX x64 : 2846.35 2713 55.1 943 1102 668 1494.0 40.6 9.72 2810.53 10.47 54.6
18 Fizbo 1.6 x64 : 2845.63 2713 55.0 861 1261 591 1491.5 46.5 10.01 2810.54 10.46 54.6
19 Senpai 1.0 SSE42 x64 : 2843.16 2713 54.7 893 1180 640 1483.0 43.5 9.99 2810.59 10.46 54.6
20 Hakkapeliitta 3.0 x64 : 2836.38 2713 53.8 915 1089 709 1459.5 40.1 9.98 2810.71 10.46 54.6
----------------------------------------------------------------------------------------------------------------------------
21 Hiarcs 14 WCSC w32 : 2830.76 2713 53.1 867 1146 700 1440.0 42.2 9.85 2810.82 10.47 54.6
22 Sjeng c't 2010 w32 : 2807.27 2713 50.1 804 1108 801 1358.0 40.8 9.85 2811.25 10.47 54.6
23 Vajolet2 2.1 POP x64 : 2800.41 2713 49.2 735 1198 780 1334.0 44.2 9.93 2811.38 10.46 54.6
24 Cheng 4.39 x64 : 2799.70 2713 49.1 757 1149 807 1331.5 42.4 9.72 2811.39 10.47 54.6
25 Shredder 12 x64 : 2797.55 2713 48.8 724 1200 789 1324.0 44.2 9.52 2811.43 10.47 54.6
26 Junior 13.3.00 x64 : 2794.39 2713 48.4 757 1112 844 1313.0 41.0 9.64 2811.49 10.47 54.6
27 Spike 1.4 Leiden w32 : 2784.58 2713 47.1 712 1133 868 1278.5 41.8 9.66 2811.67 10.47 54.6
28 DiscoCheck 5.2.1 x64 : 2782.43 2713 46.8 733 1076 904 1271.0 39.7 10.21 2811.71 10.46 54.6
29 Quazar 0.4 x64 : 2775.70 2713 46.0 653 1189 871 1247.5 43.8 10.21 2811.83 10.46 54.6
30 Booot 5.2.0 x64 : 2772.02 2713 45.5 687 1095 931 1234.5 40.4 9.83 2811.90 10.47 54.6
31 Deuterium 14.3.34.130 POP x64 : 2762.95 2713 44.3 591 1224 898 1203.0 45.1 10.20 2812.07 10.46 54.6
32 Arasan 18.3 POP x64 : 2758.21 2713 43.7 594 1185 934 1186.5 43.7 9.83 2812.15 10.47 54.6
33 Zappa Mexico II x64 : 2753.45 2713 43.1 572 1196 945 1170.0 44.1 10.33 2812.24 10.46 54.6
34 Spark 1.0 x64 : 2752.45 2713 43.0 586 1161 966 1166.5 42.8 9.50 2812.26 10.47 54.6
35 Doch 1.3.4 JA x64 : 2749.12 2713 42.6 567 1176 970 1155.0 43.3 9.75 2812.32 10.47 54.6
36 Thinker 5.4d Inert x64 : 2743.63 2713 41.9 558 1156 999 1136.0 42.6 10.05 2812.42 10.46 54.6
37 TogaII 280513 Intel w32 : 2736.35 2713 41.0 570 1082 1061 1111.0 39.9 10.25 2812.56 10.46 54.6
38 Tornado 5.0 SSE4 x64 : 2730.96 2713 40.3 569 1047 1097 1092.5 38.6 9.99 2812.66 10.46 54.6
39 Crafty 25.0.1 JA POP x64 : 2730.81 2713 40.3 543 1098 1072 1092.0 40.5 10.33 2812.66 10.46 54.6
----------------------------------------------------------------------------------------------------------------------------
40 Atlas 3.80 x64 : 2729.93 2713 40.1 546 1086 1081 1089.0 40.0 9.95 2812.68 10.46 54.6
41 Gaviota 1.0 AVX x64 : 2729.05 2713 40.0 551 1070 1092 1086.0 39.4 9.72 2812.69 10.47 54.6
42 EXchess 7.88b x64 : 2728.64 2713 40.0 553 1063 1097 1084.5 39.2 10.15 2812.70 10.46 54.6
43 Dirty 03NOV2015 POP x64 : 2726.58 2713 39.7 530 1095 1088 1077.5 40.4 10.00 2812.74 10.46 54.6
44 Nemo 1.01 Beta POP x64 : 2712.47 2713 38.0 485 1090 1138 1030.0 40.2 9.89 2813.00 10.47 54.6
45 Bobcat 7.1 x64 : 2709.69 2713 37.6 477 1087 1149 1020.5 40.1 10.25 2813.05 10.46 54.6
46 Murka 3 x64 : 2707.61 2713 37.4 430 1167 1116 1013.5 43.0 10.21 2813.09 10.46 54.6
47 Pedone 1.2 BMI2 x64 : 2700.85 2713 36.5 426 1130 1157 991.0 41.7 10.00 2813.21 10.46 54.6
48 Rodent II 0.8.7 x64 : 2686.73 2713 34.8 432 1025 1256 944.5 37.8 10.09 2813.47 10.46 54.6
49 DisasterArea 1.54 x64 : 2685.65 2713 34.7 371 1140 1202 941.0 42.0 10.24 2813.49 10.46 54.6
50 GNU Chess5 5.60 x64 : 2677.02 2713 33.7 388 1050 1275 913.0 38.7 10.38 2813.65 10.46 54.6
51 Scorpio 2.77 JA POP x64 : 2673.90 2713 33.3 415 976 1322 903.0 36.0 10.59 2813.71 10.45 54.6
52 Glaurung 2.2 JA x64 : 2662.45 2713 31.9 340 1053 1320 866.5 38.8 10.42 2813.92 10.46 54.6
53 Rhetoric 1.4.3 POP x64 : 2656.15 2713 31.2 346 1002 1365 847.0 36.9 10.48 2814.04 10.45 54.6
54 The Baron 3.29 x64 : 2651.52 2713 30.7 368 929 1416 832.5 34.2 10.98 2814.12 10.44 54.6
55 Octochess r5190 SSE4 x64 : 2643.20 2713 29.7 316 982 1415 807.0 36.2 10.77 2814.27 10.45 54.6
56 BugChess2 1.9 POP x64 : 2621.58 2713 27.4 299 887 1527 742.5 32.7 10.60 2814.67 10.45 54.6
----------------------------------------------------------------------------------------------------------------------------
White advantage = 35.64 +/- 0.99
Draw rate (equal opponents) = 48.75 % +/- 0.21-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: After round 27/50 ... 3191.94 = +25,47 Elo *Nr 2*
Code: Select all
FCP Rating List (during a LIVE Test-Run)
----------------------------------------
Date : May 03rd, 2016 (07:15)
Games : 75.735
Version : 4.09 Beta 2
Conditions : http://www.amateurschach.de/main/_conditions.htm
Generated with : Ordo 1.2.6 x64 (Ratings for chess and other games)
https://sites.google.com/site/gaviotachessengine/ordo
by Miguel A. Ballicora (Argentina)
**********************************************
Still running after round 27/50 = 54%
Stockfish 24Apr2016 BMI2 x64 C10 = +25.47 Elo
# Player : Elo Games Score% won draw lost Points Draw% Error OppAvg OppE OppD
----------------------------------------------------------------------------------------------------------------------------
2 Stockfish 24Apr2016 BMI2 x64 : 3191.94 1485 87.8 1132 343 10 1303.5 23.1 19.02 2808.94 10.29 55.0
2 Stockfish 7 KP BMI2 x64 : 3166.47 2750 86.3 2017 714 19 2374.0 26.0 14.29 2808.34 10.27 55.0
****************************************************************************************************************************
# Player : Elo Games Score% won draw lost Points Draw% Error OppAvg OppE OppD
----------------------------------------------------------------------------------------------------------------------------
1 Komodo 9.42 x64 : 3195.57 2727 88.3 2120 576 31 2408.0 21.1 14.87 2805.64 10.29 54.9
2 Stockfish 24Apr2016 BMI2 x64 : 3191.94 1485 87.8 1132 343 10 1303.5 23.1 19.02 2808.94 10.29 55.0
3 Houdini 4 STD B x64 : 3098.91 2727 81.7 1847 764 116 2229.0 28.0 12.87 2807.41 10.33 54.9
4 GullChess 3.0 BMI2 x64 : 3044.47 2727 77.1 1633 938 156 2102.0 34.4 11.41 2808.41 10.35 54.9
5 Fire 4 x64 : 3043.64 2727 77.0 1624 952 151 2100.0 34.9 11.49 2808.43 10.35 54.9
6 Equinox 3.30 x64 : 3001.15 2727 72.9 1456 1063 208 1987.5 39.0 11.15 2809.21 10.36 54.9
7 Fritz 15 x64 : 2997.35 2727 72.5 1455 1044 228 1977.0 38.3 10.83 2809.28 10.36 54.9
8 Critter 1.6a x64 : 2993.09 2727 72.1 1428 1074 225 1965.0 39.4 10.18 2809.35 10.38 54.9
9 Protector 1.9.0 x64 : 2954.26 2727 67.9 1284 1134 309 1851.0 41.6 10.42 2810.07 10.37 54.9
10 Andscacs 0.85 POP x64 : 2949.70 2727 67.4 1297 1080 350 1837.0 39.6 10.20 2810.15 10.38 54.9
11 Nirvanachess 2.2 POP x64 : 2938.86 2727 66.1 1232 1143 352 1803.5 41.9 10.61 2810.35 10.37 54.9
12 iCE 3.0 v658 POP x64 : 2927.29 2727 64.8 1218 1098 411 1767.0 40.3 10.43 2810.56 10.37 54.9
13 Chiron 2.0 x64 : 2915.11 2727 63.4 1186 1084 457 1728.0 39.8 10.01 2810.78 10.38 54.9
14 Hannibal 1.5 x64 : 2912.34 2727 63.0 1149 1140 438 1719.0 41.8 9.85 2810.83 10.38 54.9
15 Texel 1.05 x64 : 2908.79 2727 62.6 1141 1133 453 1707.5 41.5 10.18 2810.90 10.38 54.9
16 Naum 4.6 x64 : 2890.00 2727 60.3 1043 1205 479 1645.5 44.2 10.30 2811.24 10.37 54.9
17 SmarThink 1.80 AVX x64 : 2845.59 2727 54.8 943 1103 681 1494.5 40.4 10.01 2812.06 10.38 54.9
18 Fizbo 1.6 x64 : 2845.45 2727 54.8 861 1266 600 1494.0 46.4 9.76 2812.06 10.38 54.9
19 Senpai 1.0 SSE42 x64 : 2842.70 2727 54.4 893 1183 651 1484.5 43.4 9.88 2812.11 10.38 54.9
20 Hakkapeliitta 3.0 x64 : 2835.79 2727 53.6 915 1091 721 1460.5 40.0 10.04 2812.24 10.38 54.9
----------------------------------------------------------------------------------------------------------------------------
21 Hiarcs 14 WCSC w32 : 2830.32 2727 52.9 867 1149 711 1441.5 42.1 9.83 2812.34 10.38 54.9
22 Sjeng c't 2010 w32 : 2807.03 2727 49.9 804 1112 811 1360.0 40.8 9.81 2812.76 10.38 54.9
23 Vajolet2 2.1 POP x64 : 2800.04 2727 49.0 735 1201 791 1335.5 44.0 9.83 2812.89 10.38 54.9
24 Cheng 4.39 x64 : 2799.05 2727 48.8 757 1150 820 1332.0 42.2 9.83 2812.91 10.38 54.9
25 Shredder 12 x64 : 2797.33 2727 48.6 724 1204 799 1326.0 44.2 9.79 2812.94 10.38 54.9
26 Junior 13.3.00 x64 : 2793.61 2727 48.1 757 1112 858 1313.0 40.8 9.79 2813.01 10.38 54.9
27 Spike 1.4 Leiden w32 : 2784.24 2727 46.9 712 1136 879 1280.0 41.7 10.16 2813.18 10.38 54.9
28 DiscoCheck 5.2.1 x64 : 2781.95 2727 46.6 733 1078 916 1272.0 39.5 9.84 2813.22 10.38 54.9
29 Quazar 0.4 x64 : 2775.09 2727 45.8 653 1190 884 1248.0 43.6 10.10 2813.35 10.38 54.9
30 Booot 5.2.0 x64 : 2771.70 2727 45.3 687 1098 942 1236.0 40.3 9.74 2813.41 10.38 54.9
31 Deuterium 14.3.34.130 POP x64 : 2762.79 2727 44.2 591 1228 908 1205.0 45.0 10.35 2813.58 10.37 54.9
32 Arasan 18.3 POP x64 : 2758.35 2727 43.6 595 1189 943 1189.5 43.6 9.41 2813.66 10.39 54.9
33 Zappa Mexico II x64 : 2752.88 2727 42.9 572 1197 958 1170.5 43.9 9.64 2813.76 10.39 54.9
34 Spark 1.0 x64 : 2752.02 2727 42.8 586 1163 978 1167.5 42.6 10.04 2813.77 10.38 54.9
35 Doch 1.3.4 JA x64 : 2748.99 2727 42.4 567 1180 980 1157.0 43.3 10.13 2813.83 10.38 54.9
36 Thinker 5.4d Inert x64 : 2743.21 2727 41.7 558 1158 1011 1137.0 42.5 10.12 2813.94 10.38 54.9
37 TogaII 280513 Intel w32 : 2735.80 2727 40.8 570 1083 1074 1111.5 39.7 9.97 2814.07 10.38 54.9
38 Crafty 25.0.1 JA POP x64 : 2730.56 2727 40.1 543 1101 1083 1093.5 40.4 10.04 2814.17 10.38 54.9
39 Tornado 5.0 SSE4 x64 : 2730.42 2727 40.1 569 1048 1110 1093.0 38.4 9.96 2814.17 10.38 54.9
40 Atlas 3.80 x64 : 2729.39 2727 40.0 546 1087 1094 1089.5 39.9 10.04 2814.19 10.38 54.9
----------------------------------------------------------------------------------------------------------------------------
41 Gaviota 1.0 AVX x64 : 2728.66 2727 39.9 551 1072 1104 1087.0 39.3 9.78 2814.20 10.38 54.9
42 EXchess 7.88b x64 : 2728.10 2727 39.8 553 1064 1110 1085.0 39.0 10.22 2814.21 10.37 54.9
43 Dirty 03NOV2015 POP x64 : 2726.49 2727 39.6 530 1099 1098 1079.5 40.3 9.88 2814.24 10.38 54.9
44 Nemo 1.01 Beta POP x64 : 2712.10 2727 37.8 485 1092 1150 1031.0 40.0 9.66 2814.51 10.38 54.9
45 Bobcat 7.1 x64 : 2709.33 2727 37.5 477 1089 1161 1021.5 39.9 9.75 2814.56 10.38 54.9
46 Murka 3 x64 : 2707.39 2727 37.2 430 1170 1127 1015.0 42.9 10.12 2814.59 10.38 54.9
47 Pedone 1.2 BMI2 x64 : 2700.20 2727 36.3 426 1130 1171 991.0 41.4 10.07 2814.72 10.38 54.9
48 Rodent II 0.8.7 x64 : 2686.70 2727 34.7 432 1029 1266 946.5 37.7 10.14 2814.97 10.38 54.9
49 DisasterArea 1.54 x64 : 2685.17 2727 34.5 371 1141 1215 941.5 41.8 9.84 2815.00 10.38 54.9
50 GNU Chess5 5.60 x64 : 2676.38 2727 33.5 388 1050 1289 913.0 38.5 10.44 2815.16 10.37 54.9
51 Scorpio 2.77 JA POP x64 : 2673.27 2727 33.1 415 976 1336 903.0 35.8 10.66 2815.22 10.37 54.9
52 Glaurung 2.2 JA x64 : 2661.98 2727 31.8 340 1054 1333 867.0 38.7 10.30 2815.42 10.37 54.9
53 Rhetoric 1.4.3 POP x64 : 2655.69 2727 31.1 346 1003 1378 847.5 36.8 10.44 2815.54 10.37 54.9
54 The Baron 3.29 x64 : 2651.07 2727 30.5 368 930 1429 833.0 34.1 10.37 2815.62 10.37 54.9
55 Octochess r5190 SSE4 x64 : 2642.92 2727 29.6 316 984 1427 808.0 36.1 10.36 2815.77 10.37 54.9
56 BugChess2 1.9 POP x64 : 2621.33 2727 27.3 299 889 1539 743.5 32.6 10.82 2816.17 10.36 54.9
----------------------------------------------------------------------------------------------------------------------------
White advantage = 35.85 +/- 1.00
Draw rate (equal opponents) = 48.81 % +/- 0.21
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: First good comparsion ...
Hi there,
after I have each one against each one the move stats are available on my start site (menue system).
Under "Test-Run res., comparsion ... you can find after each test-run a nice information. Will give a small information to improvements of new tested engine version.
Much better as I do from time to time in the past. Means now for all tested engines in the future this information will be available.
At first to the still running Stockfish dev. version.
Best
Frank
after I have each one against each one the move stats are available on my start site (menue system).
Under "Test-Run res., comparsion ... you can find after each test-run a nice information. Will give a small information to improvements of new tested engine version.
Much better as I do from time to time in the past. Means now for all tested engines in the future this information will be available.
At first to the still running Stockfish dev. version.
Best
Frank
-
Tdunbug
- Posts: 47
- Joined: Mon Mar 07, 2016 12:46 am
Re: First good comparsion ...
Hi Frank thank for your interesting tests. By chance do you plan to run a test with K 9.42 with dynamisms set to 110. As I have found that it gives slightly better results in my small tests?
-
Frank Quisinsky
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: First good comparsion ...
Hi,
shortly in bad English.
The work I do is very boring (all need to long ... six days for one test-run ... to long) but I hope nevertheless interesting.
I have at the moment 28 cores ... 26 i7 cores and I am using for my rating system 14 cores. To expensive if I am using all cores. If I do that the red card from my girlfriend should be absolutely clear. At the moment I hold the light red card only in my hand.
With 14 cores one test-run with 2.750 games need around 6 days. 425-450 games per day can be play (if all works fine).
56 engines in my list ... around 25-30 engines updating from time to time by the programmers. The others are more or less ... older sparing engines.
12 months x 6 test-runs per month = 60 test-runs p.a.
You can see that I organice all p.a. ... means with 14 cores = 56 engines building max. test-constellation.
I think all the interesting setting ... Rodent personalities or other great ideas by programmers, like Komodo, Hakkapeliitta I can't test. Should be the work by beta testers. For that reason I give my files free for all.
A beta tester can used my databases, the engine configuration, and all other things (can be found in my download site) and can simulate all what I do ... can test special settings on own PCs.
Three people I know (one programmer) do that!
Your question made sense and is logical for me but I am chess mate with 56 engines and 14 cores with the time control I am using ... one game need around 45-minutes.
I can't do more, only better stats with the database I have. Here I can do more but statistics need time. In additional to talk big from time to time (I do often ... pssst, a secret) but the biggest problem ... the girlfriend problem you know ...
More isn't possible from my site in testing engines!
But I hope you have time to look on my site for additional information ... in combination with the work others do.
Have fun ...
Best
Frank
shortly in bad English.
The work I do is very boring (all need to long ... six days for one test-run ... to long) but I hope nevertheless interesting.
I have at the moment 28 cores ... 26 i7 cores and I am using for my rating system 14 cores. To expensive if I am using all cores. If I do that the red card from my girlfriend should be absolutely clear. At the moment I hold the light red card only in my hand.
With 14 cores one test-run with 2.750 games need around 6 days. 425-450 games per day can be play (if all works fine).
56 engines in my list ... around 25-30 engines updating from time to time by the programmers. The others are more or less ... older sparing engines.
12 months x 6 test-runs per month = 60 test-runs p.a.
You can see that I organice all p.a. ... means with 14 cores = 56 engines building max. test-constellation.
I think all the interesting setting ... Rodent personalities or other great ideas by programmers, like Komodo, Hakkapeliitta I can't test. Should be the work by beta testers. For that reason I give my files free for all.
A beta tester can used my databases, the engine configuration, and all other things (can be found in my download site) and can simulate all what I do ... can test special settings on own PCs.
Three people I know (one programmer) do that!
Your question made sense and is logical for me but I am chess mate with 56 engines and 14 cores with the time control I am using ... one game need around 45-minutes.
I can't do more, only better stats with the database I have. Here I can do more but statistics need time. In additional to talk big from time to time (I do often ... pssst, a secret) but the biggest problem ... the girlfriend problem you know ...
More isn't possible from my site in testing engines!
But I hope you have time to look on my site for additional information ... in combination with the work others do.
Have fun ...
Best
Frank