Please check the general remarks!


2017.09.23 Andscacs 0.92 with a gain of 27 Elo from Version 0.90 AND 26 Elo from one IPON-RRRL to the next. 2017.09.21 Houdini 6 with a jump of 73 Elo in the full list and from one TOP 16 RRRL to the next! 39 Elo ahead of the next entry. I have to admit that this is much more than I would have expected. 2017.09.02 Komodo 11.2.2 increased 17 Elo in the full list and 20 Elo form one IPON RRRL to the next (see archive). Not to bad for a free upgrade! 2017.09.01 Ginkgo 2.0 included. 10 Elo plus over the predessesor. Over half a million PONDER ON games for the Full list now :-). 2017.08.30 Nirvanachess 2.4 added. Plus 37 Elo from the previous version - and a re-entry into the IPON-RRRL! And the IPON RRRL has a new number one as SF performed much better against this particular opponent (and it might change with the next engine again :-) ) 2017.06.02 Ginkgo 1.96 included. +11 Elo to its predecessor in the RRRL and 13 in the full list. 2017.05.28 New No. 8, 81 Elo from one RRRL to the next and 90 Elo in the full list! The new Booot 6.2 is really surprising, especially as it is the only Engine in my TOP16 written not in C/C++ but PASCAL! 2017.05.23 The new Komodo 11.01 gained 19 Elo in the IPON-RRRL, is the new leader and the first engine to pass 3300 Elo! In the full List it improved by 14 Elo but is still second. If you are interested please check the archive to compare the old TOP16 with the new TOP16. 2017.03.27 New release of Andscacs 0.90 included. 64 Elo from Version 0.88 to 0.90 in the full liist and 59 Elo from one TOP16 to the next one (see archive)! 2017.03.17 Komodo 10.4 with a 25 Elo jump in the complete list and a new No.2 in the Round Robin list - and have a look at the individual results you can download in the archive. :-) 2017.01.09 Chiron 4.0 included in both list. Very nice 65 Elo in the full list and 63 in the RRRL! It seems many engines are jumping into the TOP 10 recently and 3000 Elo seems to be "normal". 2017.01.03 Fizbo 1.9 was released on 2016.12.31 with an increase of 67 Elo in the RRRL, a new No .5 in this list. That is an excelent start in 2017! Lets see who the others perform over the year. 2016.12.23 I tried the SF8 version of ASM-fish (4th of November) to see how much beef there really is. At first I have to say that it is running rock stable with PONDER ON in a "networked" tourney! Not a single crash or time loss, not many enignes can do this that good! This is the results: Original modern SF8 compile: # PLAYER : RATING ERROR (%) D(%) OppAvg CFS(next) POINTS W D L PLAYED 1 Stockfish 8 : 3294 10 79.6% 37.5 3036 96 2626.5 2008 1237 55 3300 2 Houdini 5.01 : 3282 10 78.4% 36.2 3037 100 2587.0 1989 1196 115 3300 3 Komodo 10.2 : 3257 9 75.9% 35.8 3038 100 2503.5 1912 1183 205 3300 4 Shredder 13 : 3123 8 60.0% 49.8 3047 100 1981.5 1159 1645 496 3300 5 Gull 3 : 3065 8 52.4% 47.7 3051 100 1729.0 942 1574 784 3300 6 Ginkgo 1.9u : 3041 8 49.2% 50.1 3053 97 1622.0 796 1652 852 3300 7 Jonny 8.00 : 3030 8 47.6% 45.6 3053 100 1571.5 819 1505 976 3300 8 Equinox 3.30 : 3008 8 44.6% 49.1 3055 97 1472.5 662 1621 1017 3300 9 Fizbo 1.8 : 2997 8 43.1% 40.5 3056 67 1421.0 752 1338 1210 3300 10 Fritz 15 : 2994 8 42.7% 47.7 3056 56 1408.5 622 1573 1105 3300 11 Critter 1.6a : 2993 8 42.6% 47.4 3056 100 1404.5 622 1565 1113 3300 12 Andscacs 0.88 : 2966 8 39.0% 44.2 3058 63 1287.0 557 1460 1283 3300 13 Hannibal 1.7 : 2964 8 38.7% 45.6 3058 100 1278.0 525 1506 1269 3300 14 Booot 6.1 : 2946 8 36.3% 46.0 3059 85 1198.5 440 1517 1343 3300 15 Chiron 3.01 : 2940 8 35.5% 43.6 3059 90 1171.5 452 1439 1409 3300 16 Protector 1.9.0 : 2932 8 34.5% 43.8 3060 --- 1137.5 415 1445 1440 3300 asmFishW_2016-11-04_popcnt: # PLAYER : RATING ERROR (%) D(%) OppAvg CFS(next) POINTS W D L PLAYED 1 asm_11-04_SYZ : 3314 10 81.3% 34.6 3035 100 2682.5 2112 1141 47 3300 2 Houdini 5.01 : 3284 10 78.4% 36.2 3037 100 2588.0 1991 1194 115 3300 3 Komodo 10.2 : 3256 9 75.7% 35.4 3039 100 2497.0 1913 1168 219 3300 4 Shredder 13 : 3122 8 59.8% 49.6 3048 100 1974.5 1156 1637 507 3300 5 Gull 3 : 3064 8 52.2% 47.5 3052 100 1723.5 940 1567 793 3300 6 Ginkgo 1.9u : 3041 8 49.0% 49.8 3054 98 1618.0 797 1642 861 3300 7 Jonny 8.00 : 3029 8 47.4% 45.4 3054 100 1564.5 816 1497 987 3300 8 Equinox 3.30 : 3008 8 44.5% 49.2 3056 96 1470.0 659 1622 1019 3300 9 Fizbo 1.8 : 2998 8 43.2% 40.7 3057 72 1424.0 753 1342 1205 3300 10 Fritz 15 : 2994 8 42.7% 47.8 3057 57 1408.5 619 1579 1102 3300 11 Critter 1.6a : 2993 8 42.5% 47.4 3057 100 1403.5 622 1563 1115 3300 12 Andscacs 0.88 : 2965 8 38.8% 43.9 3059 60 1282.0 557 1450 1293 3300 13 Hannibal 1.7 : 2964 8 38.6% 45.4 3059 100 1275.0 526 1498 1276 3300 14 Booot 6.1 : 2945 8 36.2% 45.7 3060 84 1194.5 441 1507 1352 3300 15 Chiron 3.01 : 2939 9 35.4% 43.5 3060 95 1168.5 451 1435 1414 3300 16 Protector 1.9.0 : 2929 8 34.1% 43.1 3061 --- 1126.0 415 1422 1463 3300 That is a gain of 20 Elo over the original (only +15 with Bayeselo, and less with better SF8 compiles) For a while I was tempted to include it in my list, but then I realized that there are faster compiles than the original SF8 as well, which means that the difference is smaller in reality. Then I realized that this is basicaly "just" another "compile" of the same thing without any chess related improvement - I just have to add ~20 Elo for future asmFishes. That will save a lot of money and is better for the enviroment :-) I decided to stick to my principle of orginality because "AsmFish needs a host to benefit" and can't live on its own! 2016.12.12 Ginkgo 1.9u included. 5 Elo gain from old RRRL to ne RRRL, see Archive. 2016.12.05 I played 12 long games of SF8, H5.01 and K10.2. 6x 60m + 20s (12 cores), 3x 90m+30s (16 cores) from starting positions (no book). The Ponder hit rate is as follows: A. Players list: 1 Houdini 5.01 2 Komodo 10.2 3 Stockfish 8 B. Ponder hit and miss table: nr player : hit miss hit% games pts% 1 Stockfish 8 : 254 103 71.1 7 50.0 2 Komodo 10.2 : 214 87 71.1 6 33.3 3 Houdini 5.01 : 159 84 65.4 5 70.0 Average hit % : 69.6 File : 12LONG2L.pgn Date : 2016-12-05 08:25:43 Elapsed (h:m:s): 00:00:09 69.9%. With a slightly higher ponder hit (72.6) rate it seems possible that one gets better games with 4 cores Ponder ON than with 8 Cores Ponder off! 2016.11.30 Ferdinand Mosca was so kind to write a programm to analyse my database for Ponder hit rates! This is the result: A. Players list: 1 Andscacs 0.88 2 Booot 6.1 3 Chiron 3.01 4 Critter 1.6a 5 Equinox 3.30 6 Fizbo 1.8 7 Fritz 15 8 Ginkgo 1.8 9 Gull 3 10 Hannibal 1.7 11 Houdini 5 12 Jonny 8.00 13 Komodo 10.2 14 Protector 1.9.0 15 Shredder 13 16 Stockfish 8 B. Ponder hit and miss table: nr player : hit miss hit% games pts% 1 Komodo 10.2 : 107261 72022 59.8 3300 52.1 2 Houdini 5 : 101643 72338 58.4 3300 52.1 3 Shredder 13 : 105672 76146 58.1 3300 52.1 4 Fritz 15 : 113329 81693 58.1 3300 52.1 5 Critter 1.6a : 113195 81988 58.0 3300 52.1 6 Stockfish 8 : 91870 66839 57.9 3300 52.1 7 Protector 1.9.0 : 104780 76961 57.7 3300 52.1 8 Chiron 3.01 : 103214 76467 57.4 3300 52.1 9 Equinox 3.30 : 108174 80217 57.4 3300 52.1 10 Booot 6.1 : 109026 81087 57.3 3300 52.1 11 Ginkgo 1.8 : 107277 80097 57.3 3300 52.1 12 Andscacs 0.88 : 114635 86850 56.9 3300 52.1 13 Gull 3 : 114069 88119 56.4 3300 52.1 14 Jonny 8.00 : 103241 82396 55.6 3300 52.1 15 Fizbo 1.8 : 108686 88565 55.1 3300 52.1 16 Hannibal 1.7 : 104148 85945 54.8 3300 52.1 Average hit % : 57.2 File : TOP16L.pgn Date : 2016-11-30 11:47:33 Elapsed (sec) : 6435.2 This '57.2%' mean, that is is more efficient to play PONDER ON games as soon as you CPU has 16 cores. If you play matches between top engines on longer time controls, which means a higher draw rate, is is very likely that this cutoff is below 16 cores. Here are a few hit rates and their cutoff cpu number: The effective speedup for many cores for SF8 (Lazy SMP) is: Effective Speed-Up = 1 / (1 - 0.955 + 0.955/n_cores) and hit rate calculates by speedup at n cores / speedup at (n/2) Core Speedup Hit rate 2 1,91 1,91 4 3,52 1,84 6 4,90 1,78 8 6,08 1,73 10 7,12 1,68 12 8,03 1,64 14 8,83 1,60 16 9,55 1,57 18 10,20 1,54 20 10,78 1,51 22 11,31 1,49 24 11,79 1,47 26 12,24 1,45 28 12,64 1,43 30 13,02 1,41 32 13,36 1,40 Assuming you have a Ponder hit rate of 57% it is better to run 2 Engines with 8 cores Ponder ON on a 16 core CPU. With a Ponder hit rate of 64% you get the better games on 12 core CPUs already by playing Ponder ON with 2x6cores not Ponder OFF ... Data sources: http://www.talkchess.com/forum/viewtopic.php?topic_view=threads&p=694478&t=62146 Please read the whole thread. I will keep that updated on the archive for future reference. Short update on 2017.02.05 because this is mentioned in a CSS topic: 1. The 57.2% is at 5+3 with a wide range of opponents. Opponents with a closer Elo distance will have higher Ponder Hit Rates, the same happens for games with longer time controls! If you combine long time control games with equally strong engines, at least with todays top engines, I would not be supprised about Ponder Hit rates over 70%! 2. The 57.2% is with my conditions. I stop games when the loosing side evaluates its own position with -5 for 3 consecutive moves. If one side sees a mate the game is played to the end (Classic GUI Etiquette setting). If someone insists in playing games until the "bitter" end (for whatever reason!?) the draw rate will increase even more, as most endgames are evaluated very equally by most engines. 2016.11.10 Houdini 5 released and integrated into both list. 165 Elo+ over H4 in the full list and 179 from one TOP16 to the next! That is a new No2 in both lists. Now we have some kind of three-way fight :-) Looking back the last 2 weeks the next big thing should be released in 4 days :-D 2016.11.06 Booot 6.1 included into both list. Very remarkable 84 Elo increase since 6.0.2. Even more remarkable as it proves that a top engine does not have to be developed in C/C++! In this case it is an engine by Alex Morozov in Pascal (Delphi) ! 2016.11.03 Stockfish 8 released after 10 month development. Very surprising 67 Elo jump in the full list and 69 from one Top16 to the next! That is more than I would have expected and is asking for an earlier release next time!? :-) Congratulations to a new No.1 spot with a convienient margin to the No.2 in all my lists. In case someone is interested in the statisticaly irrelevant individual results -> ARCHIVE ! 2016.10.31 Komodo 10.2 and Andscacs 0.88(r) included in both lists. Komodo with a 17 Elo win over 10.1 in the full list and 20 Elo from one RRRL to the next one. Andscacs with 4 Elo less in the full list but 4 Elo more from one Top 16 to the next. I had similar close calls in the past and stick with the newer version. See archive to compare older TOP 16 list and to download the individual stats. 2016.10.30 Shredder 13 released and included in both lists. All new engine with an elo increase of 334. New No. 3 in the IPON-RRRL. (More details after the current test run, TBD = Shredder 13). Originaly I mentioned a time frame from the last release. It seems I was wrong because I have different releases of S12. Chessprogramming Wiki mentions a S12 release in Jan 2010! 2016.08.29 A new entry and a new No.10 for the RRRL: Fizbo 1.8! (+65 Elo to Version 1.7). No. 6 to 16 with only 92 Elo Elo spread. That is a nice field to watch in the weeks to come :-) 2016.08.10 Hannibal 1.7 included in both lists. A jump of 105 Elo from Version 1.4a. A very convincing re-entry into the IPON-RRRL as a new Number 10. This version will stay in for a while! Interesting fact: The lower 5 engines are only 22 Elo apart and the lower 10 less than 100 Elo! 2016.08.02 Nirvanachess 2.3 included. 16 Elo increase in the full list, 19 in the RRRL. 2016.07.23 Komodo 10.1 with a plus of 8 Elo in the full list and 11 Elo from one RRRL to the next. This version would not be tested because it is to close to the last release but as it is the reigning world champion it should be integrated. 2016.07.21 New Ginkgo 1.8 included in the full list and the IPON-RRRL. It gained 25 Elo from the previous version and is getting close to Gull! Nice competition up there :-) 2016.07.17 Booot 6.0.2 included. Unfortunately it missed the No. 16 spot by 5 Elo. Nonetheless a 133 Elo jump from Booot 5.2.0! 2016.07.14 Texel 1.06 gained 29 Elo for the complete list with this latest release. 30 Elo jump in the IPON-RRRL. 2016.07.06 Jonny 8.00 with an increase of 130 Elo over J7.01! This is a new No. 5 in the IRRRL! With this speed Jonny will pass everyone next year ;-). Impressive increase, amazing result. 2016.06.22 Andscacs 0.871 added to both lists. 10 Elo plus in the complete list, 7 Elo in the RRRL (see Archive). 2016.06.20 To conclude my "contempt" test I run the same set of opponents against Komodo 10 with a contempt of 0 1 Komodo 10 C0 : 3241 10 79.8% 34.4 2982 94 2633.0 2065 1136 99 3300 2 Stockfish 7 : 3229 11 78.7% 36.9 2983 100 2595.5 1986 1219 95 3300 3 Houdini 4 : 3120 9 66.6% 37.0 2990 100 2196.5 1586 1221 493 3300 4 Gull 3 : 3068 9 59.9% 45.4 2994 100 1976.0 1227 1498 575 3300 5 Ginkgo 1.7 : 3021 8 53.5% 49.7 2997 100 1766.5 946 1641 713 3300 6 Equinox 3.30 : 3001 8 50.8% 50.3 2998 60 1676.5 847 1659 794 3300 7 Fritz 15 : 3000 8 50.6% 49.6 2998 87 1669.5 851 1637 812 3300 8 Critter 1.6a : 2993 8 49.7% 49.1 2999 100 1639.5 829 1621 850 3300 9 Andscacs 0.86 : 2963 8 45.5% 45.8 3001 100 1502.5 746 1513 1041 3300 10 Protector 1.9.0 : 2937 9 42.0% 47.4 3003 56 1387.5 605 1565 1130 3300 11 Chiron 3 : 2936 8 41.9% 45.5 3003 100 1383.0 632 1502 1166 3300 12 Nirvanachess 2.2 : 2918 8 39.5% 46.3 3004 99 1302.5 538 1529 1233 3300 13 iCE 3.0 : 2903 8 37.5% 42.5 3005 88 1238.0 537 1402 1361 3300 14 Jonny 7.01 : 2896 8 36.6% 40.8 3005 87 1207.0 533 1348 1419 3300 15 Texel 1.05 : 2889 9 35.7% 40.1 3006 100 1177.0 516 1322 1462 3300 16 Naum 4.6 : 2859 9 31.8% 40.9 3008 --- 1049.5 375 1349 1576 3300 and this is the original ranking with default settings: 1 Komodo 10 : 3248 10 80.4% 30.7 2983 97 2653.0 2147 1012 141 3300 2 Stockfish 7 : 3234 10 79.1% 36.7 2984 100 2609.0 2003 1212 85 3300 3 Houdini 4 : 3119 9 66.3% 36.2 2991 100 2186.5 1589 1195 516 3300 4 Gull 3 : 3066 8 59.5% 44.9 2995 100 1963.5 1222 1483 595 3300 5 Ginkgo 1.7 : 3021 8 53.5% 49.4 2998 100 1764.5 950 1629 721 3300 6 Equinox 3.30 : 3003 8 50.9% 50.2 2999 68 1681.0 853 1656 791 3300 7 Fritz 15 : 3000 8 50.5% 49.5 2999 87 1668.0 852 1632 816 3300 8 Critter 1.6a : 2993 8 49.6% 48.8 3000 100 1636.5 831 1611 858 3300 9 Andscacs 0.86 : 2963 8 45.5% 45.7 3002 100 1501.5 748 1507 1045 3300 10 Protector 1.9.0 : 2937 8 41.9% 47.5 3004 59 1384.0 601 1566 1133 3300 11 Chiron 3 : 2936 8 41.7% 45.0 3004 100 1377.5 635 1485 1180 3300 12 Nirvanachess 2.2 : 2917 8 39.2% 45.8 3005 97 1294.0 538 1512 1250 3300 13 iCE 3.0 : 2906 8 37.7% 42.5 3006 88 1245.0 544 1402 1354 3300 14 Jonny 7.01 : 2898 8 36.8% 40.9 3006 94 1213.5 538 1351 1411 3300 15 Texel 1.05 : 2889 9 35.5% 39.8 3007 100 1172.5 516 1313 1471 3300 16 Naum 4.6 : 2860 8 31.8% 40.8 3009 --- 1050.0 376 1348 1576 3300 Again not a big change (performance -0.6%). It might be only statistical noise (see previous test) but for sure the effect of the default contempt is neglectable. (Distance to the second best is 14 Elo with default settings and 12 Elo with contempt 0) I dont want to emphasise the individual results because it is just 220 games vs each engine, nonetheless this is intereting: K10 vs Stockfish 7 : 220 ( 35, 135, 50), 46.6 K10C0 vs Stockfish 7 : 220 ( 45, 142, 33), 52.7 That is a difference of 13.5 points because of contempt. One time K10 is loosing and the other run is easily winning ... 2016.06.18 I was interested how Stockfish would perform for the IPON if a contempt is used. I played a full round against all opponents with C=20. This is the result: 1 Komodo 10 : 3253 10 80.7% 30.5 2984 98 2663.0 2159 1008 133 3300 2 Stockfish 7 C20 : 3238 10 79.3% 33.0 2985 100 2616.0 2071 1090 139 3300 3 Houdini 4 : 3121 9 66.3% 35.7 2992 100 2189.5 1600 1179 521 3300 4 Gull 3 : 3068 8 59.6% 44.8 2996 100 1965.5 1227 1477 596 3300 5 Ginkgo 1.7 : 3020 8 53.2% 48.6 2999 100 1755.5 954 1603 743 3300 6 Equinox 3.30 : 3004 8 51.0% 50.1 3000 65 1683.0 856 1654 790 3300 7 Fritz 15 : 3002 8 50.7% 49.5 3000 93 1672.0 855 1634 811 3300 8 Critter 1.6a : 2993 8 49.5% 48.5 3001 100 1632.5 833 1599 868 3300 9 Andscacs 0.86 : 2964 8 45.4% 45.4 3003 100 1499.0 750 1498 1052 3300 10 Protector 1.9.0 : 2937 9 41.8% 47.0 3005 56 1381.0 605 1552 1143 3300 11 Chiron 3 : 2936 8 41.7% 44.9 3005 100 1376.5 635 1483 1182 3300 12 Nirvanachess 2.2 : 2916 9 39.0% 45.4 3006 94 1286.5 538 1497 1265 3300 13 iCE 3.0 : 2906 8 37.7% 42.2 3007 78 1243.0 546 1394 1360 3300 14 Jonny 7.01 : 2901 8 37.0% 41.2 3007 98 1222.5 542 1361 1397 3300 15 Texel 1.05 : 2889 9 35.4% 39.5 3008 100 1169.0 517 1304 1479 3300 16 Naum 4.6 : 2859 9 31.7% 40.5 3010 --- 1045.5 377 1337 1586 3300 And this was the original SF7: 1 Komodo 10 : 3248 10 80.4% 30.7 2983 97 2653.0 2147 1012 141 3300 2 Stockfish 7 : 3234 10 79.1% 36.7 2984 100 2609.0 2003 1212 85 3300 3 Houdini 4 : 3119 9 66.3% 36.2 2991 100 2186.5 1589 1195 516 3300 4 Gull 3 : 3066 8 59.5% 44.9 2995 100 1963.5 1222 1483 595 3300 5 Ginkgo 1.7 : 3021 8 53.5% 49.4 2998 100 1764.5 950 1629 721 3300 6 Equinox 3.30 : 3003 8 50.9% 50.2 2999 68 1681.0 853 1656 791 3300 7 Fritz 15 : 3000 8 50.5% 49.5 2999 87 1668.0 852 1632 816 3300 8 Critter 1.6a : 2993 8 49.6% 48.8 3000 100 1636.5 831 1611 858 3300 9 Andscacs 0.86 : 2963 8 45.5% 45.7 3002 100 1501.5 748 1507 1045 3300 10 Protector 1.9.0 : 2937 8 41.9% 47.5 3004 59 1384.0 601 1566 1133 3300 11 Chiron 3 : 2936 8 41.7% 45.0 3004 100 1377.5 635 1485 1180 3300 12 Nirvanachess 2.2 : 2917 8 39.2% 45.8 3005 97 1294.0 538 1512 1250 3300 13 iCE 3.0 : 2906 8 37.7% 42.5 3006 88 1245.0 544 1402 1354 3300 14 Jonny 7.01 : 2898 8 36.8% 40.9 3006 94 1213.5 538 1351 1411 3300 15 Texel 1.05 : 2889 9 35.5% 39.8 3007 100 1172.5 516 1313 1471 3300 16 Naum 4.6 : 2860 8 31.8% 40.8 3009 --- 1050.0 376 1348 1576 3300 Stockfish with contempt gained 'just' 4 Elo, which is well within my errorbar (79.3% vs 79.1%, actually that percentage is the most interesting number). Quite remarkably K10 gained 5 Elo with only 220 different games compared to the full list. All this could be just statistical noise but I have the feeling that contempt as it is implmented in SF doesn't work very well. (Comparing the +/=/- between the two SFs is interesting too) In April this year I tested Fizbo 1.7. It performed quite good, it even would be a new No.12 in the IPON RRRL. Unfortunately it has 67 losses on time in 3300 games. Which is over 2% of all games. The problem is, that with such an error rate I would have to carry that problem with me if I would include it in the IPON-RRRL. I decided to integrate it with the lost games in the full list. You can find it today on rank 52, just behind Protector 1.8.0. I am crossing fingers that the next release will be more reliable - it would be a great entry for the main list! 2016.06.17 I run a little experiment with "asmFish 2016.06.16". After 1718 games it is at 79.8% vs my set of opponents. That is just 0.7% over SF7 and 0.7% is within my error bar. Biggest problem is that it is crashing very often under IPON conditions. The usual suspect here is a problem with pondering so I stopped the test as it is about as good as SF7 and will not pass K10. 2016.05.30 Komodo 10 added to all lists. 62 Elo from the last full release. 10 Elo from 9.42 in the full list. However, from one RRRL to the next, with the same opponents, it gained 12 Elo (see archive lists). Full individual results is available in the archive section. 2016.05.06 Chiron 3 added. 53 Elo jump for the latest release! Very nice. 2016.04.11 Ginkgo 1.7 included into both lists. 18 elo above Ginkgo 1.5 2016.03.26 Andscacs 0.86 added to the lists. A new entry to the Top 10 now! Nice jump of 51 Elo within 3 monthes (from 0.84 to 0.86) 2016.03.24 Komodo 9.42, a bug fix of K9.4, included in both lists. Komodo 9.42 is now 51 Elo over the initial 9.0 release. Komodo 9.4 is removed as the release is too close to 9.42. 2016.03.19 Komodo 9.4 is the leader in both IPON lists again. Now 58 Elo over the initial 9.0 release and worth a full version jump! 2016.01.08 iCE 3 is a new strong entry for the IPON-RRRL and with 2909 Elo a new no. 12! 2016.01.04 Stockfish 7 added to both lists. +58 Elo from SF6! That is much more than I would have expected. Outstanding! You can download the current individual statistics in the archive section. 2015.12.10 Andscacs 0.84 included in the list. 73 Elo plus to the 0.82 version. No. 11 in the IPON-RRRL. 2015.12.02 Jonny 7.01 included. +102 Elo from Jonny 6.00 and a new No.11 for the IPON RRRL. 2015.11.27 Ginkgo 1.5 added. Moderate 17 Elo added, but good enough to be the new No. 5 in the RRRL. 2015.11.26 I added a new rule: 5. Engines with a high failure rate will be eliminated by new entries prior to the last entry. Many engines have problems when being tested automaticaly with my conditions. As this is causing to much work to repeat games and sometimes stopping the whole automatic run these engines will either be eliminated or not even included in the IPON-RRRL. Nonetheless, new versions of an excluded engine will be included again if the problem is solved. 2015.11.20 Fritz 15 added to both lists. Remarkable 103 Elo to Fritz 14 (predecessor). 35 Elo to Rybka 4.1 (same author). New No. 5 in the IPON-RRRL! 2015.10.23 Protector 1.9.0 included in the full list. 9 Elo gain over its predecessor. 2015.10.20 Nirvanachess 2.2 and Andscacs 0.82 added to the complete list. Andscacs 0.82 is a nice new entry with 2842 Elo and Nirvanachess 2.2 is 35 Elo over its predecessor. Very nice as well. The entry to the IPON RRRL will follow at a later date. 2015.09.18 Komodo 9.2 included. Impressive 27 Elo increase from 9.01. Especially for a (nearly) free update! Komodo 9.2 is the first engine to crash the 3200 Elo barrier on the IPON lists! 2015.09.16 Ginkgo 1.3 added to the list. New entry and already No.7 at the IPON-RRRL! The ranks 5, 6 and 7 are very close! 2015.06.01 Protector 1.8.0 included. 50 Elo increase over the last release. Nice! The competiton gets closer :-) 2015.05.20 Nirvanachess 2.1c with a 56 Elo jump included in the list. Quite an achievement! 2015.04.28 After just 3 monthes both IPON lists have a new No.1! Komodo 9 could replace SF6 with an impressive increase of 50 Elo over K8! Right now there is a real battle going on to get the first place. Computer chess wasn't that exciting for years! The last column in the rating lists is new. It is the "Confidence For Superiority" to the next player in the list. Useful to check if an engine is really better :-) - thanks to Miguel Ballicora for his wonderful program ORDO 1.0! 2015.03.26 Because of a few requests: Hannibal 1.5 can't be tested as it crashes too often in the Shredder Classic GUI. I suspect a UCI Problem as the Classic is "Standard" per definition ... Unfortunately I can't use another GUI as the Classic seems to be the only one capable of running one tournament via multiple networked computers with a shared directory. 2015.01.29 Stockfish 6 is the new No. 1 in both of my lists. Very remarkable 39 Elo in the Complete list! Impressive! (With Bayes it would be a 33, with Elostat 35 Elo win over SF5) Just to show the differences: This is SF5 in the IPON-RRRL: 1 Komodo 8 : 3143 9 2462.0 3300 74.6% 2 Stockfish 5s : 3142 10 2458.5 3300 74.5% 3 Houdini 4 : 3125 9 2396.0 3300 72.6% 4 Gull 3 : 3076 9 2206.0 3300 66.8% 5 Equinox 3.30 : 3001 8 1884.0 3300 57.1% 6 Critter 1.6a : 2993 9 1849.5 3300 56.0% 7 Deep Rybka 4.1 : 2959 8 1699.0 3300 51.5% 8 Deep Fritz 14 : 2897 8 1419.5 3300 43.0% 9 Texel 1.05 : 2892 9 1396.5 3300 42.3% 10 Chiron 2 : 2888 8 1376.5 3300 41.7% 11 Protector 1.7.0 : 2883 9 1355.0 3300 41.1% 12 Naum 4.6 : 2867 9 1287.5 3300 39.0% 13 Hannibal 1.4b : 2864 9 1274.5 3300 38.6% 14 Senpai 1.0 : 2834 9 1145.5 3300 34.7% 15 Nirvanachess 2.0a : 2829 9 1123.5 3300 34.0% 16 HIARCS 14 WCSC 32b : 2815 9 1066.5 3300 32.3% And this is SF6 vs the same opponents, openings ... 1 Stockfish 6 : 3178 9 2574.0 3300 78.0% 2 Komodo 8 : 3138 9 2436.0 3300 73.8% 3 Houdini 4 : 3123 9 2383.0 3300 72.2% 4 Gull 3 : 3075 9 2196.0 3300 66.5% 5 Equinox 3.30 : 3000 9 1877.0 3300 56.9% 6 Critter 1.6a : 2993 8 1845.5 3300 55.9% 7 Deep Rybka 4.1 : 2959 8 1695.5 3300 51.4% 8 Deep Fritz 14 : 2896 8 1410.0 3300 42.7% 9 Texel 1.05 : 2890 8 1386.5 3300 42.0% 10 Chiron 2 : 2886 8 1367.0 3300 41.4% 11 Protector 1.7.0 : 2882 8 1348.5 3300 40.9% 12 Naum 4.6 : 2867 9 1282.0 3300 38.8% 13 Hannibal 1.4b : 2864 8 1272.5 3300 38.6% 14 Senpai 1.0 : 2834 9 1142.0 3300 34.6% 15 Nirvanachess 2.0a : 2829 9 1120.5 3300 34.0% 16 HIARCS 14 WCSC 32b : 2815 9 1064.0 3300 32.2% Looking at this the win of SF6 over SF5 would be 35 Ordo-Elo (and less with Bayes or Elostat) but the average Elo stregth of the opponents is lower ... 2015.01.26 Texel 1.05 is the new No. 9 in the IPON-RRRL. Remarkable +56 Elo to it's predecessor! 2015.01.08 Nirvanachess 2.0a added to the IPON. New No. 15 in the IPON-RRRL. Nice new entry in the TOP 16 2015.01.04 Equinox 3.30 added. No improvement over 3.00. Changed the rating again to 2800 for Shredder 12 as the Zero offset for the No.1 might be the best solution but looks to "strange". 2014.10.02 Naum 4.6 added. With a nice gain of 38 Elo over the old 4.2 Version. 2014.09.24 Protector 1.7.0 added. A plus of 10 Elo to the last release. 2014.09.13 2420 additional games for Critter 1.6a. Still below the previous version 1.4a. No change for the IPON-RRRL. 2014.09.12 I switched to the latest ORDO-v0.9.7. With this version it is possible to remove the last digit to get full Elo. As I personally think the smallest atom part of the Elo scale is "ONE ELO" I prefer this kind of presentation. The maximum difference to the old style is 0.5 Elo which should be negligible. 2014.09.06 The IPON has a new No.1! Komodo 8 made a jump of 41 Elo and is leading by a small margin in all statistical programs (see Archive -> Individual statistics, you find all stats there). It is the first time that Komodo gained a No.1 spot, congratulations to Larry Kaufman, Mark Lefler and, of course, Don Dailey. 2014.08.09 The explanation as described on 2014.05.31 is much to complicated to be maintained. I decided to simply show what all rating lists are about. The difference between the engines. (As any rating number for engines (and humans) is arbitrary and doesn't mean anything the No.1 Engine will be rated with 0 Elo and all others lower than that. The TOP 16 are for sure better playing chess than any human (living or dead), which means you can add whatever you like to get a comparable number to believe in. ;-) ) 2014.07.06 Equinox 3.00 included. 25 Elo gain from version 2.02. It passed Critter and gained one rank. I changed the name of "Stockfish 5 Syzygy" to "Stockfish 5s" to indicate the Syzygy version. As H4 performed a bit better versus Equinox 3 than Stockfish 5s it changed ranking within the Bayeselo list. H4 would be No.1 there again. You can find all lists in the download on the "Archive" section. 2014.06.15 Two administrative changes. 1. The lists are switched to Ordo 0.8 as this seems to be more accurate in some close cases 2. Stockfish 5 (no SYZYGY bases) removed from the list as I don't want to have 2 nearly identical engines 2014.06.06 Stockfish 5 with 4pc Syzygy bases added. The engine gained 6 Elo (in the one on one TOP 16 comparision) and just passed H4. With this the IPON-RRRL has a new No.1! It is that close that the next new entry might change this! New anchor for the lists is Stockfish 5 Syzygy with 3113 Elo. For the "experts": I added Bayes, Ordo and Elostat results in the IPON Archive. 2014.06.01 Stockfish 5 entered the complete list with a very impressiv plus of 40 Elo to Stockfish DD and ended just 5 Elo below H4. As a little remark I have to say that ELOSTAT would have Stockfish in front in the Complete List and in the RRRL. It is very close. 2014.05.31 A new No. 12 entered the IPON-RRRL. Texel 1.04 with a rating of 2835 Elo. Congratulations! As Texel pushed out my reference engine of the TOP 16 I will try a new method of "reference". It is always the No. 1 which is giving the reference point. Today it is Houdini 4 with 3111 Elo. As soon as a an engine passes the No.1 it initial rating became the new reference point. With this I always have a reference and the ratings hopefully doesn't change to much. My "theoretical" gut feeling tell me that over time (which means 2, 3, 4 new No1 engines) the overall rating will drop a bit e.g. the old reference, Shredder 12, will drop below 2800 Elo. Most important is the difference of the engines and that is not different than with any other reference point. 2014.05.26 Komodo 7a now with all games in the Main List. A new No.2 with a very impressive 32 Elo jump! I got requests about statistics for individual results. Please check the Archive section. It is all available there. 2014.05.25 Komodo 7a included in the Complete List. 31 Elo increase over K-TCECr Statistics and Main List will be updated after the last games. 2014.05.04 Protector 1.6.0 added to all lists. 33 Elo better than the last release. Rank 11 in the IPON-RRRL! 2014.04.20 Finished the missing games of Gull 3 vs DF14 and made a new main list. No. 2, 3 and 4 are all within one SD basically they are that close that their playing strength is hard to distinguish. 2014.04.19 Gull 3 added to the complete list. Nice increase of 40 Elo and a new No. 3. The IPON-RRRL will follow as soon as the missing games are finished. 2014.03.20 Senpai 1.0 included. An entry of 2841 Elo for the main list. Impressive first release! As Senpai is pushing a number 16 to 17 the last engine will be excluded out of the TOP 16 enignes. It was very close but it is Depp Sjeng c't 2010, an engine which I really liked to have in my list. This would be a TOP 17 List of the IPON: 1 Houdini 4 3119 9 9 3520 77% 2906 29% 2 Stockfish DD 3072 8 8 3520 72% 2909 41% 3 Komodo TCECr 3057 8 8 3520 70% 2909 38% 4 Gull 2.8 3023 8 8 3520 65% 2912 40% 5 Critter 1.4a 2982 8 8 3520 59% 2914 46% 6 Equinox 2.02 2978 8 8 3520 59% 2914 46% 7 Deep Rybka 4.1 2968 8 8 3520 57% 2915 47% 8 Deep Fritz 14 2901 8 8 3520 47% 2919 45% 9 Chiron 2 2893 8 8 3520 46% 2920 45% 10 Hannibal 1.4b 2875 8 8 3520 44% 2921 43% 11 Senpai 1.0 2843 8 8 3520 39% 2923 41% 12 Naum 4.2 2838 8 8 3520 39% 2923 41% 13 Protector 1.5.0 2836 8 8 3520 38% 2923 43% 14 HIARCS 14 WCSC 32b 2822 8 8 3520 36% 2924 40% 15 Jonny 6.00 2805 8 8 3520 34% 2925 37% 16 Deep Shredder 12 2800 8 8 3520 33% 2926 38% 17 Deep Sjeng c't 2010 32b 2798 8 8 3520 33% 2926 39% 2014.03.17 Gull 2.8 included. 3023 Elo in the main list. Ranked 4th! 2014.03.14 All counters removed - No vanity, no liabilities! 2014.01.05 1. Equinox 2.02 included. 2978 Elo in the MAIN lst. 2. For historical reasons the 4 year old Robbolito 0.085g is included. It is possible to compare it with the alleged source but keep in mind that Robbo did not play the same opponents. It's average opponent elo is much higher. As "the source" is playing with a contempt there is the possibility that it would lose more points when playing the same strong engines. 3. The IPON is fixed to 2800 elo for Shredder 12. This is an offset of 2783 over ALL engines. I applied this offset to the three ECO based lists. This is more accurate than having a fixed engine there. I will do that principle with all ECO lists in the future. 2014.01.03 The three rating lists sorted by opening system have to be recalculated. It makes no seance to fix this with a certain engine as this might cover an in- or decrease of another engine in case the fixed engines is weaker or stronger in one of the lists. It is better to give the lists the same average elo value as all list have the same engines. This way it is just the distribution which counts, not the relative distance to a specific engine. Unfortunately I deleted the individual PGN data for the lists. I will remodel the list with the next engine I include. 2014.01.01 I restarted the IPON 2014 with some changes. 1. The old 75 opening positions set is increased by 35 new position to 110 opening positions. The ECO distribution is 21% open games (ECO C20-C99), 30% half open games (ECO B + C00-C19) and 49% closed games (ECO A,D, E). This is less 'GM tourney' practice but more suited to the average chess player and his analytical needs. 2. Only 16 top engines are tested. The reasons are a smaller width of opponent elo and less games to play. Additionally the error bar is below 10 Elo now for all top 16 engines in the IPON-RRRL. These 10 elo are the "border of irrelevance" for me, as no one can feel or distinguish a + or - of 10 elo. (Having smaller error margins in a testing environment is a different thing of course). 3. I will provide live games from time to time but not necessarily the day an engine is released ... With 110 opening positions and a O20/H30/C50 distribution the individual comparison of 220 games becomes interesting to a certain extend. It should not be stretched to much, but 220 games can be decisive in some cases. In the rating section I offer three new lists sorted by closed, half open and open openings. This is just an experiment to show some interesting trends. It should not be taken too serious as the number of games as well as the number of used openings is limited. Nonetheless, in some cases it might give a hint what can happen if a certain opening distribution is used (here and somewhere else). It shows too, that there is no 'right' or 'wrong' set of openings. Everything is correct and the observer has to ask if a particular set-up suits his needs! This is a free service. If you don't like it please have a look at some of the other excellent lists. There are plenty available to satisfy anyone! Bye Ingo

2015 by Ingo Bauer