Author: George Tsavdaris
Date: 13:12:10 01/01/04
Go up one level in this thread
On January 01, 2004 at 14:27:55, Brian Katz wrote: >On January 01, 2004 at 12:44:44, George Tsavdaris wrote: > >>On January 01, 2004 at 11:17:32, Brian Katz wrote: >> >>>I have found that Fritz 8 (8.0.0.8 and 8.0.0.23)has trouble solving some >>>positions that other engines solve in a matter of seconds. Such is the case with >>>the recent position: >>> >>>8/8/K6p/6p1/8/8/k5PP/8 w >>> >>>Of course Tablebases are needed to find the mate in 29, which is the case with >>>Ruffian, Yace and Crafty 19.06. Hiarcs finds it in 30. >>> >>>Fritz 8.0.0.8 did not find it at all as yet. Fritz 8.0.0.23 just found it as I >>>am sending this post after 00:08:26, but it shows mate in 31 instead. >>>Once Fritz 8 finds a mate, it seems to end it's search whereas, some other >>>engines continue to search and find quicker mates. >>> >>>Any and all opinions welcomed. >>>Brian >> >> Fritz 8 has no trouble at all to solve this position. On a PIV-1500MHz with >>48 MB hash tables and all 3-4 and some 5 piece endgame tablebases it finds the >>winning move instantly and shows it is a winning move (+5) in 5 seconds. >> >> Fritz 8 is not a mate solver, but a chess player, so if it isn't able to >>anounce a mate in x, this is not bad. >> > > > > >Hi Thanks for the reply. > >This is true, Fritz 8 is not a MATE solving engine, even though at times it >finds long mate quite rapidly. >I just wonder why other "NOT MATE SOLVING ENGINES" find the Mate quite fast, and >Fritz 80023 finds MATE in 31 in 5 min/17 seconds and after over 45 minutes, >Fritz 8008 cannot find it at all. > > First off, just out of curiosity, do yo really have all the 5 man Tablebases? I have "Fritz endgame turbo" I and II so yes. But i haven't saved all 5-piece in the hard drive. > I bought the 9 disc set of Nalimov Tablebases, and there were still SOME >(ONLY SOME) missing. I used the Tablebase Generator to find just which ones were >missing and I then used the Tablebase Generator to GENERATE the missing >endgames. This took a few days to do. Perhaps it would have taken only one day >if I didn't have to break in between. I then copied the Generated endgames onto >a DVD. >Secondly, which engine version of Fritz 8 do you have? I have Fritz 8.00.8 and Fritz 8.00.23. > >I had the Hashtables set at 64 MB in this case. I often have the much higher, >sometimes upto 768 MB Sometimes the engine stalls at this setting and I then >lower it to 640. > >Yes, Fritz 8 Version 8.0.0.8 solves the position in only 3 seconds where it >shows a +-(1.41) (Plus Minus) rather than the Plus over Minus(0.84) it gave >after only 2 seconds. Also in the 3 second time frame it shows +-(5.50. > >Fritz 8 Version 8.0.0.23 finds the position as +-(1.31) and +-(5.50)in the ONE >second time frame. It also goes on to find the mate in 31 rather than 29. It >finds this in 00:05:17. >This shows quite a bit of difference between Engine Versions. I am curious as to >just how much better is the recent 8.0.0.26 Version? >The other odd thing to question, is........Many people are not happy with the >80023 version because of the e4 problem that so many have referred to on this >site, and also, many believe that 8008 is stronger. Yet, 80023 is able to solve >this position much quicker. >It must also be noted, that, there are weaker engines out there that are able to >solve some positions much faster than some stronger engines. Right, top engines can't be top in every chess position. Ikarus v0.18 for example finds the mate rather quickly: (Interesting is at depth 1/24, it says Kb5 +- (#30) ---> a bug??) Analysis by Ikarus V0.18: 1.Ka6-b5 +- (#30) Depth: 1/24 00:00:00 1.Ka6-b5 Ka2-b3 = (-0.17) Depth: 2/2 00:00:00 1.Ka6-b6 Ka2-b3 = (0.02) Depth: 2/2 00:00:00 1.Ka6-b6 Ka2-b3 2.Kb6-b5 ² (0.33) Depth: 3/3 00:00:00 1.Ka6-b6-- ² (0.33) Depth: 4/4 00:00:00 1.Ka6-b6 Ka2-b3 2.Kb6-b5 h6-h5 = (-0.15) Depth: 4/4 00:00:00 1.h2-h4 g5-g4 2.Ka6-b6 = (-0.03) Depth: 4/5 00:00:00 1.h2-h4 g5xh4 2.Ka6-b6 = (-0.02) Depth: 5/5 00:00:00 1.Ka6-b6 Ka2-b3 2.Kb6-b5 Kb3-c3 3.Kb5-c5 ² (0.32) Depth: 5/6 00:00:00 1.Ka6-b6-- ² (0.32) Depth: 6/7 00:00:00 1.Ka6-b6 Ka2-b3 2.Kb6-b5 Kb3-c3 3.Kb5-c5 h6-h5 = (-0.16) Depth: 6/7 00:00:00 1.Ka6-b6! = (-0.16) Depth: 7/9 00:00:00 1.Ka6-b6 Ka2-b3 2.Kb6-b5 Kb3-c3 3.Kb5-c5 Kc3-d3 4.Kc5-d5 ² (0.32) Depth: 7/9 00:00:00 3kN, tb=56 1.Ka6-b6-- ² (0.32) Depth: 8/10 00:00:00 4kN, tb=68 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 = (-0.09) Depth: 8/10 00:00:00 6kN, tb=101 1.Ka6-b6! = (-0.09) Depth: 9/11 00:00:00 10kN, tb=117 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e3 ² (0.32) Depth: 9/11 00:00:00 14kN, tb=164 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e2 5.Ke5-f5 Ke2-e3 = (0.05) Depth: 10/12 00:00:00 19kN, tb=234 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.g2-g3 Kd3-e3 = (0.20) Depth: 11/13 00:00:00 31kN, tb=388 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e3 5.Ke5-f6 Ke3-f2 = (0.00) Depth: 12/15 00:00:00 46kN, tb=632 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e3 5.Ke5-f6 Ke3-f2 6.g2-g4 Kf2-g2 7.Kf6-g6 = (0.04) Depth: 13/17 00:00:01 73kN, tb=1056 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e3 5.Ke5-f6 Ke3-f2 6.g2-g3 Kf2-g2 7.Kf6-g6 = (0.00) Depth: 14/19 00:00:01 112kN, tb=2088 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e3 5.Ke5-f6 Ke3-f2 6.g2-g4 Kf2-f3 7.Kf6-f5 = (0.04) Depth: 15/19 00:00:02 163kN, tb=3307 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e3 5.Ke5-f6 Ke3-f2 6.g2-g4 Kf2-f3 7.Kf6-f5 = (0.01) Depth: 16/21 00:00:03 226kN, tb=5447 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c3 3.Kc5-d5 Kc3-d3 4.Kd5-e5 Kd3-e3 5.Ke5-f6 Ke3-f2 6.g2-g4 Kf2-f3 7.Kf6-f5 = (0.03) Depth: 17/21 00:00:04 307kN, tb=7772 1.Ka6-b6 Ka2-b3 2.Kb6-c5 Kb3-c2 3.Kc5-d5 = (0.03) Depth: 18/23 00:00:06 400kN, tb=11047 1.g2-g4 Ka2-b3 2.Ka6-b5 Kb3-c3 = (0.09) Depth: 18/23 00:00:07 577kN, tb=16402 1.g2-g4! = (0.09) Depth: 19/22 00:00:07 602kN, tb=17258 1.g2-g4 Ka2-b3 2.Ka6-b5 Kb3-c3 3.Kb5-c5 Kc3-d3 4.Kc5-d5 Kd3-d2 5.Kd5-e6 Kd2-e3 6.Ke6-e5 Ke3-e2 7.Ke5-f6 +- (#30) Depth: 19/22 00:00:07 613kN, tb=17413 >The evaluations of Fritz 8008 and Fritz 80023 are shown below. >Regards >Happy New Year >
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.