Author: Martin Grabriel
Date: 17:18:31 11/15/99
Go up one level in this thread
Hi, you have certainly set a new standard in holding tournament games in terms of quality of paticipants, hardware used, reporting results, and methodology. Excellent work! I will visit your site regularly. One question though: Your "Genius 5 = Mephisto Genius 98 version" confuses me. I have Mephisto Genius 99 and I always believe it contains a version of Genius earlier than Genius 5. Can you or anyone reading this me help clairfy? Thanks in advance. On November 14, 1999 at 20:48:36, Heiko Mikala wrote: >Hi everybody! > >During the last weeks I have played a big tournament, g/60 on two computers, >including some of the brand-new and some of the strongest programs available. > >I would like to show you the result, and tell you about my impressions. > >The following programs played in this tournament: >------------------------------------------------- > >Fritz 6 >Rebel Century >Nimzo 7.32 >Chessmaster 6000 >MChess Pro 8 >Genius 5 (Mephisto Genius 98, using version 5.001 of the engine) > >The conditions were: >-------------------- > >Game in 60 minutes (60 minutes for each of the programs) >2 computers: IBM 6x86MX PR-300, 64/128 MB Ram (approx. as fast as PII-300) >Pondering enabled >Each program used it's own opening book, strongest settings, learning >enabled where possible, EGTB's where possible (3,4 and a few 5-men) > > >And this is the resulting crosstable: >------------------------------------- > > > 1 2 3 4 5 6 7 >1 Fritz 6 ** 1½ ½1 ½½ ½0 ½1 11 8.0/12 >2 Nimzo 7.32 0½ ** 0½ ½½ 1½ 11 11 7.5/12 >3 MChess Pro 8 ½0 1½ ** 01 01 10 1½ 6.5/12 >4 Hiarcs 7.32 ½½ ½½ 10 ** 10 10 01 6.0/12 >5 CM6000 ½1 0½ 10 01 ** ½0 ½½ 5.5/12 >6 Genius 5 ½0 00 01 01 ½1 ** ½1 5.5/12 >7 Rebel Century 00 00 0½ 10 ½½ ½0 ** 3.0/12 > > >You can find the PGN of all the games at: > >http://www.online-club.de/~rp45195/tournaments > > >I would like to point out, that I played this tournament to find out more >about the playing-styles of the new programs, not to find out which one >is the strongest or which one would win the tournament. > >So, to make this very clear, I'm not saying that the winner is the strongest >program and the last one is the weakest. > >I originally started to play Fritz 6 against some of the strongest opponents >I had available, two games each (1 white, 1 black) to find out more about >the new Fritz. I then ordered Rebel Century and thought it would be interesting >to play Century against the same opponents to find out more about Century and to >be able to compare the results. After these games had been finished, I thought >it would be interesting to make this a complete tournament, especially since >I wanted to see my personal favourite program (MChess Pro) to play some more >games. >During the time I played the rest of the games to make this a complete tourna- >ment, I heard of some extremely impressive results of Nimzo 7.32 (you can find >some of them at the Gambit Soft forum, tournament games played on two comps), >so I decided to order Nimzo 7.32 (yes, I'm crazy! ;-) and enter it into this >tournament too (crazy might be the wrong word, stupid may be better :) > >The games have been played using the Autoplayer wherever possible, with some >exceptions. All games including Chessmaster and Genius 5 have been played by >hand, the games between Rebel and MChess have been played by hand too. No matter >which method had been used, I have watched all games live, since the reason >for these games was to find out more about the playing style of the programs. >I didn't see any anomalies in the games that were played using Auto232, and I'ld >like to point out especially, that the games with Rebel involved went absolutely >fine. Rebels nodes/second rate has always been normal (around 100k np/s on my >computers). > > >My impressions of the programs and the games: > > >Fritz 6: >-------- >Well, this is one of the strongest programs available, no doubt about it. I've >heard some rumours about F6 not being stronger than F5.32, but I don't think so. >At least it won this tournament, it lost only 1 game out of 12 against this >extremely strong opposition, and it's also leading my private blitz rating-list, >which consists of 2500 games (g/5) in front of Hiarcs 7.32, Fritz 3.1 (!) and >Fritz 5.32. >My impression of F6 is, that it is one of the strongest programs tactically (we >all knew this already, eh?) but that it could be a bit more aggressive. Not that >it's a sheep, it's aggressive already, but maybe it should be even more >aggressive. >It seems to use it's tactical superiority mainly to prevent to fall into traps, >and >to defend all attacks, biting as soon as the opponent makes a mistake. But >this way it plays a lot of draws, and imagine what could happen, if F6 would >aggressively attack all the time!? > >Nimzo 7.32: >----------- >This is an extremely interesting program, and I'm happy that I finally bought >it. It's extremely strong, and it could well have won this tournament if it >would >not have played a bad opening line in the first game against F6 (the games >Nimzo-Fritz were the last games I played). Just to comment on the book, with the >exception of this one game I had the impression that Nimzo's opening book is >extraordinarily good. >Nimzo seems to be an extremely aggressive program, although in a different sense >than for example MChess. While MChess always tries to attack the enemies king, >Nimzo tries to attack where it sees the best chances. I have seen two games, >where >Nimzo agressively pushed forward it's pawns on one side of the board, when the >opponent didn't expect anything like that to happen. My impression is, that >Nimzo >has some knowledge to help it make plans in certain situations. On the other >hand, >there are situations, where Nimzo seems unable to build a plan, but then it >simply >plays in a style like all the other top programs, which isn't bad at all ;-) > >MChess Pro 8: >------------- >Well, if MChess had not been my favourite program already, it would have become >so definitely after this tournament! Some of the most interesting, most >thrilling >games of this tournament have been played when MChess was involved. >This program is simply unbelievable and it's style is so different than all the >others. MChess is by far the most agressive program I've ever seen, it always >tries to attack the opposing king, when possible. The most interesting thing >about >this is, that it often makes you believe, that long-range planning is involved. >It doesn't attack with only one or two pieces, resulting in short fights without >much being won, like some other programs do. Instead it slowly develops it's >pieces, building up an attack for some moves and then suddenly launches the >strike. >These attacks are often decisive for the outcome of the game, although, of >course, >sometimes the attacks fail too (but much more often they are succesful). >Another interesting thing about MChess is, that it's evaluation very often >indicates >the outcome of the game much earlier (often many, many moves) than the >evaluations >of the opponents. And it's always fascinating to watch it's mainlines. >MChess must have a fantastic evaluation function, and I'm sure it uses some >search extensions, which no one else uses. It's extremely slow compared to >others, >but still seems to be one of the best tactical programs around. >It plays an extremely strong endgame too. >Unfortunately I have the feeling, that there is a small but severe bug in >MChess's >search function, because I have seen in two games, in which MChess was leading >or at least equal, that it had a good PV-move with a good evaluation for a >long time, then suddenly switched to another move, showing a worse evaluation >than >before, immediately played this move, and directly after playing this move >discovered (while pondering) that this move was much worse than the move it had >thought about before. This should of course normally be impossible. If there is >a best move on a certain depth already, it should never switch to and play an >inferior move, if it sees that this move is worse... > >If you want to see some of MChess best games, please have a look at the first >game against Rebel, the two games against Hiarcs (these were thrillers!) or >at the first game against Nimzo, although in the game against Nimzo, MChess >didn't play a king-attack, but you can see it's agressivenes very good here. > >Hiarcs 7.32: >------------ >A very strong program too, but I still can not decide about it's playing style. >Sometimes it plays brilliant attacks, in other games it doesn't seem to be able >to find a plan at all. What makes me wonder is, that it sometimes plays >brilliant >king-attacks, but also seems to have a big weakness in it's own king-safety. You >can see this weakness very good in Hiarcs games against MChess. Nevertheless, >Hiarcs has become one of my favourites, because boring games with Hiarcs >involved >are very rare. >Tactically Hiarcs is extremely strong, which could make you wonder, considering >it's low nps rate. Must be using some great extensions, and some great selective >techniques too, because it often reaches deep search depths despite of it's low >nps-rate. > >Chessmaster 6000 >---------------- >A very solid program. Doesn't seem to have any extraordinary strenghts or weak- >nesses. It simply plays rock-solid chess. I have heard some people speak about >Chessmaster beeing an agressive program, sacrificing pieces sometimes to win a >game, but I couldn't witness such things in all the games I have played so far. >To all you CM-freaks, planning to lynch me now, let me tell you, that I've been >a collector of CM-versions for a long time, and really like the King engine. >But if you really want to see an agressive, attacking program, perhaps you >should have a look at CStal or MChess, maybe even Nimzo. Still, Chessmaster is >a great program, I enjoy to see it play. > >Genius 5: >--------- >The legend. Still one of the strongest, most impressive programs out there. I >don't understand all the rumours about Genius beeing much weaker than the other >top programs, it simply isn't. For example, look at the crosstable of the tour- >nament, before Nimzo 7.32 entered the stage: > > 1 2 3 4 5 6 >1 Fritz 6 ** ½1 ½0 ½½ ½1 11 6.5/10 >2 Genius 5 ½0 ** ½1 01 01 ½1 5.5/10 >3 CM6000 ½1 ½0 ** 01 10 ½½ 5.0/10 >4 Hiarcs 7.32 ½½ 10 10 ** 10 01 5.0/10 >5 MChess Pro 8 ½0 10 01 01 ** 1½ 5.0/10 >6 Rebel Century 00 ½0 ½½ 10 0½ ** 3.0/10 > >Genius was in second place here. Unfortunately it lost both games against Nimzo >badly, and my impression is, that it has huge problems against the very good, >very fast Nullmovers, like Fritz and Nimzo. But it still is one of the best >defenders around, and it definitely still plays one of the best endgames of all >programs available, although it doesn't use EGTB's. >If you like Genius style of play is another question, though. I, personaly, >think >that it plays much too defensive, and so I often find Genius's games against >other >computers a bit boring. But if it manages to reach the endgame, you'll see some >great chess! And I still can't let my hands of it. It's a legend, it will always >be a legend and legends tend to be fascinating ;-) > >Rebel Century: >-------------- >I don't understand what happened to Rebel. Rebel 8 has been one of my favourite >programs for a very long time, because it played the most solid chess I had ever >seen in a chess program. It *never* made a bad mistake and it's opponents had to >play extraordinarily good to beat Rebel. I had expected Rebel Century to finish >in one of the first places in this tournament, but well, it didn't. I had >published all the games of F6 and Century on the german Gambit-Soft forum some >weeks ago, with some comments, and then asked others to help me with Century, to >find out about some potential problems in my installation. I even asked them to >check the moves, but I didn't get a response from anyone being unable to repro- >duce the moves. I'm sure I used the strongest settings, I did reload the Rebel >personality, set Rebel to strongest settings, even did let it play in pure >MS-Dos >mode, which is said to be a few percent faster for Rebel, than a dos-window in >Win9x. > >Well, after all, these have only been 12 games, and there has been a positive >trend in the games against Chessmaster and Hiarcs. The last two games against >Nimzo on the other hand have been bad losses again. But then again, Rebel has >always had big problems against fast Nullmovers like Fritz and Nimzo. > >To make the impression of Rebel not too bad, there have been some very positive >results of Rebel Century, posted on the german CSS and Gambit-Soft forums by >some >other testers. So maybe Century was simply unlucky in my tournament. I really, >really hope so, because I like Rebel. > > >Finally: >-------- >I already have ordered the new HCC/Millennium package, which includes Shredder 4 >and Genius 6.5 (crazy? stupid?? You name it! ;-), and if I find the time and >the energy, I would really like to enter these two programs in my tournament. >Other possibilities would be CSTal II or Crafty 17.x, but I would like to enter >the very new programs first, and in case of Crafty, I would have to play all the >games by hand... >I'd also like to see, how Tiger would do in this tournament, but it seems we >will >have to wait some more time, before Tiger is available. > > >Ok, enough for today, I hope you enjoy the games (remember, you can download >them >at the address mentioned above), > >greetings, > >Heiko.
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.