I normally spend a lot of time with dedicated chess computers and old home computers. Why? Because they amaze me on how much was done back in the old days with so little. So little memory. So little speed. So little technology compared to today.
Recently I finished a monster tournament between Commodore 64 software and comparable dedicated chess computers, which I found very enjoyable because I was permanently amazed by the little 1 MHz computer as the written software of the top programmers of the day churned out amazingly good moves while rarely moving beyond 3 ply in their search.
Well, this caused me to think nostalgically, wouldnt it be nice to just see how these old programs would play against the modern programs of today. My first thoughts on this were that they would likely get hammered.....but then I wondered why not try it out and see what would really happen.
Picking up my TASC Final Chesscard for the Commodore 64, I then decided to play a 5 ply game (figuring this is easy since the engine and the Chesscard can both be set to play at this level setting). Well the Chesscard took a few seconds for it's moves and the engine fractions of a second.... but interestingly both programs seemed to show search extensions (variation extensions) occasionally beyond 5 ply although they both conformed to the 5 ply search. Now this seemed interesting to me, because it seems that even in ply both searched selectively. I then played Brute Force with the Final Chesscard and at that point I could definately see that in Brute Force a 5 ply search would take many minutes to play.
Now this excited me, because the idea of playing old against new became a much more realistic possibility. Taking out my Atari 800, I then loaded Colossus 4 and set it to Infinite Level and played a few moves. It took the program over 30 minutes (1.79 MHz) to search through 5 Ply Brute Force. Now setting it to 30 seconds per move for a real game it would reach 4 ply but not go beyond that. At 40/2hrs it would reach 5 ply but rarely unless it is a simple endgame would it go beyond 5 ply. I then tried the same on my C64 and the results were the same (taking into consideration the speed difference between a C64 and Atari 800).
Therefore it seems to me that the way to play matches against these old programs and new programs is to let them do what is natural as much as possible. Ideally the best way would be if Winboard or some other platform had a way to control moves at fractions of a second, which unfortunately they do not. Another difficulty with this is knowing how much allowance to provide to an engine for it's start delay.
Since the above is currently not possible, the only way to try this out is to play ply games with modern engines, knowing that this might curb some of the techniques that modern engines use. But, I also have a feeling since they continue to search within Ply selectively that this is not much different to the old software techniques for playing games with level settings.
Therefore I decided to play a Test Tournament anyway under the following conditions which I feel would allow the programs to behave naturally.
The Tournament is a Team tournament and I am playing this for fun and for the experience of understanding and enjoying the respective chess programs. All the moves will be manually input (even between engines) because I feel this allows me to follow their games and enjoy them. The contestants for the Tournament were placed into 6 sections as follows:
Section 1 comprises of Teams of Old Home Computers. Several of the software are the same (there just weren't too many choices back then), but I wanted to play them anyway because of wanting to see if the softwares remained the same while ported to the various platforms (Often the programmer was different ie.. Skeet Hannigan for Sargon III C64 and Alex Ford for Amiga) I also know from a programmer back then that sometimes for particular platforms they had to be creative because of lack of ROM and RAM. Therefore I am curious how they all play in this Tournament. The Final Chesscard I will use as a special test to see if there are differences between Agressive and Normal game style and also differences between 5 Ply selective search and Level game play selective search.
From my trial and errors with the C64, I have decided to allow all the old chess computers to play their Level settings. The starting base is 225 seconds per move for the C64 (= 40/2.1/2 hrs), or a level that is closest to this. An Atari 800 for example at 1.79 MHz playing Colossus 4 would get 2 Hrs 5 minutes and 42 seconds to play 60 moves (7542s/60). The nearest lower level for Sargon III is Level 5.
Section 2 - Is a dedicated chess computer section. I have played around with these these and here also I find that the fairest setting for everyone would be to use corresponding appropriate level settings. Ply does not work either because for example in the case of Mephisto Vancouver 68020 5 ply would mean that it would still selectively search variations up to 12 deep, which I think would be a great disadvantage to Modern Engines. Therefore instead Mephisto Vancouver (12 MHz) would play 60/19m or in the case of Mephisto Magellan 10 seconds per move.
Section 3 - Has a mixture of dedicated chess computers which will play with appropriate level settings or ply in TASC's case, and DOS/Windows software. The DOS/Windows software will play 5 ply. With DOS/Windows software I do not know if they all play selectively or brute force but we will finds out....
Section 4 - Are mostly classic chess engines that I enjoyed playing over the years. It is a chess engine nostalgia section. All these engines will play 5 ply. For 5 PLY games I will use Arena whereever possible, but some engines will have to play in Chessbase/Fritz gui.
Section 5 - Comprises of some old and some newer engines. Kind of in the middle between the Classic engines and the New Engines. Again all will play 5 Ply.
Section 6 - Represented here are all the Modern TOP engines of the world. Some of the latest top PRO engines I don't have (been spending all my money in the last couple of years on dedicateds
) but I think there is a good representation in this section. I would have liked to have included some others like Junior but these fail the 5 ply test since the mimum ply is 6.I used http://www.random.org/ to randomply pick the teams for the Tournament Groups (I also did the same for most of the engine groups). The Tournament Groups are as follows:
Because I don't want to have too many computers and dedicated computers spread across the room, I will be playing each individual group from start to finish, before I move on to the next Group.
GROUP A has started and looks as follows:
Round 1
36 games were played in Round 1 and most of them were really hard contested. The Old Home Computers posted some surprisingly good results. Poor DOS/Windows, it seems that they will struggle in the Group.
Standings after 1 Round
The Round 2 schedule is as follows:
All the schdules are randomly picked using http://www.random.org/, even the coin toss to see who starts with white first
.If there is an interest in the actual games played in this Tournament, please let me know and I will post them here also.
I know this is a little different to most of the Tournaments played in this Forum, but I hope you will find it interesting. I am playing this for the sheer enjoyment and for very little science
Best regards
Nick
as in dedicateds