Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: summer tournament Tue Aug 31, 2021 11:08 pm
The sensation has happened.
The unknown Clockchess89 by Chris Whittington beats the Tiger Grenadier and climbs up in the tournament , has 5/7 together with Steinitz and leads the tournament so far.
Beginn des Turniers: 2021.08.18, 14:37:29 Letzte Aktualisierung: 2021.08.18, 14:38:09 Ort/ Land: ORION8, Deutschland Spielstufe: Turnier 40/120 20/60 Hardware: Dual Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz mit 1.016 MB Speicher Betriebssystem: Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790) PGN-Datei:Schachcompus.pgn Tabelle erstellt mit:Arena 1.99beta4
TheSelfImprover
Posts : 3028 Join date : 2020-11-18
Subject: Re: summer tournament Tue Aug 31, 2021 11:54 pm
Mclane wrote:
The sensation has happened.
The unknown Clockchess89 by Chris Whittington beats the Tiger Grenadier and climbs up in the tournament , has 5/7 together with Steinitz and leads the tournament so far.
Once again, it seems to have a certain cleverness about it. It looks as though Chris had some success in getting this 8-bit program to play the way he wanted it to play. Did he release any other programs after Clock Chess but before his masterpiece, CS-Tal?
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: summer tournament Wed Sep 01, 2021 12:09 am
This was all before i met chris. I had a zx spectrum. I even had a chess cassette tape with one of his programs, but i had only the 16 K spectrum and i guess i had the wrong program.
Then i changed to cpc464 and i came to chris later when i had the atari ST. There i saw the engine was interesting and i contacted him.
But i oversaw the 8Bit engines he did. When i first got superchess3.5 on a spectrum emulator i was surprised how typical chris alike it was. Detailed information how it is working is written in the help text of Superchess 3.5. I chose clockchess89 because i think superchess has no permanent brain while clockchess has. I have no clue if superchess3.5 is weaker or if clockchess89 is a superchess3.5 with permanent brain.
The next thing was to give colossus chess and clockchess similar spped like the dedicated chess computers have.
Therefore i put colossus chess4 to 5 mhz and clockchess89 from 3.5 to 10 mhz. This way the software engines are competitive to the dedicated chess computers.
He released of course many engines on atari ST and PC before CSTal appeared. That was his main job to produce several chess engines and generate money.
TheSelfImprover likes this post
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Beginn des Turniers: 2021.08.18, 14:37:29 Letzte Aktualisierung: 2021.08.18, 14:38:09 Ort/ Land: ORION8, Deutschland Spielstufe: Turnier 40/120 20/60 Hardware: Dual Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz mit 1.016 MB Speicher Betriebssystem: Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790) PGN-Datei:Schachcompus.pgn Tabelle erstellt mit:Arena 1.99beta4
An enjoyable and easy to follow game, but a couple of SuperStar's moves were a bit weak. The programmer, Julio Kaplan, was a strong player, but maybe lacked Chris's skill level in programming and in getting the program to play how he wanted.
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
I am different opinion. Julio kaplans engines tried to play really good chess. While dave Kittinger beasts concentrated on tactics.
Julios chess engines were selective.
Of course if we relate this to chris Whittington, we have to admit that chris was able to make good playing engines VERY EARLY in the age of computers.
One cannot say that julio kaplan was a weak chess player. But he had maybe not the best programmers in his heuristic alpha company, While chris was boss and programmer in one person. Chris is IMO one of those rare renaissance genius, He was teacher, carpenter, programmer and chess player. Having 6 children his family was a huge job, He managed to be father and boss and all kind of duties. He never forgot his vision. His earliest chess engines are similar to the last creations. He never changed. The machines changed. And so changed the outcome.
Julio Kaplan needed more time then chris Whittington. He was capable to beat the others with the D and D+ and D++ version but it took many many programs before to refine the engine.
With chris it was different, even his first engines had the same strength (if you forget about the different hardware platforms).
So the spectrum programs were weaker because the hardware was weak. But the software was as strong as on the later hardware platforms.
Yes i would say chris is one of those rare renaissance people. When i was with him at the Aegon tournament in the hague and he explained me his latest ideas he took out a paper and a pencil and drew his ideas and algorithms on it. Like a painter.
He never stopped this “flue” while talking, it poured out of him like a spirit.
This is a very good thread: the story of 8-bit chess computers unfolding in the form of a live tournament with understandable games, 30 years after their time!
I'm struggling to think of a similar experience!
Mclane likes this post
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: summer tournament Thu Sep 02, 2021 12:12 am
Concerning kittinger machines. When i tested (for computer schach and spiele, a big german computerchess magazine) dedicated chess computers from novag, i had to test novag forte A and B with Kittinger engines.
After i wrote the article for chess engines. I telephoned with the german distributor of novag machines ZENS. And i told him: it was a hard time to test your machines because they have almost ZERO understanding about chess.
Later on ZENS complained to CSS (Computer schach and Spiele) editorial because he got this telephone call of a tester who complained his dedicated chess computers do not play chess at all.
Very funny.
A few years later NOVAG hired don dailey and larry kaufmann to help them making NOVAG computers playing more KNOWLEDGE able, so don dailey programmed piece square tables into novag 8 bit machines. As a result suddenly the Novag super Forte/Expert C played strategical accurate. Suddenly the machines put the pieces into the right squares and developed the pieces accurate. In opposite to the machines i had to test before.
Novag never told us about. It came out to the public much later.
But it fit with my experience at that time.
TheSelfImprover likes this post
Tibono
Posts : 26 Join date : 2020-11-27 Location : France
Subject: Re: summer tournament Thu Sep 02, 2021 8:07 am
Hello!
I performed several 15s/move tournaments including ChrisW's Z80 programs (ZX emulator, standard 3.5Mhz). CElo=computer Elo (ref. Excellence=1780 CElo); Rperf=relative performance (hardware agnostic, ref. Excellence=100%)
- 1983 SuperChess (level 2) 1241 CElo (73% Rperf) - 1985 SuperChess 3.5 1398 CElo (82% Rperf) - 1989 Clock Chess 89 1430 CElo (84% Rperf)
A breakthrough then in 1990, with Chess Simulator/ChessPlayer 2150 (Amiga/PC DOS) leveraging far more powerful CPUs and enabling much more performing chess software:
- 1990 ChessPlayer 2150 on PC DOS 80386 @16Mhz 1702 CElo (92% Rperf) - 1990 Chess Simulator (Player 2150) on Amiga 68020 @14.1Mhz 1751 CElo (92% Rperf)
Following ones were Chess Champion 2175, Complete Chess System, Chess System Tal... This later one, indeed a masterpiece; CSTal reaching 109% Rperf = on par with best/latest Spracklen software.
Last edited by Tibono on Thu Sep 02, 2021 8:09 am; edited 1 time in total (Reason for editing : typo)
Mclane and TheSelfImprover like this post
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: summer tournament Thu Sep 02, 2021 11:34 am
You have to consider that I play 180" per move and spectrum 3x oc.
I've always wondered how the developers used the 1-4k of RAM that used to be available in the 8-bit chess computers. Some will be needed for recording moves, a few bytes will be needed to memorise various aspects of the system's state (and game state - e.g. castling rights), some will be needed to evaluate positions, and I imagine that most of it will be used to traverse the game tree, and memorise the scores of the best (= "least worst") positions in the game tree.
It has just occurred to me that there are people here who know the answer, so this is actually the ideal place to ask!
One more question: why did even expensive 8-bit chess computers have so little RAM?
Tibono
Posts : 26 Join date : 2020-11-27 Location : France
Subject: Re: summer tournament Fri Sep 03, 2021 7:59 am
TheSelfImprover wrote:
why did even expensive 8-bit chess computers have so little RAM?
In the early 80's I expanded my TRS-80 RAM (from 16K to 48K) and I remember the RAM itself was MUCH expensive. Since, RAM price has dropped down several orders of magnitude! Usual 8bits CPUs (such as Z80 and 6502) gather 2 x 8bits registers for addressing memory up to 64K addresses, so providing 4K chess computer was not a matter of limited technology - just a cost concern. And worth pointing out, those chess programs were assembler (machine language) coded, which can be very compact, and enable handling data at bit level. So, game state can be stored in a few bits rather than bytes.
Admin and TheSelfImprover like this post
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: summer tournament Fri Sep 03, 2021 10:15 am
All true but most intelligent chess engines of that time had at least 8 kb ram. And rom was at least 32 kb. Only the first brute force machines had 4 kb because for brute force you dont need so many space. The more selective and intelligent the search and the evaluation is, the more kb you need. This is the reason IMO ed schroeders early Amsterdam prototype (here rebel mystery modul) on the rathsmann pcb had only 4kb and could not generate the playing strength it later created with rebel 5.0 on the new PCB H+G gave ED after his successful appearance in cologne 1986.
Most softwares for homecomputers (e.g. zx spectrum 16k/48k) was split e.g. between the 16k version and the 48k version of the hardware. So when buying the software you had to decide which spectrum you have in order to get the most out of the software/hardware.
I remeber some companys put the low ram software on cassette side A and the high ram version on cassette side B.
If we consider at that time c64 had 64 kb ram giving the end user arround 30 kb size, cpc464 similar, spectrum 48k, oric atmos etc. Or bbc acorn B, this means most homecomputers had more ram then the usual dedicated chess computer with its 32 / 8 kb platform.
The main difference was mhz.
Most dedicated chess computers were 3.7 mhz, 4 mhz, 5 mhz or more.
The 6502 beeing used were special versions that could run faster then the homecomputer cpus.
C64 and even first apple machines had only 1 mhz cpu !! Zx spectrum was 3.5 on a z80a.
In the early years of dedicated chess computers the 6502 was also slow and limited to 1-2 mhz. But then faster versions were used and the mhz increased from these 3,3.6 or 3.7 to 4 and 5 and later even 6-8.
E.g. mkV 1981 was only 2 mhz. Constellation from novag 1983 was 2. Superstar36K from 1984 was 2 mhz but is overclocked in my tournament.
Rathsmann was 3.7 mhz . You maybe remember novag came out with a model called constellation 3.6 mhz. And super constellation was 4 mhz from the beginning.
There was a race going on, a competition between those companies Fidelity, novag, HG and scisys and therefore they had to put faster hardware and better software into their expensive machines.
We and the programmers profit from this competition. Because the end users got better hardware and software and the platforms formthe programmers also devloped into better platforms.
If you could win championships or tournaments you could make a better deal and the ICCA was the organisation were the deals were prepared.
Obviously the organisation to benefit the chess programmers was mainly an organisation for some to exploit the situation and make money.
TheSelfImprover and Tibono like this post
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Black suffers badly from the horizon effect - one of the biggest problems at that time.
Mclane likes this post
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: summer tournament Sun Sep 12, 2021 4:16 pm
The next game is Ed Schröders Rebel Mystery aka rebel amsterdam prototype vs. Chafitz steinitz encore with 4 mhz.
The Steinitz machine is by Larry Atkin. Larry Atkin
Normally it runs at 2 mhz but this version has 4 mhz to be competitive.
The hardware has no sensor board but keys instead and neither LEDs nor LCDs but a glass tube like it was in the early pocket calculators and arm watches.
Maybe you remember.
The machine can be made very tiny by sliding the display and the keyboard under the chess board. It was modular at that time and you were Able to buy other engines for it, a module called morphy, one called capablanca etc.
Even sargon 2.5.
TheSelfImprover likes this post
TheSelfImprover
Posts : 3028 Join date : 2020-11-18
Subject: Re: summer tournament Sun Sep 12, 2021 9:06 pm
Mclane wrote:
The next game is Ed Schröders Rebel Mystery aka rebel amsterdam prototype vs. Chafitz steinitz encore with 4 mhz.
The Steinitz machine is by Larry Atkin. Larry Atkin
Normally it runs at 2 mhz but this version has 4 mhz to be competitive.
The hardware has no sensor board but keys instead and neither LEDs nor LCDs but a glass tube like it was in the early pocket calculators and arm watches.
Maybe you remember.
The machine can be made very tiny by sliding the display and the keyboard under the chess board. It was modular at that time and you were Able to buy other engines for it, a module called morphy, one called capablanca etc.
Even sargon 2.5.
Looking at that Steinitz display, it looks like an LED display to me rather than a nixie tube (link).
Mclane
Posts : 2976 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: summer tournament Sun Sep 12, 2021 9:10 pm
Zooming in to 500%, it's clear that the display is an 8 segment display (standard LED displays were 7 segments for displaying digits).
Even more interesting, the "nixie clock" linked in the quoted text does not use nixie tubes either! It's using VFD (Vacuum Fluorescent Display), so calling it a "Nixie Tube Clock" is misleading.
See the images linked in my previous post to see what nixie tubes look like.
On a device made in the 1980s, you would expect them to use an LED display. However, they might have used an 8-segment VFD display so that letters could be displayed clearly. If so, maybe the extra cost could have been spent on higher spec CPU or more memory? Having said that, there is no shortage of products where appearance has been put ahead of functionality. You'd be amazed how much poor functionality can be forgiven when the appearance is good!
Home 
