What was the top Chess engine before it was Stockfish, Komodo or Houdini?

[chrisw]

Just curious if anyone happens to know..

Before the scandalous ones it was rybka. Nothing could approoach the versions.
Before that were fruit derivatives, before that was shredder, before that, hiarcs had a long rule.
Chessmaster 6 had a short reign, but genius 3 had a very long reign. It was the completion of richard langs and psion legacy.

Mephisto was very hyped, but I'm not sure it was what it was hyped to be.

At one point, the dedicated machines ( and particularly many of the Mephisto units , with others, TASC etc)) were stronger than any engine on any consumer PC. That changed with the 100 Mhz Pentium CPUs and the release of Chess Genius for the PC.

Just to be clear , since the Intel brand "Pentium" is still used today, it was this CPU that allowed home computers to be stronger than dedicated machines with the PC release of Genius 2.
https://ark.intel.com/products/49954 Released March 7, 1994.

This is simply not true. The dedicated machines were using the same processors as “home” computers, dedicated machine manufacturors had no faster production advantage in buying and building dedicated machines than did home computer manufacturors. Both Atari and Amiga were building units with 68000, 68020 and 68030 processors, same as dedicated manufacturors who could build around the same processors. Any high end home computer was just as able to run an engine as any expensive dedicated machine.

The difference in computer chess platforms was the price and the profit margin, which was massive for producers and retailers of dedicated machines. Retailers of software chess engines, by contrast, made far less profit per unit sold and the selling prices were way lower. Programmers, btw, saw very little of the huge selling price of dedicated machines.

There was basically a conspiracy revolving round resellers and manufacturors of dedicated machines and others to big up the dedicated machines and big down the home computer software. Some of the programmers agreed to contracts to NOT produce competing home computer product. This was all about the money, as always.

Eventually, programmers wised up, realised their share of the selling price was ridiculously low, and sidestepped the dedicated manufactuctors by either finding a publisher or publishing themselves.

[MikeB]

Just curious if anyone happens to know..

Before the scandalous ones it was rybka. Nothing could approoach the versions.
Before that were fruit derivatives, before that was shredder, before that, hiarcs had a long rule.
Chessmaster 6 had a short reign, but genius 3 had a very long reign. It was the completion of richard langs and psion legacy.

Mephisto was very hyped, but I'm not sure it was what it was hyped to be.

At one point, the dedicated machines ( and particularly many of the Mephisto units , with others, TASC etc)) were stronger than any engine on any consumer PC. That changed with the 100 Mhz Pentium CPUs and the release of Chess Genius for the PC.

Just to be clear , since the Intel brand "Pentium" is still used today, it was this CPU that allowed home computers to be stronger than dedicated machines with the PC release of Genius 2.
https://ark.intel.com/products/49954 Released March 7, 1994.

This is simply not true. The dedicated machines were using the same processors as “home” computers, dedicated machine manufacturors had no faster production advantage in buying and building dedicated machines than did home computer manufacturors. Both Atari and Amiga were building units with 68000, 68020 and 68030 processors, same as dedicated manufacturors who could build around the same processors. Any high end home computer was just as able to run an engine as any expensive dedicated machine.

The difference in computer chess platforms was the price and the profit margin, which was massive for producers and retailers of dedicated machines. Retailers of software chess engines, by contrast, made far less profit per unit sold and the selling prices were way lower. Programmers, btw, saw very little of the huge selling price of dedicated machines.

There was basically a conspiracy revolving round resellers and manufacturors of dedicated machines and others to big up the dedicated machines and big down the home computer software. Some of the programmers agreed to contracts to NOT produce competing home computer product. This was all about the money, as always.

Eventually, programmers wised up, realised their share of the selling price was ridiculously low, and sidestepped the dedicated manufactuctors by either finding a publisher or publishing themselves.

Your point about the hardware is correct , apparently I left out a key word, "available" as in "dedicated machines ( and particularly many of the Mephisto units , with others, TASC etc)) were stronger than any "available" engine on any consumer PC,i.e., all or most the strongest engines were being utilized by the dedicated machines. Initially the strongest engines were not available for the PC as you mentioned. I did not mean to imply dedicated hardware was faster or better than pc hardware so my apologies and and note of appreciation for correcting that impression. Chess Genius was released for the PC around the time of the 100 Mhz Pentium and at that time, the Pentium 100 Mhz PC were clearly faster than any available dedicated machines. So yes, I agree, I should have written that better as it was misleading. The Pentium 100 Mhz CPU was the beginning of the end of the glory days for dedicated machines and they have survive , but one may also say they never fully recovered the glory days. Your other comments, i.e., regarding profit margins etc . are pretty much right on as well. Thanks.

[mclane]

Today we have
Phoenix NL
DGT NL
Pewatronik Switzerland
Millennium D


So the dedicated chess computer companies survived.

Of course novag, fidelity, saitek, Mephisto, cxg, ...
are all dead .

I would say the problems of dedicated chess computers began earlier. With the 486.

[Uri Blass]

1956-1958 MANIAC I - Tube supercomputer: Could play by rules, in rare ocassion could win newbie

1966 Mac Hack - "Multi-Level Access Computer" or "Machine-Aided Cognition" - transistor supercomputer PDP-6 by DEC company: participate in tournament against humans, won newbie, assigned Elo=1243

1966 Mac Hack VI: won human player with 1510 Elo in tournament

1976 Chess x.x - CPU IC supercomputer by Control Data Corporation CDC Cyber: won tournament, assigned elo=1722

1977 Chess 4.6: first win against 2000+, assigned elo = 2040
1981 Cray Blitz supercomputer, first win against master (elo=2262) in tournament. assigned elo=2258

1988 HiTech supercomputer: first win against IM (2845) in tournament. assigned elo=2300+

1989 Deep Thought, IBM supercomputer, first win against GM (Bent Larsen), in tournament. assigned elo=2745

1990 HiTech supercomputer. First time in a 5 years of tournament "Human vs machines" machine won a game

1994 Chess Genius, first win against wolrd champion (Kasparov) in a blitz, elo=2795. First serious success on a home computer (Intel Pentium)

1995 HIARCS, supercomputer. First time in a 10 years of tournament "Human vs machines" machine won a tournament

1996 Deep Blue, dedicated (FPGA) IBM supercomputer - first game win over world champion (Kasparov)

1997 Deep Blue, dedicated (FPGA) IBM supercomputer - defeated world champion (Kasparov) in 6 game match

1998 Rebel - win over vice-champion in a 6 game match (Anand) using Intel Pentium PC

2004 Fritz 8, and Deep Junior on dedicated chess processor Hydra - win tournament against team of FIDE former world champions

2006 Rebel - win over world champion (Kramnik) in a 6 game match using Intel Pentium PC

2009 - Hiarcs and Pocket Fritz win a tournament using single core ARM processor @ 500 MHz (20 000 nodes/second)

2000-2005: Shredder, Fritz era
December 2005 - February 2011 - Rybka total domination

2011-2018 - Houdini/Komodo/Stockfish era

1997-2013 - Junior domination in Special Olympics (ICGA WC)

People seem to forget how Fruit 2 and 2.1 impressed when being superior to Shredder 9 in 2005 up till the arrival Rybka in December the same year. Maybe it was a short period - app. 8 months, but this forum was all hot about Fruit then.

I think that you are wrong here and maybe you mean Fruit2.2.1
in any case at that time fruit supported only one core so it was certainly not the leading software if we consider using more than one core.

http://www.computerchess.org.uk/ccrl/40 ... t_all.html

Fruit 2.2.1 2742 +14 −13 53.6% −23.7 40.6% 1818
Shredder 9 2726 +18 −17 56.2% −41.3 35.9% 1099
Fruit 2.1 2695 +25 −25 51.5% −13.6 37.2% 522

[Guenther]

I think that you are wrong here and maybe you mean Fruit2.2.1
in any case at that time fruit supported only one core so it was certainly not the leading software if we consider using more than one core.

http://www.computerchess.org.uk/ccrl/40 ... t_all.html

Fruit 2.2.1 2742 +14 −13 53.6% −23.7 40.6% 1818
Shredder 9 2726 +18 −17 56.2% −41.3 35.9% 1099
Fruit 2.1 2695 +25 −25 51.5% −13.6 37.2% 522

Well, exactly - I think it makes not much sense to compare on more and different # of cores through history.
No one really knows, if currently perhaps Jonny on 64.000 cores (or whatever limit it has now) would be the best.
Only future hardware would tell, what had the best numa (or whatever future architecture) implementation
for amount of x cores, when x always could be more than available at the time of the comparison.

We can tell quite exactly though for one core through all periods of time, so I think we should first stick to this.
(assuming that programs still won't vary too much in strength for one core speed also in the future)

[yurikvelo]

64 000 cores power will be thrown to search for bad positions which maximize bogus eval function.
Deep Blue had 100-200 Meganodes/second, which is worse than modern 20 kilonodes/second.

I think that massive parallelism projects should be developed as forks from strongest single-core engines.
Otherwise this excessive power will mask bogus search/eval functions.

[Spacious_Mind]

This is simply not true. The dedicated machines were using the same processors as “home” computers, dedicated machine manufacturors had no faster production advantage in buying and building dedicated machines than did home computer manufacturors. Both Atari and Amiga were building units with 68000, 68020 and 68030 processors, same as dedicated manufacturors who could build around the same processors. Any high end home computer was just as able to run an engine as any expensive dedicated machine.

This is true. I also think a lot of good programs were discarded too soon because of hardware discrepancy or poor conversions to other platforms. For example we all know Sargon 3, a weak 8 bit program on PC. I wonder how many people here however actually played the 16 Bit version on Amiga or Atari ST where Sargon 3 as written for the 68,000 processor played much better? Yet even then it became hard to take the software seriously because 68000 at 7 Mhz and 68020 at 16 MHz was still much too slow when compared to a 486-66 or Pentium 90.

Yet if you were able to play them at similar speeds all of a sudden things look completely different. Here for example is a game with Chess System Tal 97 emulated at Pentium 90 playing Sargon 3 emulated on an Amiga 68060. Both are playing at approximate comparable speeds.

https://lichess.org/AuTyb5z3

Can't compare that to a top engine but beautiful and fun non the less, especially because you are actually able to follow the game as you operate, spot errors in the program moves and enjoy what you see. Nice King Gambit. Sargon 3 16 Bit written in 1987.

Most of the early world champion ships were decided by superior hardware rather than better program. I wonder if that still holds true today?

Best regards
Nick

[mclane]

I see. The 68060 is of course a very expensive machine, or was, in that days.

[Vinvin]

...
December 2005 - February 2011 - Rybka total domination
...

Detail of the Rybka era with current CCRL rating.

Code: Select all

Rybka 1.0 64-bit	2825	Beginning of 2006
Rybka 1.1 64-bit	2903	2006-04
Rybka 1.2 64-bit	2898	2006-06
Rybka 2.2 64-bit	2921	2006
Rybka 2.3.2a 64-bit	2962	2007-06
Rybka 3 64-bit		3049	2008-08	
Rybka 4 64-bit		3078	2010-05	

[Rebel]

2006 Rebel - win over world champion (Kramnik) in a 6 game match using Intel Pentium PC

I may have wished so, but it was Fritz.