The future of computer chess

[Rom77]

Current hardware would mean 1 cpu since the older chess engines cannot run more then one core.
In a uniform platform tournament, so same hardware, one can try to find out.

I overlooked your reply.
Yes, single core Fruit 2.1 vs single core Stockfish, last release.
Fruit will be crushed.
Do you doubt this?

Already done:

[Rom77]

See also:

So somewhere around 100 nodes per second is likely where performance becomes superhuman. What kind of hardware would that be? It’s hard to say since modern versions of Stockfish would take a lot of work to get running on truly old hardware, if it is possible at all. But ignoring that, this user reported getting Stockfish 6 running on a 386 at about 1,000 nodes per second. On my machines SF 17.1 gets about 35% as many nodes per second as SF 6, so let’s say a 386 would run it at 350 nodes per second. That would still result in 3000+ play. Perhaps a 286 would run Stockfish 17.1 in the 100 nps range. Of course with 16-bit architecture and nowhere near enough RAM to fit the neural net, this would be pretty much impossible, but this experiment suggests that it really is ancient hardware like this we would need to reference if we want modern Stockfish to sink to the level of top humans.

[Rom77]

See also:

Minifish, NNUE chess engine that compresses to less than 64kb:
https://github.com/linrock/minifish

Winner FIDE & Google Efficient Chess AI Challenge:
https://www.kaggle.com/competitions/fid ... eaderboard

[Rom77]

See also:

ice4, is a chess engine which fits in 4096 bytes:
https://github.com/MinusKelvin/ice4

Winner TCEC Season 28 - 4K V MiniFinal:
https://tcec-chess.com/#div=4kf&game=1&season=28

CCRL Rating - 3151:
https://computerchess.org.uk/ccrl/4040/ ... 6_0_64-bit

[syzygy]

Current hardware would mean 1 cpu since the older chess engines cannot run more then one core.
In a uniform platform tournament, so same hardware, one can try to find out.

I overlooked your reply.
Yes, single core Fruit 2.1 vs single core Stockfish, last release.
Fruit will be crushed.
Do you doubt this?

Already done:

Thanks! And, wow... Fruit 2.1 from June 2005 against single-core pre-NNUE Stockfish from 2020, both on 2020 hardware, with 1000:1 time odds for Fruit 2.1, and Fruit 2.1 got crushed.

And Fruit 2.1 was not a weak engine at all!
One could still "blame" Fruit 2.1 though, because its release as open source certainly contributed to the progress in software that we have seen since 2005.

[mclane]

True.
But at least two 8 bit pioneers (ed Schroeder and Chris W.) demonstrated on todays hardware/ressources
that they are capable to create strong competitive and even interesting playing engines in todays technics.

Would the younger generations of programmers be capable to beat ed schroeders 8 bit 6502 computers if they would use similar ressources ? I highly doubt that.

Take a look here:

http://ssdf.bosjo.net/long.txt

354 Mephisto Nigel Short 6502 5 MHz 1950 ELO

This is a dedicated unit of ed schroeder using 5 mhz cpu 8 bit 6502.
https://www.schach-computer.info/wiki/i ... igel_Short


379 Novag Super Expert C 6502 6 MHz Sel 5 1868 ELO
Dave Kittingers strongest 8 bit 6502 engine
https://www.schach-computer.info/wiki/i ... r_Expert_C

393 Mephisto Modena 6502 4 MHz 1805 ELO
Frans Morsch strongest 8 bit 6502 engine.
https://www.schach-computer.info/wiki/i ... sto_Modena

399 Saitek Turboking II 6502 5 MHz 1777 ELO
Julio Kaplan’s strongest 8 bit unit.
https://www.schach-computer.info/wiki/i ... bo_King_II


400 Conchess Plymate Victoria 6502 5.5 MHz 1776 ELO

Ulf Rathsmanns strongest 8bit

https://www.schach-computer.info/wiki/i ... e_Victoria

404 Fidelity Par Excellence 6502 5 MHz 1745 ELO
Spracklens strongest 8 bit unit.
https://www.schach-computer.info/wiki/i ... Excellence

So where are today’s programmers if they had the ressources of the past ????

Choose assembler, c++ , c, pascal or whatever compiler.
But use the hardware ed schroeder made the most ELO out of 5 mhz.

If you would ask me to tell about ed schroeder, i would begin with showing you a bronze monument.
How good we still can ask him about those days.

1950 ELO Ed Schroeder was able to create out of the 6502 5 mhz 8 bit cpu 64/8 kb rom/ram.

How much can YOU do with today’s algorithms??

I guess that you can add at least 200 elo for Mephisto Nigel Short 6502 5 MHZ on the same hardware thanks to the knowledge we have today but somebody will need to pay for it because programmers are not going to work for something nobody is going to use unless you give them enough money.

The question is how much money do you offer for it.
I guess you need to pay at least the same money that Ed earned from Mephisto Nigel Short 6502 5 MHZ that was a commercial program.

I doubt ed earned a lot for the mephisto nigel short.
The Mephisto nigel short is mainly a mephisto milano with different Software.
Why did Mephisto sold this with the name nigel short:

Although I cannot be certain, I wonder whether the Mephisto Nigel Short machine emerged as a result of threatened legal actions between Jan Timman / Nigel Short against Hegener and Glaser (Mephisto) back in 1992.

Apparently back in 1989, Hegener and Glaser had offered a prize of 370,000 UK Pounds to the first Western player to challenge Kasparov. But in April 1992 (just after Timman had defeated Yusupov and Short was about to defeat Karpov in the world championship contender playoffs), Hegener and Glaser stated that the award had actually been withdrawn in 1991 - concurrent with the collapse of the Soviet Union. Hegener and Glaser had not, however, publicised the withdrawal of the award.

Both Timman and Short threatened legal action against Hegener and Glaser when it was clear one of those two players would be challenging Kasparov for the World Title. This culminated in a retreat by the company and the award was reinstated, though on altered terms.

One of those altered terms included the requirement for Short to endorse Mephisto products. Those terms included sporting mephisto logos on breast pockets, and (my guess), a chess computer named the Nigel Short.

https://www.independent.co.uk/news/uk/s ... 77682.html

Source:
Hiarcs chess forum:
https://www.hiarcs.net/forums/viewtopic.php?t=2541

As far as i see it mephisto milano plays a litte more passive then mephisto nigel short.
So the difference is mainly a playing style difference. Not really a big playing strength thing. Mephisto milano is also nearly as strong as the nigel short.

Hoping to take advantage of a commercial opportunity Mephisto released the Nigel Short at the time of the match. Perhaps there was also a deal in the background consequent upon Hegener + Glaser’s earlier withdrawal of a million Swiss Franc prize fund for the first Western challenger to play for the World Championship, and the threat of a law suit the withdrawal provoked?

Anyway the chess computer briefly sold quite well in the UK, but the manner in which Kasparov crushed Short would not have helped longer term sales. Within 12 months the price fell from £269, to £229, and then down to £199.

In the UK there are probably quite a few Nigel Shorts bought in that first year. Because they are useful machines for good club players the owners may not have considered selling them on eBay, as more often happens with chess computers of less practical value. Outside of the UK Mephisto Nigel Shorts are hard to find at all. This is a chess computer which is rare, but how rare?

The Nigel Short is based on the Mephisto Milano. I do not know the exact hardware and software differences between the two. I have seen comments by the programmer Ed Schroeder to the effect that an updated program was probably the only chess playing difference between the two machines. On the latest Selective Search list the Nigel Short stands 19 Elo points above the Milano.

The one you see was bought from Eric Hallsworth in July 2006. It is a version 2.00 machine. To get the program version press Memory then Info.

Source:
http://www.chesscomputeruk.com/html/mep ... short.html

[towforce]

Current hardware would mean 1 cpu since the older chess engines cannot run more then one core.
In a uniform platform tournament, so same hardware, one can try to find out.

To make it fair, you could run them all in the same simulator. However, if you built a simulator the 8-bit programs could run in, the new engines won't run in it.

[mclane]

I would like to see how today's programs run on 6502 or z80a or Motorola M6809,
and if they are capable to beat Ed Schröders engines from
1991
https://www.schach-computer.info/wiki/i ... sto_Milano
and 1993
https://www.schach-computer.info/wiki/i ... igel_Short

If an algorithm progress has been made, it should be possible to put it into 64 KB engine size.
It it doesn't fit into 64 KB ROM, 8 KB RAM it is maybe not a software progress but a hardware progress.

[chrisw]

Current hardware would mean 1 cpu since the older chess engines cannot run more then one core.
In a uniform platform tournament, so same hardware, one can try to find out.

I overlooked your reply.
Yes, single core Fruit 2.1 vs single core Stockfish, last release.
Fruit will be crushed.
Do you doubt this?

Already done:

Thanks! And, wow... Fruit 2.1 from June 2005 against single-core pre-NNUE Stockfish from 2020, both on 2020 hardware, with 1000:1 time odds for Fruit 2.1, and Fruit 2.1 got crushed.

And Fruit 2.1 was not a weak engine at all!
One could still "blame" Fruit 2.1 though, because its release as open source certainly contributed to the progress in software that we have seen since 2005.

I feel you are conflating software progress from algorithmic progress with software progress made possible by having large amounts of memory available.
For example NNUE is only possible with loadsa memory, likewise continuation histories and similar important stuff.
Conflation also with what is possible just as a result of speed. SF builds substantial position knowledge on the fly in the first milliseconds of searching which it then uses (and develops) for search guidance.
The question of this thread should be whether there is algorithmic knowledge in SF outside of algorithmic progress made possible by a) speed and b) memory. I would guess the answer is yes, but not that much, mostly it came as a result of hardware allowing it.

[flok]

For example NNUE is only possible with loadsa memory,

Exactly! I could barely fit it in 320 kB ram!

likewise continuation histories and similar important stuff.
Conflation also with what is possible just as a result of speed. SF builds substantial position knowledge on the fly in the first milliseconds of searching which it then uses (and develops) for search guidance.
The question of this thread should be whether there is algorithmic knowledge in SF outside of algorithmic progress made possible by a) speed and b) memory. I would guess the answer is yes, but not that much, mostly it came as a result of hardware allowing it.