Harware and software in chess

[Greyguy]

This rating list:
https://en.wikipedia.org/wiki/Swedish_C ... ssociation

Your opinion about hardware and software?
For Athlon 1200 software is clearly better year after year.
6502 is not so easy. What could 16 MHz mean instead of 4 MHz?
Compute is faster but they could need extra waitstate reading RAM or ROM.

[Dann Corbit]

This rating list:
https://en.wikipedia.org/wiki/Swedish_C ... ssociation

Your opinion about hardware and software?

Hardware gets exponentially better over time.
Software gets exponentially better over time.

For Athlon 1200 software is clearly better year after year.

And for every other platform.

6502 is not so easy. What could 16 MHz mean instead of 4 MHz?

Four times as fast.

Compute is faster but they could need extra waitstate reading RAM or ROM.

None of that is important.
What is important is that the SSDF tests interesting programs at long time controls until the Elo values of the programs are very well known.
For some of the programs, you have to sit and babysit, so it is a big effort.

Like every computer engine calibration list, it has its flaws.
But for long time control, it is one of the best and also a cornerstone of computer chess history.

IMO-YMMV.

[mclane]

I want to see stockfish 10 running on 1 MHz 6502 with 32 kB ROM and 8 kB ram.
How strong is it on such a hardware ?

[Greyguy]

This rating list:
https://en.wikipedia.org/wiki/Swedish_C ... ssociation

Your opinion about hardware and software?

Hardware gets exponentially better over time.
Software gets exponentially better over time.

For Athlon 1200 software is clearly better year after year.

And for every other platform.

6502 is not so easy. What could 16 MHz mean instead of 4 MHz?

Four times as fast.

Compute is faster but they could need extra waitstate reading RAM or ROM.

None of that is important.
What is important is that the SSDF tests interesting programs at long time controls until the Elo values of the programs are very well known.
For some of the programs, you have to sit and babysit, so it is a big effort.

Like every computer engine calibration list, it has its flaws.
But for long time control, it is one of the best and also a cornerstone of computer chess history.

IMO-YMMV.

Do you mean multiplication was so slow that it was totally dominating?
For x86 I have found IMUL 13-24 clocks for a 286 and today they are down to 3 clocks.

[Greyguy]

I want to see stockfish 10 running on 1 MHz 6502 with 32 kB ROM and 8 kB ram.
How strong is it on such a hardware ?

Should be problematic to make it so small. Why not 16bit DOS? It should be hard enough....

[mclane]

The idea is to measure how strong stockfish is on fossile Hardware the spracklen and others had to use.

This means same MHz, same maximum program size (32 kB) and same maximum ram size (8 kB).

[Uri Blass]

The idea is to measure how strong stockfish is on fossile Hardware the spracklen and others had to use.

This means same MHz, same maximum program size (32 kB) and same maximum ram size (8 kB).

Stockfish is not designed to be able to run on that hardware because it is more than 32 kB.

The question should not be about stockfish but if there is an improvement in the best software that can use the relevant hardware in the last years.
I do not know and I do not know if somebody developed in the last years some software that can use the hardware.

[smatovic]

This rating list:
https://en.wikipedia.org/wiki/Swedish_C ... ssociation

Your opinion about hardware and software?
For Athlon 1200 software is clearly better year after year.
6502 is not so easy. What could 16 MHz mean instead of 4 MHz?
Compute is faster but they could need extra waitstate reading RAM or ROM.

There was an Turbo-Kit for the 6502 with up to 20 MHz:

https://www.chessprogramming.org/6502#TK20

IIRC chess articles mention 1 ply deeper search, resulting in +100 Elo.

Maybe Ed Schröder has his one still up and running?

Myself tried to write an engine in C for the 6502 (Atari 800 XE, 1,77 Mhz, 64KB RAM),
and the result was pretty disappointing, about 10 nps (400 nps in perft),
a rewrite in 6502 assembly is still on my bucket list...

--
Srdja

[smatovic]

This rating list:
https://en.wikipedia.org/wiki/Swedish_C ... ssociation

Your opinion about hardware and software?
For Athlon 1200 software is clearly better year after year.
6502 is not so easy. What could 16 MHz mean instead of 4 MHz?
Compute is faster but they could need extra waitstate reading RAM or ROM.

There was an Turbo-Kit for the 6502 with up to 20 MHz:

https://www.chessprogramming.org/6502#TK20

IIRC chess articles mention 1 ply deeper search, resulting in +100 Elo.

Maybe Ed Schröder has his one still up and running?

Myself tried to write an engine in C for the 6502 (Atari 800 XE, 1,77 Mhz, 64KB RAM),
and the result was pretty disappointing, about 10 nps (400 nps in perft),
search depth 3 reached in 3 minutes from start position,
a rewrite in 6502 assembly is still on my bucket list...

--
Srdja

[smatovic]

***edit***

...search depth 3 reached in 3 minutes from start position.
--
Srdja