Ras wrote: ↑Sun Apr 21, 2024 12:42 am
mclane wrote: ↑Sat Apr 20, 2024 11:44 pmE.g. neuronal nets are possible because fast graphic cards make progress.
You are aware that NNUE engines such as SF don't even use GPUs?
Yes. Nnue. But nn use them. And try nnue on 6502

)
My point was that it was the hardware possibilities that pushed the todays progress in „software“.
With only 8 bit hash is difficult. Old cpus were not capable to use big data.
They could not calculate with more then one core.
The cpus were mainly limited concerning RAM and ROM. It was KByte and not Megabytes or Gigabytes.
Try reloading data when the machine has only a cassette tape.
Reload tablebases or other overlays when the machine has only 5 1/4 floppys with 360 Kbyte data.
Nevertheless David Broughton did use overlays for his Philidor PC engine.
And Martin Bryant colossus chess used an opening book on disk that enhanced itself after the games.
This sums up in: most „software“ progress is in fact hidden hardware progress.
What seems like a fairy tale today may be reality tomorrow.
Here we have a fairy tale of the day after tomorrow....