Discussion
Someone put a modern chess engine on a 1996 Pentium Pro, and he thinks it would have beaten Kasparov...
IBM says Deep Blue "was able to evaluate 200 million chess positions per second," and Nicholas Carlini argues a desktop PC from that same era could have beaten Garry Kasparov anyway. In a November 1 post, he describes the machine sitting next to him: a 200 MHz Intel Pentium Pro with 64MB of RAM, running a custom port of Stockfish.
The ratings he gives are lopsided.
Carlini says "the majority of the Elo gains (+720 of the +1335 gained) can be attributed to better programming, and not better hardware," and his conclusion is blunt: "we could easily have beaten Kasparov without building Deep Blue, using only commodity hardware." IBM remembers 1997 as the year Deep Blue "became the first computer system to defeat a reigning world chess champion in a match under standard tournament controls."
Not everybody buys the clean split between software and hardware.
One Hacker News commenter argued that "a lot of that better programming can be attributed to better hardware, it wouldn't have been possible to implement NNUE or Fishtest without modern hardware." That's fair, because the ideas were tested on fast modern machines even if the finished engine fits on an old one. Another just wrote "I had a Pentium Pro 200 back in the day," and I felt that.
The same thinking applies to the computer in your closet.
Before I replaced an old machine that feels slow, I'd try newer, lighter software on it first, because what it runs can matter as much as the chip inside it.
Kasparov lost to a supercomputer. He might have lost to a beige box.
Carlini estimates his port plays "at an estimated Elo of 3300 ± 87 at classical time controls," far past Kasparov's "Elo of just under 2800," and he writes that "Deep Blue was rated 2700." That would make the beige box next to him the strongest player of the three, by a mile.
The improvement lives in the code.