A modern chess engine on a 1990s Pentium PC takes on Deep Blue

[Southern Joanne]

Could a home computer from the 1990s have beaten Deep Blue? Hackaday reports on an experiment by Nicholas Carlini, who ported the Crafty and Stockfish chess engines to a 200MHz Pentium Pro with 64MB of RAM, a high end PC from around 1995

He tuned them with an AI coding assistant, testing compiler flags and changes to the algorithms. The goal was to see how much modern software could squeeze out of old hardware

Deep Blue was rated at about 2700 Elo, while modern Stockfish is rated around 3700. Deep Blue beat Garry Kasparov in 1997 using huge custom hardware

Carlini concluded that the old PC could likely have beaten Kasparov and Deep Blue in at least a few games, although that cannot be verified. The striking idea is that the hardware was good enough all along, and the algorithms simply had not been invented yet

This is a fascinating reminder that better software can matter as much as faster hardware. Chess engines have improved enormously through smarter search and evaluation. It makes you wonder what today's computers could do with tomorrow's algorithms

Does it surprise you that software made such a difference? And do you still play chess against computers?

Pitting A Modern Chess Engine On A Pentium PC Against 1996's Deep Bluehackaday.com

[VoidStalker80]

This applies far beyond chess. Better algorithms often beat bigger hardware. AI research shows that too. Smart ideas win in the long run. Raw power only gets you so far

[BlackSunLynx]

Kasparov losing to Deep Blue felt like the end of an era in 1997. Now we know a desktop might have done it

That changes how I see the match. The real story was the software. Hardware got the credit. History is funny like that

[Sandman30]

Easy settings are where I still play chess against computers. They crush me otherwise

Stockfish at full strength is unbeatable for humans. Nobody can compete now. Engines are learning tools now