It took a while between Kasparov losing to Deep Blue, and Rybka 4, the first true "super" chess engine to really hit the market, and computer technology in general massively overtook the super chess computer that Kasparov was playing against. I suppose they might get to 5nn, but that's about the practical limit, if it's even possible to get below 7nn, which is really getting close to the hypothetical limit. Someday, don't know when, a cloud verson of a NN engine will play at the level of Alpha Zero Chess or even better, but it will be several years at least, perhaps a decade or more, unless there is a fundamental change in CPU chips, because the industry is getting close to not being able to shrink silicon much more. I do think in a few years, single computer NN engines will overtake regular chess engines, but not at a level that pushes chess theory anymore than a regular super engine would. I don't see them being viable without some sort of cloud funcion, because a single personal computer, at this time, can't even scratch the amount of resources it needs to play at a level that pushes the boundaries of chess theory. I think neural network (NN) engines will eventually overtake regular engines. The issue with Stockfish is that it's open source, so you don't have to wait too long before someone releases a derivative of the program that plays a bit better than whatever other Stockfish version or more recent derivative that was the current toast of the town. It's pretty high in the rating, but probably a bit dated by now.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |