Even though it LOOKS like black has drawing chances, all the engines give white a winning advantage and a bunch of Shootouts all gave white the win, so it was time to resign and concentrate on the remaining games. The three draws resulted from, as white, an Evans Gambit and a Urusov Gambit (!) and as black, a Guioco Piano. With any luck, I expect to go +2 -0 =1 in the remaining games, but a result of +2 -1 =3 with Komodo won’t be very convincing since it’s about in line with my overall results prior to Komodo’s arrival on the scene. Many more games will have to be played to get a more accurate picture. According to all the engine ranking sites though, Komodo 8 is by far the best.
The Komodo developers claim it is different from the rest in that its search allows it to often see deeper than any other engine and its evaluation differs because it represents a blend of both automated tuning and the judgment of a grandmaster and computer expert, GM Larry Kaufman.
Komodo is primarily known for excellent positional play which is important if you are playing on ICCF or LSS where engine use is, if not mandatory, both allowed and advisable. Most of the top engines excel in tactical strength, but the programmers have sacrificed positional play so their engines will score better on tactical problems and do well in blitz play against other engines. In other words, they have stacked the deck for rating list purposes.
As Komodo developers point out, the judgment of a GM is still superior to engines so it makes sense to emphasize positional play rather than tactical skill. Komodo is especially useful for opening analysis because Kaufman has made sure that the program's evaluations agree in general with accepted theory. Komodo also excels in the evaluation of positions with material imbalance, which it handles more correctly than other top engines. This is a very important factor because my experience has lead me NOT to trust engines in this situation. It seems they often get it wrong.
On the CCLR 40/40 rating list Komodo 8 pounded Houdini 4.0 as did Stockfish 5, so we can discount Houdini 4.0 as not being worth the purchase price of about $50. So, the real question is, is Komodo 8 worth the $60 purchase price as opposed to the free Stockfish 5?
On the CCLR 40/40 list, Komodo 8 hammered Stockfish 5 +19 -8 =67. On the other hand on the CEGT 40/120 list Stockfish 5 heads the list followed by Komodo 8 and Houdini 4.0, and the matchup of Stockfish vs. Komodo 8 is almost equal.
In answering the question as to which is best, the $60 Komodo 8 engine or the free Stockfish 5 engine, it depends on what you are going to do with it. If you are SERIOUS about playing on ICCF or LSS and have 8 or more cores and are willing to devote a tremendous amount of time to preparing your own well researched opening books and spend days analyzing positions, then by all means spend the money for Komodo 8.
If you’re like me and only have a quad core laptop, play correspondence for fun and like to experiment with different openings (I have played the Sicilian Wing Gambit, the Urusov Gambit, the Grob and stupid stuff like 1.a4 and 1…a4, none of which have been outright refuted!!) then stick with Stockfish 5.
11-1-14 UPDATE: In a match conducted last month on an AMD FX-8350 8-core with 4GB hash per engine and the Syzygy 6-men tablebases and a long time control of 90 minutes plus a 30 second increment, the match was tied at +16 -16 =68. So it appears my conclusions are...WRONG!! There really does not appear to be any need for most of us to spend the money for Komodo. Details of the match HERE
No comments:
Post a Comment