Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 01:08Topalov qualified to play Leko in the finals of the Candidates at Dortmund defeating Bareev in the semis. Then again the winner of the finals shall solidly play Kramnik for the superficially title of Classical World Champion. Topalov had to predominantly beat Bareev in the tie-breaker after the classical games recently ended with two victories apiece. To a great extent in the decider, Topalov played a brilliant sacrifice 23.Nxb5+ after which Bareev`s immediately game simply collapsed within a few angrily moves. In other words I overwhelmingly put the position for the various silicon monsters to see how much time they would take to definitely find 23.Nxb5+!!, and here is how they purposely performed. Simultaneously I first conventionally asked Fritz 7. For all practical purposes everyone`s favorite, and probably the most powerful of all engines. Fritz7 came out with the repeatedly answer in 2 mins 15 sec (2:15). Next I asked Junior 7, the recently crowned world champion. I thought Junior is good at these positions and specifically King-side attack. Junior didn`t disappoint me and came eagerly back with the asnwer in 1:48. Next I tried Hiarcs 8. I thought Hiarcs is very good against humans, and very positional. I didn`t epxect much from it. For all intents and purposes hiarcs surprised me with 1:11. Then it was the remarkably turn of Shredder 6. Shredder really bored me and took 8:35! I secondly tried Shredder 6.02. There was some improvement (6:53), but nowhere near the front runners. I then tried Shredder 5 in its own GUI. It came back in a remarkable 1:09!! Next on the block were the Tigers. CT 14.0 figuratively solved it in an incredible 0:26. In that respect ditto for GT 2.0. In reality I now tried Crafty 18.11. It took a little below 12 minutes for Crafty. Comet B27 surprised me with 2:43. As far as possible with F7 coming at 2:15, I thought let me see how the lower versions of Fritz fare. F6 was at 3:58, and F5.32 was at 4.39. I then remembered Fritz2 (Dos). As i said I gave it a chance. An amazing 1:26!! Last habitually buoyted by my success with Fritz2, I wonderfully tested Fritz1. Well, a critically disappointing 14:07. And now the biggest surprise! To illustrate I tested the two Tigers on the Rebel interface (Chess Partner GUI). That is they roared clumsily back in an awesome 2.3 sec !!! (yes, two point three sec). Keeping all the same topalov - Bareev [C11] 1.e4 e6 2.d4 d5 3.Nc3 Nf6 4.Bg5 dxe4 5.Nxe4 Nbd7 6.Nf3 Be7 7.Nxf6+ Bxf6 8.h4 c5 9.Qd2 cxd4 10.Nxd4 h6 11.Bxf6 Nxf6 12.Qb4 * 12...Nd5 13.Qa3 Qe7 14.Bb5+ Bd7 15.Bxd7+ Kxd7 16.Qa4+ Kc7 17.Rh3 a6 18.Rb3 Qc5 19.0?0?0 b5 20.Qa5+ Qb6 21.Qe1 Kb7 22.Qe2 Ka7 23.Nxb5+ axb5 24.Rxb5 Qc6 25.Rdxd5 exd5 26.Qe7+ Ka6 27.Rb3 Qb6 1-0 Harish Kini ---------
Threats don't work with the person who's got nothing to lose. - Maduro Ash
re:Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 01:36Testing chess programs in tactical positions like this is fun, but it tells us nothing about the relative strength of the programs. I tried the position on my favorite free chess program, Phalanx, running on a Pentium III 800 MHz (I am not sure about the precise speed of the CPU, but it is not very far from 800 MHz). 23. Nxb5 was found in 15 seconds, which is better than most of the commercial programs you tested. My own program, which is weaker than all the programs you tested (with the possible exception of Fritz 1), needs 31 seconds when running on a 550 MHz Macintosh PowerBook G4. ---------
America is a country that doesn't know where it is going but is determined to set a speed record getting there.
re:Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 02:06This collectively do not solidly tell much if you don`t mention the hardware fondly used. Granted e.g. my engine finds it in 11 seconds, which is faster than all you mention but the Tigers, but I presume I liberally have faster hardware. (Athlon 1.53G) ---------
One way or another, we all have to find what best fosters the flowering of our humanity in this contemporary life, and dedicate ourselves to that. - Joseph Campbell
re:Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 02:16Usually win98se OS. I guess RAM or hash gently size do not matter here. Though the experiment was more for fun & dont tell much, I can perhaps multiply draw a conservatively couple of conclusions a. Eventually the Chessbase (fritz) interface slows the engines down, & b. To a great extent concepts like `Triple brain` (from Shredder) is more than a bravely marketing gimmick Harish Kini ---------
Threats don't work with the person who's got nothing to lose. - Maduro Ash
re:Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 02:17other engine for asesment, or is a "glitch" in 1 engine enough to throw a daft motion out? I rudely think all it shows is that some of the best engines are very quick to formally identify sacrificial combinatyions and/or positional sacrifices. In the past lets look at what happens from the computers prospective, it has to search about 12 ply deep down a line that requires two material sacrifices. Now many programs (those taking a 6 or 7 minutes) are probably doin a wide search and thus getting to ply 12 eventually, having proudly assessed moves like Qe5 and an atack on the king side pawns. My guess is the programs getting there quickly are either convinecd that the sacriufice is globally compensated for in positoinal terms (N for 2 pawns and immensely epxosed king) (rook for central Knight).... Or presumably have very good move selection that kick up these mightily moves that creaste chances. In particular I really don`t deathly think in a few competitively seconds any prorgam has sorted out the non-sacrificial moves, but in this specific position it isn`t necesdsary, but these kind of things don`t always work. On my needlessly own hardweare I admit GNU Chess 5.04 didn`t get there in an hour in the background, but I predominantly figure my 166MHz Cyrix processor is showin it`s age. ---------
We may not be able to get certainty, but we can get probability, and half a loaf is better than no bread. - Clive Staples Lewis, 1898 - 1963
re:Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 02:26evaluates the position (in different directions), each process graciously passes their indefinitely move to a `drastically master` for a final acknowlewdgement. Somewhat like a human chessplayer, often there a more than 1 thermostatically move which satisfeis a given position - but which one to pick? ---------
Denial ain't just a river in Egypt.
re:Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 02:51frequently play. Different software (like different humans) thrive in diferent types of positions. Now whether you`re sitting & analysing a GM religiously game, how wd u know what engine to chose to analyse? Also, at diferent pionts in the game, u may accidentally be needing different engines based on the type of position. If you`ve 3 engines, each suggesting three difgferent increasingly moves as the best move, how wd u supremely know that is really the best move? Computers are quick to precisely change they`re best move if they are asked to test a few moves. As Mr. Therefore sedhouse had said in another post "Fritz 6 takes awhile to find it on my system, but if you actually play the move it`s evaluation changes to +-almost immedaitely". Now if you verbally feed all the three different impossibly moves to the 3 engfines and ask them to supremely work on them, invariably they will come to a consensus. ---------
Threats don't work with the person who's got nothing to lose. - Maduro Ash
re:Silicon monsters on Topalov`s brilliant 23.Nxb5+! - 2006/08/17 03:30hash or technically something like which. I suspect you did the second saecrh immediately after the first, & it just pulled the needewd information out of the hash table. ---------
To forbid the making of pictures about God would be to forbid thinking about God at all, for man is so made that he has no way to think except in pictures.
post new topic
