Author: stuart taylor
Date: 05:50:47 02/27/06
Go up one level in this thread
On February 26, 2006 at 22:12:51, Jon Dart wrote: >A few notes from Arasan's games in CCT8: > >Arasan is playing on faster hardware this year and >also has gotten a significant speed boost (20% or so) >from being built with Visual C++ and Profile-Guided >Optimization (PGO). I tried the Intel compiler with >PGO, but did not get as good results. > >It has also had quite a bit of eval tuning in the >last year and now is using history pruning, besides >all the other pruning tweaks version 8.4 has. > >The first game was against "Crash Test Dummy", ctd, >which Arasan won decisively. ctd has no opening book but >played quite a reasonable opening. However, it allowed >too many pawn weaknesses. 24. b4 starts an attack that >opens up the position, gains material and leads to a >passed pawn that can't be stopped > >Arasan was smashed by Ikarus in round 2. Ikarus played >1. g3 and Arasan was out of book fairly quickly. Arasan >had a + score through most of the opening, but 31. >bxc5 was unexpected and after that it was in negative >scores. Ikarus pushed two pawns to the 6th, cutting off >the Black bishop and posing threats. The end was ugly >as Arasan tried to throw away material to stave off mate. > >witchess had Black against Arasan and played a weird kind >of Nimzo/QGA hybrid (it is another program without a book). >The game looked ok when I stepped out to run an errand, but >when I came back Arasan had lost. 33.. Rxa3 was surprising - >Arasan did not expect this, but also didn't see any big >problem from it. 57 .. f4+ was another unexpected blow, much >more decisive. Good game from witchess. > >I was still gone and didn't watch the game against >Chompster. Chompster played 1. d4 Nf6 2. f3, which put >Arasan out of book right away. Then it followed up with 4. >g4 and 5. g5, which might make Grob fans happy, but can't be >good. After 25. Rxf7 Qxf7 26. Rxf7, Black has two Rooks for >the Queen, which Arasan was happy to take. Arasan ate some >pawns and eventually pieces, to finish up with a big >material plus. > >Next round, jsbam played some strange Gruenfeld variation >and was in trouble right away (Arasan's score on move 7 was >0.5 out of book). Arasan won easily. > >Next game was Crafty. I remarked online that Arasan >has beaten Crafty in casual ICC games, but has never scored >against it in a computer chess tournament. I know Bob has >been tweaking the eval recently and that has probably led to >some decrease in playing strength. Crafty is still >incredibly fast - it was searching 4 million NPS or so (vs. >1-1.5M for me). But Arasan was getting more depth in some >cases (may not mean much because Arasan is fairly selective >and you can miss a lot that way). > >Arasan played the Meran variation and was in book until move >23. The Meran is full of opportunities for both sides >to go wrong, but against a properly booked opponent >it seems to be drawish (in my experience) and that >proved to be the case here. I had a little scare when >Crafty's eval went briefly above +1, but basically it >was a boring endgame, ending with draw by repetition. > >The next game was against Quark, which is a pretty strong >program. Arasan got an advantage soon out of the opening >(Ruy Lopez). 19. Nf5 looks strong and Quark failed low >at this point. At move 23, Arasan expected .. a5, but >Quark played .. Re6, and in the following tactics, Quark >lost a Rook for a Bishop. Arasan simplified down, but >in the end it had a rook file passer vs. Quark's one on >the c file. Quark was able to block the pawn and draw by >repetition. > >The 8th round Arasan was Black against Hossa. Hossa has >been challenging competition in the past, but now I think >Arasan is stronger. The opening was a fairly even English >Opening. I don't think Arasan would play 18. f4, which >opens up the King side. Also 28. b4 and 29. bxc5 open up >the b file, which Arasan winds up occupying. Arasan's >35 .. Rdb8 doubled Rooks on the b file and led to a >massive advantage. > >Against Glaurung, Arasan played a QGD Tarkatower. >I was not very happy to see this opening since Arasan >doesn't have a particularly good track record against >it. Arasan was out of book on move 10 (older versions >have 10 .. c5 as a book move but the current version >doesn't). I am not quite sure where Glaurung went wrong. >The culmination of a lot of piece manouvering and some >exchanges was that Arasan had an outside passer on the >b file, and this proved to be decisive. > >Overall I was glad to score 6/9, which is a personal best >for Arasan in CCT. > >I greatly appreciate Peter's organizing and directing >this. > > >--Jon I'm very happy for Arasan. I've been quite friendly with it especially due to the fact that it was easy (for me) to download, at difficult times, when I didn't have much easy computer access. Also, I got my first copy from a games CD, a few years ago. You seem to have now come up to the competition near the top. I used it alot, to play against me. I always found it very sharp, even at quick levels (atleast when it was against me, personally). The only problem was th at it was a bad loser! When it was losing, it started going bezerk and changing the colors around and changing a king into a pawn, etc. This has still been the case in very recent versions. regards S.Taylor
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.