Author: Amir Ban
Date: 17:15:01 11/04/99
Go up one level in this thread
On November 04, 1999 at 20:04:46, Fernando Villegas wrote: >On November 04, 1999 at 18:21:09, Amir Ban wrote: > >>On November 04, 1999 at 16:09:21, Peter Kappler wrote: >> >>>On November 04, 1999 at 12:49:49, Amir Ban wrote: >>> >>>>On November 03, 1999 at 23:12:31, Richard Bean wrote: >>>> >>>>>http://bbs.msnbc.com/bbs/kasparov-team/posts/hj/101641.asp >>>>> >>>>>"... I suspect that Krush's team used a computer >>>>>with computational capabilities comparable >>>>>to those of ``Deep Junior''. My guess >>>>>is that it probably was american program >>>>>``Ferret'' which just like ``Deep Junior'' >>>>>works off 4 processors. My assumption >>>>>here is based on the queen ending analysis >>>>>that we've seen. It's comparable in >>>>>size and quality to ours..." >>>>> >>>>>Hey, is that true? I've seen LOTS of >>>>>people ask ``When is Ferret going to be >>>>>sold to the public''... perhaps it needs >>>>>a web page detailing the predicted release date, >>>>>and then have the web page address added to >>>>>some CCC FAQ. The "message ferret on ICC" system leaves >>>>>the date quite uncertain... >>>>>The other thing is that I found the mate in 25 >>>>>announcement from Deep Junior after 62...Qc6+ 63. Kg5 quite >>>>>incredible. I had crafty with 768Mb hash & 486Mb egtb >>>>>cache run for 96 hours on that position KQQKQQ + 5 piece TBs, no mate, >>>>>just failing low with a +9.96 eval for White. >>>>>I suspect that Deep Junior had "63...Qd5+ 64. Qf5" >>>>>added to the position as well. >>>>> >>>>>(since DJ allegedly had 512Mb of RAM, and the computer >>>>>I was using only benched at ~135000 NPS. Still, >>>>>the large amount of hash should make up for it.) >>>> >>>>Kasparov probably knows about Ferret from Paderborn. He was following at least >>>>some of it. I overheard a conversation between him and Alterman the evening >>>>after Junior-Nimzo. Boris asked him if he saw it, and Kasparov said yes, even >>>>commented that the sacrifice was "safe", meaning that white has at least a draw. >>>> >>>>Boris Alterman was at Paderborn as a journalist covering the tournament for Club >>>>Kasparov. They paid for his trip. >>>> >>>>Amir >>> >>> >>>Amir, >>> >>>Thanks for the info. I'm curious about the version of Junior that will play in >>>the Club Kasparov Grand Prix, beginning December 18. >>> >>>What hardware will it use? >>> >> >>It will come from Club Kasparov. There's a Quad 500 Pentium III Dell machine at >>the CK office at Herzlia. I think that will be it. >> >> >>>Will it be the standard Junior 6, or a special development version? >>> >> >>Standard Junior 6 is not out yet, but will be out between now and the >>tournament. We'll use the best version we have available at the time. I don't >>have any special Anti-GM feature. Actually, now that you mention it, I'll give >>it some thought. >> >> >>>Will it use a special opening book? >>> >> >>There will be opening preparation within the standard J6 book. >> >> >>>Does Kasparov have access to an early version of Junior 6, or did he get >>>analysis from you for his game against the world? >>> >> >>I don't know what took place there, but Shay is the development manager of Club >>Kasparov, and Boris Alterman is working there too. They have early versions of >>Deep Junior and Junior 6 of course. If Alterman was analyzing the game for >>Kasparov, it would be natural for him to use Junior. >> >> >>>What are some of the major differences between Junior 5 and 6? SMP support? >>> >> >>Good timing: I've just this evening frozen the new version, so this is a good >>time to talk about it. >> >>There are two products, Junior 6 and Deep Junior 6. The difference between them >>is SMP support. >> >>The other differences over Junior 5, are: >> >>- Many changes in evaluation function, whose effect on evaluation & style ranges >>from subtle to major. >>- New book prepared by GM Boris Alterman. >>- No clearing of hashtables between moves. This is good for searching, but the >>major benefit is in analysis, as Junior is able to "learn" a position as you >>step through variations. There's a "clear hash" engine parameter and a "clear >>hash" menu option if you want the old behavior. >>- Tablebase support. >>- Some search algorithm improvements. >>- Underpromotion support. >>- There's of course the new GUI that comes from ChessBase, with a new look, and >>many new features and options. >> >>Amir > >Hi Amir: >Let me add a couple of questions to those of Peter. I have not a dual processor >computer; muy very best one is a decent but not top-notch Armada with around >100Mg in Ram and 266 Mhz. My question is: what can I get from Junior 6 with this >machine? It will be apreciable stronger -let us say, 50 to 75 elo points- than >my actual J. 4,6? And what is more, will give me more chances to improve my game >with features adequate to that task? I mean, aditional features to what >Chessbase products actually gives? Or said in another word: I will be beaten >just because superior tactical ability as any top proam an do or I will see a >positional lesson, a lesson also in endings -if I get there-, a lesson in >everuthings hat counts? >Or just I am asking too much? :-) >Fernando the sadomasochist... Hello Fernando ! It's a big two-year step from J46 to J6. Compared to Junior 4.6 I can safely say you will see a 50 elo points improvement. I can't be sure about 75 elo, but I think yes. It will certainly play positionally better. It will also be faster, J4.6 was 16-bit, later engines are 32-bit. Amir
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.