Author: Robert Hyatt
Date: 11:38:04 05/07/01
Go up one level in this thread
On May 07, 2001 at 12:27:15, Vincent Diepeveen wrote: >> >>Where are you getting this? They did _not_ do fractional ply extensions >>and this "consecutive move" extension is not in any writeup I have ever >>seen. >> >> >>> - 2 moves best (form of threat extension) 1/2 ply extended > >> >>Ditto. This didn't come from them. No idea where it _did_ come >>from. But I had already asked about fractional extensions a long while >>back and got a "no". > >This came from you Bob. > >> Didn't come from me. I have never mentioned them doing _anything_ except SE, recapture and out-of-check extensions. This is in all their papers they wrote. One of the team experimented with other extensions and published a paper in the JICCA. But those were _not_ part of the production deep thought or later deep blue. >>> >>>So for some reason they do loads of extensions show similar >>>lines as i do at about the same depth, from tactical reason seen. >>> >>>Sometimes they need 1 ply less to show it! >> >> >>Vincent: here is a problem with your statement above. I know a lot about >>what DB does. I know a lot about what Crafty does. Until you can beat >>Crafty regularly, please don't continue to point out how great your search >>is. If it can't beat my program consistently, even though you continually >>point out how bad my very simple q-search is, then how can anybody take you >>seriously when you continually point out how much better _your_ search is than >>Deep Blue's? It simply isn't credible. > >Well best compare with deep blue gives of course Gnuchess with singular >extensions and checks in qsearch. > >Deep Blue never played my program nor did they anyone else. I only >can conclude that any commercial program of today would beat them! To conclude that you need data. What data do you have? Commercial programs couldn't beat them on the deep thought hardware. Oh yes, they did lose _one_ game over 10 years. > >Just like Fischer would be beated blindfolded by any of todays 2650+ GMs. Not if you took the Fischer of 1972 he wouldn't. > >Crafty is exception here as it relies bigtime on search. If crafty >doesn't outsearch opponent then it's history as it never pushes >pawn otherwise and also tactical it goes wrong near tips. That is pure crap. Why don't you test a hypothesis before posting it? If you want examples I'll give them to you. As I have said _many_ times in the past, I will begin to notice your criticism of Crafty _after_ you begin to beat it on a regular basis. Since that hasn't happened, your argument hardly convinces me what I am doing is wrong or bad. > >Commercial progs in general are more aggressive tuned. > >Note i beat crafty with huge numbers at auto232 so do i with most >programs. Despite that i do not have learning. Please don't go there. Just come to ICC and do it in public sight. We can play on even hardware if you want. But don't start with the "I can beat everybody bigtime using auto232" until you can do it in public. > >But this really isn't the question here. I'm doing loads of >dangerous extensions, i'm doing a lot in qsearch, so that's very >comparable with what Deep Blue does. > >If we would take shredder or crafty >we have a serious problem comparing same search depth to DB: > >As both programs: > a) do nearly nothing in qsearch > b) shredder forward prunes in non-PV lines > c) c+s do very little dangerous extensions > >So that's hard to compare! > >GNUchess with singular extensions, recapture extensions and checks in >qsearch would be a great compare to deep blue. > >There is a bigtime improved gnuchess program... ...it's called Zarkov. > >Compare its searchlines with that of Deep Blue i'd say! > >You'll see an amazing similarity. Loads of bad moves which deep blue >made like the Qa5? Bc7? h6? g5? g4? all those moves also made by Zarkov! And that proves exactly what? > >The reason already described by Seirawan btw. Both DB and Zarkov count >queen mobility in the same way and both might receive bonus for bishop >covering queen... Yasser doesn't know a computer chess program (internally) from a tic-tac-toe program. Him speculating about scoring bonuses is really amusing... > >Best regards, >Vincent > >>>It makes no sense forward pruning 6 ply if you have hardware processors >>>that search 6 ply! > >>Then it makes no sense to prune _anything_. That statement has so many holes, >>I really don't want to take the time to explain why. If you are saying that the >>hardware search can't work with a selective search, that is nonsense. > >I didn't read anything from Hsu saying he has futility pruning inside DB! He stated this at several public presentations he made. This dates back to the original 10-0 match vs Rebel and Genius. He stated that "we had the futility pruning in the q-search disabled for testing..." Which was the first I had heard about it. Others heard the same thing and reported it here as well... > >You're only one who suggests it now! > Nope. Several others heard the same thing. In fact one of them reported that there weren't just 10 games played, there was a total of 40, if you recall. That wasn't me. >From my viewpoint all lines at n depth i can also produce at n depth if >i turn on the dangerous extensions in DIEP. > >So 11(6) i can reproduce at 11 ply. > >Easy. Peanut. > and your point would be? since you can't see all of DB's PV? >> >> >> >> >> >>> >>>Note that in that case we would see long mainlines which we do not >>>see. We only see tactical lines extended. Positional lines we can all >>>explain with the above extensions! >> >> >>I don't believe I have _ever_ seen anyone so hard-headed. One more time: >>The hardware will _never_ show a PV. It can _never_ do it. There is no >>way to get the last 6 plies + captures out of the hardware. So when you see >>11(6) you are going to see 11 plies of software search + possibly some >>extensions, and most of the time the PV is not even going to show all of that >>since the positions are scattered around in the memory of 32 machines, and >>reconstructing the PV is non-trivial when you have to deal with that. >> >>So get over the fact that their PV will _never_ be as long as yours or mine. >>It is simply irrelevant. Belle's PVs were _always_ exactly 2 plies long, >>unless the first move was a check, in which case its PVs were 3 plies long. >>Do you want to go back to 1980's hardware and show me how easily you could >>beat Belle since it obviously couldn't search more than 3 plies, even though >>they _say_ 9 plies??? >> >> >> >> >> >> >>> >>>Even with huge pruning it's impossible to search 17 ply. >> >> >> >>No it isn't. I just searched to 15 plies in the above position in a few >>minutes. On an Alpha I could _easily_ do 17 plies in that same time >>limit, I only need a factor of 8-10 in speed, and I could get a factor of 30+. >> >> >> >> >>> >>>FHR is a very dubious way of searching. It's very incorrect. I can't >>>imagine DB ever used it. >> >>Why the rambling? They didn't do this. It is not mentioned _anywhere_ >>in any writeup about them. >> >> >> >>> >>>FHR basically says next: >>> >>> IF evaluation + thread >= beta >>> THEN reduce depth 1 ply; >>> >>>This can only apply for half of the search depth *at most*. >>> >>>The only official search depth example i have seen from Hsu was >>>examining a 'typical' 12 ply depth search! >> >> >>If you mean Murray you should have asked him. When talking about search, >>they almost _always_ used the software search depth in discussions, without >>thinking. Since that was the "best" part of their search (using SE, recapture, >>etc) and since they could see the PV from that part of the search, that is >>what they did their debugging on. The chess procesors could be thought of as >>simple endpoint evaluation functions, except that part of the evaluation was >>a 6-ply exhaustive search + capture quiescence search + checks in the q-search.
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.