Author: Robert Hyatt
Date: 08:01:14 06/05/99
Go up one level in this thread
On June 05, 1999 at 06:37:48, Vincent Diepeveen wrote:
>Their is a weird account at ICC called DeepViolet.
>The facts about it:
>
> a) the operator is a nice fellow
> b) when it entered icc many years ago
> it had the same rating like scratchy (deep blue). didn't go up
> much since then.
this isn't correct. Check this out. I have _every_ ICC game crafty/scrappy
have ever played. Here is a 1996 game:
[DeepViolet (2358) vs. scrappy (2177) --- Mon Nov 25, 04:13 PM EST, 1996]
This was at a time when crafty was 2550 (pentium 133 or P6/200) while scrappy
was running on my P5/75 notebook. At this point scratchy's rating was 2800+
and the next closest human was 2500 or so.
> c) i have not seen it lose because of tactical mistakes yet, so
> tactical it's at ferret level.
> d) it makes the weird deep blue Bxg3 mistake, yet plays
> very strong further
> e) its rating is way higher than zarkovx (which has not only
> a high similar style as deepviolet, but from experiments
> also plays positional similar to deep blue).
> f) finger says deepviolet plays at a PII-266. operator told me
> PII-333, and updated recently his finger.
> ZarkovX runs unattended at a K6-3 at 475Mhz. zarkov runs way faster
> on a K6 than on a PII at same speeds. This K6-3 at 475 is for
> crafty though and for DIEP exactly as fast as a PII-450. For zarkov
> it's more like a PII-650Mhz.
> g) operator needs 18 seconds increment, a level which is only
> needed if you are way older than 75 or operating hardware or
> a parallel system. Operators of Rebel use 8 seconds a move (very seldom
> 9 seconds). operators of windows programs use 4 seconds a move
> (this all loss because of operating).
I don't follow the 'parallel hardware' comment. Cray Blitz used 16 cpus and
played dozens of 5 0 games with me manually operating. In fact, we played with
the _real_ chess clock, and relied on my typing speed to not lose on time.
There doesn't have to be a big delay to get going. DT played lots of speed
games at the various ACM events with no problems at all...
> h) he already plays for years 2 18. a very consequent man this operator.
>
>fingering scratchy:
>
>Information about scratchy(C) (Last disconnected Wed Mar 13 1996 19:25):
>
> rating [need] win loss draw total best
>Blitz 2871 [8] 141 1 4 146
>Standard 2781 [6] 374 14 37 425
>
> 1: Meow. I am a seafood gourmet.
>
>
>Information about DeepViolet(C) (Last disconnected Tue Jun 01 1999 22:09):
>
> rating [need] win loss draw total best
>Blitz 2857 203 105 78 386 2898 (20-Sep-1998)
>Standard 2508 [6] 94 45 21 160 2526 (02-Oct-1998)
>
> 1: DeepViolet is an experimental chess program developed by several graduate
> 2: students in the Virgin Islands and maintained by the DeepViolet Laboratories
> 3: in Northern NJ. We appreciate the stronger players taking the time
> 4: for matches with DV, and especially welcome matches from other
> 5: computers. This is version 4.0 and is manually operated, requiring
> 6: the large increment (2 18 r) for blitz games. Thank you for your
> 7: patience. DeepViolet now runs on a 333mhz Pentium II with 64 megs RAM.
>
> Address : octo@intercall.com
>
>Nowadays 2800 is not much that much anymore for programs at the internet:
>
>Statistics for crafty(C) On for:31:21 Idle: 1
>
> rating [need] win loss draw total best
>Wild 2113 [6] 185 50 8 243 2118 (17-Jan-1999)
>Bullet 2969 5703 1363 927 7993 3108 (04-Jul-1997)
>Blitz 2912 45140 12385 7824 65349 3187 (09-Apr-1999)
>Standard 2655 3333 1322 883 5538 2720 (16-May-1999)
>
> 1: Crafty v16.10 (4 cpus)
> 2: crafty now uses all 3/4/5 piece endgame databases, over 20gb.
> 3: Intel 486/66, 4mb, 40mb EIDE disk
> 4: Accumulated hours: 26,500 % of life on icc: 75.54
> 5: Crafty is freely available, both source and a windows executable, for those
>that don't have unix machines they can compile on.
> 6: anonymous ftp to ftp.cis.uab.edu/pub/hyatt
> 7: Note that I generally +noplay computers that don't fill in their notes with
>the program and machine being used...
> 8: note to computer operators: if you are only interested in rating points, as
>verified by only matching crafty when it is rated higher than you, but not when
>it is lower, then expect to end up on its +noplay list.
> 9: also note that _any_ computer operator that interrupts a GM/IM match will be
>+censored and it will be permanent.. If crafty is playing a human do _not_
>interrupt and jump in between games.
>10: I like playing "rabid".. it inflates my rating. :)
>
>finger ferret:
>
>Information about Ferret(C) (Last disconnected Fri Apr 30 1999 22:54):
>
> rating [need] win loss draw total best
>Wild 2618 [6] 2 0 0 2
>Bullet 2930 [8] 2027 498 372 2897 3155 (28-Jun-1998)
>Blitz 3132 [8] 11469 2416 2638 16523 3132 (30-Apr-1999)
>Standard 2720 [6] 663 218 237 1118 2825 (01-Apr-1998)
>
> 1: Ferret. Xeon 450 mhz x 4
> 2: Message me and I'll put you on my spam list, these people will get email
>when Ferret is released for public consumption (no other use will be made of
>this list).
> 3: I am automatic and mostly unattended. Autoflag is on. I will not resign,
>give takebacks, or accept or offer draws, unless the operator is present. I
>reserve the right to decline any challenge.
> 4: "finger schroer" for information on high quality lessons, etc.
> 5: Please send any comments via ICC "message" command to Mouse or Ferret.
> 6: If you get computer assistance during games, but are not on the computer
>list (no "C" after your name), please read "help computer" and stay away from me
>completely if you wish to continue to cheat.
> 7: I'll play unrated fast blitz with any registered player (5 2, 4 4, 3 5, 2 7,
>or faster).
> 8: I'll play rated 2 0 bullet, rated slower blitz (no 3 0 or 4 0), or quick
>rated standard (30 0, etc.), with any established human over 2400, or any other
>established player (including computers) within 300 points.
> 9: Please play an even number of games at a given time control, and no
>auto-rematch please.
>10: 1995 world microcomputer amateur chess champion, 1996-1997 world
>microcomputer speed chess champion
>
> Address : brucemo@seanet.com
>
>Search deepviolet:
>
>[Event "ICC 2 18"]
>[Site "Internet Chess Club"]
>[Date "1999.05.19"]
>[White "DiepX (4 processor version)"]
>[Black "DeepViolet"]
>[Opening "Ruy Lopez: modern Steinitz defense, 5.O-O"]
>[ECO "C72"]
>
>1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 d6 5. O-O Bd7 6. c3 Nge7 7. d4 Ng6 8.
>Re1 Be7 9. Nbd2 O-O 10. Nf1 Nh4 11. Nxh4 Bxh4 12. Ng3 exd4 13. cxd4 Re8 14.
>Bxc6 Bxc6 15. d5 Bb5 16. Bd2 Bxg3? 17. hxg3 c5 18. Bf4 a5 19. Qb3
>{Game courtesyadjourned by White}
>
>The 16..Bxg3 move was funny and quite laughable. DIEP doesn't even
>*consider* bxg3 of course. I tried several ZarkovX versions:
>don't even *consider* Bxg3 either.
>
>Now this bxg3 gets made by more programs, i don't know why actually, it's
>a beginnersmistake in chess. Giving up the bishop pair AND giving white a
>nice g3 pawn.
>
>Now let's get to a game fragment
>
> DeepViolet - other program
> 1. d4 (0:05) Nf6 (0:06)
> 2. c4 (0:05) g6 (0:04)
> 3. g3 (0:05) c6 (0:04)
> 4. Bg2 (0:07) d5 (0:04)
> 5. cxd5 (0:06) cxd5 (0:04)
> 6. Nc3 (0:07) Nc6 (0:04)
> 7. Nf3 (0:27) Bg7 (0:06)
>Weird, but not all programs play Nf3
>
> 8. O-O (0:19) Ne4 (0:10)
> 9. Bf4 (0:39) O-O (0:09)
>Most prefer Bg5
>
> 10. Rc1 (0:25) Be6 (0:06)
>
>
> 11. e3 (0:29) f6 (0:39)
>e3? the opponent of deepviolet would have made the same mistake
>
>
>
> 12. Qb3 (0:29) Qd7 (0:31)
> 13. h4 (0:07) h6 (0:12)
> 14. g4 (0:33) Rad8 (0:44)
> 15. Rb1 (0:38) g5 (0:28)
>
>Now this is a weird move 15. Rb1. it's laughably bad.
>it's not a tactical blunder though. I see junior at huge depth
>get to the same move. Zarkov is the only program that could reproduce
>this move, but deepviolet definitely isn't zarkov at a PII-333.
>It isn't zarkov at all.
>
Crafty may well do that as well. I have been looking at it to see why
this happens occasionally...
> 16. Bh2 (0:24) Rf7 (0:46)
> 17. Rfc1 (0:33) Bxg4 (0:25)
> 18. Nxe4 (0:37) dxe4 (0:10)
> 19. Nd2 (0:06) f5 (0:31)
> 20. hxg5 (0:06) hxg5 (0:08)
> 21. f3 (0:13) Bh5 (0:17)
> 22. fxe4 (0:11) Nxd4 (0:09)
> 23. exd4 (0:09) Qxd4+ (0:06)
> 24. Kh1 (0:07) Qxd2 (0:12)
> 25. exf5 (0:08) Bg4 (0:42)
> 26. Rc7 (0:25) Qe2 (0:20)
> 27. Rc2 (0:11) Qd3 (0:10)
> 28. Qxd3 (0:07) Rxd3 (0:07)
> 29. Rc8+ (0:12) Kh7 (0:10)
> 30. Rc7 (0:40) Bxf5 (0:18)
> 31. Rxb7 (0:07) Bd4 (0:49)
> 32. Bg1 (0:37) Kg6 (0:22)
> 33. Bxd4 (0:06) Rxd4 (0:15)
> 34. Rc1 (0:12) a6 (0:47)
> 35. Ra7 (0:43) Bd3 (0:06)
> 36. Kg1 (0:11) Bb5 (0:32)
> etcetera of course white lost it
>
>So i give here 4 moves:
>
>16..Bxg3?
>7.Nf3
>9.Bg5
>15.Rb1
>
>I haven't found a program yet that makes all those moves in blitz,
>or a minute later.
>
>Can someone try all his programs at home at these 4 positions and see
>whether these 4 moves get the mainline.
>
>I like to solve this deepviolet mystery. It is not
> zarkov 4.5 4.0
> diep
> fritz
> junior
> wchess
> genius 6.0
>
>So what softwareprogram/hardwarecomputer is it then?
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.