| Computer Chess Reports Vol. 4 No. 2 $10.00 ‘he Dedicated Chess Computer Comeback With software taking front stage, some new offerings, albeit more ex- pensive, from stand- alone manufacturers are making quite asplash.... (Pictured Above) The New TASC R-30 Version 2.2THIS ISSUE Winter '94 Review.. 1993 World Microcomputer Championship Aegon 1994.. The 4th Harvard Cup. Oviedo Tournament Welser Excalibur Review... Mephisto Review. Novag Review Saitek Review Tasc Review. Personal View SSDF Ratings List. Tasc Evaluation PC SOfWAFE essere Opening Power For Your Genius 2.0 Teaching PC Software .. PC Program Ratings .. PC Speed Adjustment Chart... Rating the Commercial Computers ... CCR Ratings List .. The Louguet Set Revisited The One-Hour CCR Test.. The Relative Value of the Pleces.. Bits & Pieces. Definitions The End Is Here. SSRSRRR The Computer Chess Reports is brought to you by: Senior Editor: Larry Kaufman, Stoll Editors: Ricardo Alcos, Froncesca *Chess" Levy, Steve Schwartz, Bob Sostock Design Tear Leader: Poul DeStefano Design Team Apprentice: Mark Schneider Product Coordinators: Phiip Kiet, loseph Aiello Electronic Transmissions Consultent: Bob Marzano Subscriptions: Geri Emon, Printing: George Dunlop Printing, Peter Carlucci consultant Publisher: Computer Chess Digest Incorporated All Materiol ©1994 Computer Chess Digest We welcome letters, comments, criicisms, ideas, aries, cheesecake recipes, tournament results or anything you would care to send us. Send oll material to: CCR Bits & Pieces c/o ICD, 21 Walt Whitman Rood, Huntinglon Staton, NY 11746 USA. No text or graphics within this publication may be reproduced by any means without the writen consent of the editors, ‘excep! for excerpis used in reviews. Computer Chess Reports i in no way connected to the United Stotes Chess Federation, Chess life, Chess Digest Inc. or ony of the companies producing product reviewed within Winter ‘94 Review by Larry Kaufman, International Master The big news since our last review is the takeover by Saitek of Hegener & Glaser, the manufacturers of Mephis- to and Fidelity models. This creates a situation in which ‘one large company dominates the dedicated chess com- puter market, with smaller companies (Novag, Excalibur, TASC) competing in certain segments of the market. Since dedicated chess computers can no tonger claim to be superior in playing strength to top PC software on fast PCs, sales are now limited primarily to those players who simply prefer to play on a real chessboard instead of on a screen and 10 those who have no real use for PCs outside of chess. The takeover seems to have halted the flood of new ‘models by all ofthe companies involved, but atleast sore ofthe models mentioned as planned in the last issue have made their appearance in the U.S. Models of particular interest at this time include the TASC R30, the Mephisto Berlin 68020 Professional, Saltek GK2000, Travel Champ, the Novag Emerald and Ruby, and the Excalibur Krypton ‘and Comet. {As for tournaments, the top computer at Harvard Cup was Socrates, the top computer at Welser, Austria was Mephisto Genius 2, and the World Microcomputer Cham- plonship also went to Mephisto Genius 2. Hiarcs 2.1 won the World Microcomputer software division. The ACM International Computer Chess Championship was postponed until June at last word. While new versions of some of the leading pe programs have come to market, no totaly new programs have madea splash since our last review, Genius 2s stil ‘generally rated as the top pe program, and ChessMaster 4000 seems to be the top program running under Win- dows. Al of the new versions of the leading programs seem to be somewhat stronger than their predecessors, but Genius 2 and Hiarcs 2.1 seem to show the best improvement. Because pe software has come to dominate the com- puter chess market, the emphasis in CCR needs to shift more in that direction. Since | am now the co-author of several pc programs (Rex, Socrates, Kasparov's Gambit) ‘and continue to develop new ones, itis not really ap- propriate for meto continue to review pe software, nor do have time to do justice to the many pe programs, and ‘80 | plan to relinquish the pc software portion of CCR to another writer, USCF Master Nick Schoonmaker. Nick is rated over 2300, plays regularly in tournaments in the Orlando, Florida area and has considerable experience playing against chess programs. | think he will be able to, do more in- depth reviews of individual programs than | have been able to do when trying to cover everything on the market each issue. | plan to continue to write for CCR, about dedicated machines (since this does not pose a conflict for me) and other special articles. As for this Issue, | am stil the author of all articles not otherwise attributed. ‘As usual, our apologies for the lateness of this issue, but interesting late news always tempts me to hold up Publication. in order to both solve the lateness issue and the readers’ desire to have the most up-to-date informa tion, weare going to.a quarterly or bi-monthly format, We believe that such a new format will be beneficial for everyone, 1993 World Microcomputer Championship ‘The World Microcomputer Championship was con- tested in two divisions last fallin Munich, Germany. Only three dedicated machines played in the Manufacturers Group, each playing eight games with each of the other two. In contrast, the software division was well attended, with some 28 programs contesting a 9 round Swiss. The Manufacturers Group was virtually just a match between Mephisto Genius 2 running on a very fast (est. 80 MHz) 486 and the TASC "R30" (actually an R40 as it ran at 40 MHz), since the third entrant, Saitek RISC 2500, ran on much slower hardware than the other two. Mephisto Genius just nosed out the TASC RaQ by 1/2 point (MG scored 10 1/2, TASC 10) while the RISC 2500 was left in the dust at 31/2. Since the TASC and Saitek entrants both Used similar programs by Johann de Koning, the huge disparity between their results shows the value of fast hardware. Although Genius outscored TASC, the clear Implication for the commercial market is that the commer- cial TASC R30 is stronger than the commercial Mephisto Genius 68030, since in this event the R90 was only sped Up bya.4103 ratio, while the Mephisto machine ran about 7 to 3 faster than the 68030 model. The Software section was won by Hiarcs (later released as Hiates 2.1) with 7 1/2 points, running on a 36 MHz SPARC chip, followed by The King at 7, cunning on a 150 MHz (!) DEC Alpha chip. Third at 6 were Genius 2 ‘on a Pentium 69 MHz chip and the amateur Austrian program Nimzo on a 486/80. It is thought that the SPARC chip used by Hiarcs was roughly equivalent in speed to the Pentiums in the event but far below the speed of the DEC Alpha chip used by The King. Next at 5 1/2 were Gideon on a Pentium 60, Pandix, Quest (=Fritz), andKallisto all on 486/66, and MChess Pro on a Pentium 60. ‘At 5 were Bobby2 and Mirage on 486/66 and Ulysses 1 ona SPARC 41 MHz. At 4 1/2 were Saitek Sparc on a 50 MHz SPARC, and Ecume, Now, Greif, and Nightmare, all (0n 486/66. At4 were the Russian program Centaur, SOS, and Magellan, all on 486/66. At3 1/2 came Gandalf (486), Breakthrough (Sparc), Experimental (Pentium), and Diogenes (486). At’ were Viktor 2 (486) and Chess Brain 1.0 {Pentium 66). The tallenders were BB at 2 and Pachach at 1/2, both on 486/68. Clearly Hiarcs had the ‘most to brag about in this section, The event concluded with a playoff between Mephis- to Genius and Hiares (both by British programmers), with Mephisto winning the match 1 1/2 to 1/2 to take the ‘Absolute Championship. It seems to me that Lang (Mephisto), de Koning (TASC), and Mark Uniacke (Hiarcs) all have good reasons to be proud oftheir results inthis event. It also seems to me that there is not much Point to making the distinction between dedicated machines and software anymore; the line is just too fuzzy. The same program can easily run inside a box or ona pe, as both Mephisto and TASC did in this event, so why bother with two divisions? Aegon IS94 ‘The Sth annual Aegon tournamentin the Netherlands. was won by Grandmasters Larry Christiansen and John ‘Nunn with § 1/2 outof 6 points, followed by Grandmaster David Bronstein and PeeWee van Voorthuijsen at §. The top computer program at 4 1/2 was Gideon 2, the As- ‘sembly language upgrade of Gideon Pro. As the results Just arrived on deadline day, | don't have many details yet; | presume that most if not all of the pe programs ran fon Pentium machines. The tournament matches 38 computer programs against 38 human players in a six Found Swiss-system match; five of the humans are grandmasters, the others are mostly of U.S. master strength or stronger. The total score this year was exact- ly tied; 114-114. Considering that the computers won last year (98 1/2 - 93 1/2), this has to be abit of a surprise in view of the improved hardware each year. Perhaps the average rating of the humans was higher this time (the results don't list the ratings of the players) . although thelist of players does not look so different fromlast year. Pethaps the human players are learning how to combat, computer strengths. Games were at 40/2. ‘The programs scoring 4 points were TASC R30, Hiares, The King, Mephisto Genius, Hitech, Fritz 2, Joker, Quest, and Zarkov 3.0. At3 1/2 were TASC R40, Chess Genius 2, Mephisto Berlin, Check Check, and Nightmare. 4 ‘At3 were MChess Pro, Mephisto RISC WK, Chessmaster 4000, Complete Chess System x, Ecume, Kallisto, L- Chess, and. At 2 1/2 came Gandalf, Goldbar, Greif, Now, Pandix, Saitek Brute Force, Saitek Sparo, Schach 3.0, and Socrates exp. Then, at2 points we find Anance, Isichess, Mirage, Nimzo-Guernica, and Saitek RISC 2500, The tall enders were Milobarus at 1 1/2 and Dappet at 1. ‘The TASC R30 performed especially well. Its four points include wins over two grandmasters, David Bronstein and Roberto Cifuentes Parada, The Bronstein ‘game appears below. | regret that other details are not available in time for this issue, but we will cover them in the next issue, White: TASC R90 Black: Grandmaster David Bronstein, former World ‘Champion finalist (He tied a ttle match with Botvinnik back in 1951.) 1.04 06 2.4415 3Nc3 NIG 4.03 Bb4 5 NIS 0-0 6 Bd2 b6 7 Bd3 Bb7 80-085 9.43 Bxc3 10 Bxc3 Ned 11 Act d6 12 Qe2Nd7 13 b3 Qe7 14Bb2c5 15 Ret Racs 16 Qc2 Ndi6 17 RA QI7 18 Nd2 Qhs 19 f3 Ng5 20 bél? axb4 21 axb4 oxb4 22 Qa4 e5 23 Oxb4 e4 24 Be? ext3 25 NxfS Ng4 26 Qxd6 {4 27 he (I's hard to believe that white survived this attack and won!) Ne4 28 Gxbé Ng3 29 Rice! Nxes (Black could probably have won simply by Not sacrificing the bishop) 30 Qxb7 R8 31 Qa7 Axb2 32 Ritz Kh8? (why not 92...Ng4) 33 Bd3 Abbé 34 Ra2 Ride ‘95 Qc7 Qg4 36 Qb7 Rb 37 Qa7 Rbe8 38 Qc5 Ros 39 Qg5 Qd7 40 Rxe3 tke3 41 Oxg3 RieB 42 Qg5 Ros 43 NeS Qxd4 44 Nxc6 Qxd3. 45 Ra7 Qd1+ 46 Kh2 Qd6+ 47 Ne5 96 48Qh6 Gxe5 + 49g30b2+ 50Kh3 and black resigned. The 4th Harvard Cup The 1993 Harvard Cup was played on November 6 at the Computer Museum in Boston. The format is that a team of top U.S. grandmasters (four in the first two cups, then five, and this time six) plays against a team of top ‘computer programs (all the programs after the first cup have been commercial or prototypes of commercial programs, running on microprocessors) at game/25'. Each grandmaster plays one game with each program. The grandmasters are competing for cash prizes ($1,000 first, $500 second this time), the computers for favorable publicity. This year the chief sponsor was Intel, which provided Pentium based 60 MHz PCs to the four programs that ran on PCs. For the third straight year, a program entered by Heuristic Software and authored by Don Dailey and myself was the winner of the top computer prize. Socrates Experimental scored 3 out of 6, bringing our total score against the grandmasters over the last three Cups to 8 out of 15, a plus score! Moreover, we were quite unlucky, since against former World Junior Cham- pion llya Gurevich we were about to queen a pawn when ur flag fell after 100 moves (we stil got a draw as he had ‘only bare king); we were two pawns up against former Soviet Champion Boris Gulko but allowed him to reach a drawn rook ending; and we had an easy win against the reigning U.S. Champion Patrick Wott butthe program got too greedy, snatching a pawn instead of simplitying to an ‘exchange up ending. Socrates did deteat last year's Harvard Cup winner Michael Rohde as well as Grand Prix ‘Champion Alexander Ivanov, and lost to Joel Benjamin, who won the Cup this year with a perfect 6-0 score. Second place among the computers went to the TASC R30, which scored an impressive 2 1/2 out of stx. ‘Aithough the processor in the R30!s very fast, roughly like ‘486/68, itis notin the class with the Pentium, so the R30 was in that sense handicapped relative to all the other programs but the SPARC. Third place was attie between MChess Pro v. 3.42 and —Battlechess 4000!!! Bat- techess has never been known for strength, only for its (graphics and animation, butit seems thatit was complete- Iy revised to be a strong program as well. Since it has never competed or been tested independently everyone expected it to get shut out, but It was not so Weak and. perhaps the GMs underestimated it. Fit place went to Saitek Sparc, which managed only a single draw, and Kasparov's Gambit was shut out with 0 points, though it ‘would probably have beaten Joel Benjamin if the com- puters were given a couple seconds per move operator time (its interface Is slower to operate than the programs written for tournament players, 80 itost on time) Overall the computers scored 9/36, or 25%, the same percentage as in 1991 and slightly below the 28% of 1992 (they only scored 9% the first year, even though Deep ‘Thought and HiTech were playing). In view of the use of the Pentium machines this time, this was a real disap- ointment to computer fans. Part of the problem was that ‘some of the best programs were absent (Chess Genlus2 ‘most notably), but some observers felt that what has happened is that the grandmasters are learning how to play effectively against computers. Most oral ofthe GMs have by now had substantial experience playing against computers, both in the Harvard Cups and elsewhere, and, they no longer take their opponents so lightly. Assuming that the top programs all participate nexttime, and assum- ing that the Pentiums will then run at 90 MHz or more, perhaps the next match wil be a much closer one. Since Intel would like to see the computers win, Its likely that they will do what is necessary to insure that the strongest programs do participate and that the fastest possible hardware is available. AAs for the human participants, lvanov took second at 5, Gulko, Wolff, and Gurevich all made 4 1/2, and Rohde, said to be out of practice, scored only 2 1/2. LL. GUREVICH (ELO 2575) TASC R30 1, 04 €5 2. NI3 Nc6 3, BbS a6 4, Bad Ni6 5. 0-0 Be7 6. Ret bS 7. Bb3 0-0 8 d4 Nxd4 9. Nxd4 exd4 10. 5 Ne8 11. 68 dxc3 12. Nxo3 Bb7 13. Nd5 d6 14. e615 15. Bla NIG 16. Nxe7 Qxe7 17. Ret Raed 18. Qd4 Ned 19. Qa7 05 20. xed fxed 21. Bxd6 Qxd6 22. 67 + 4.23. exi8Q+ Pxf8.24. d1 Qi6 25.Bc2 Qxb2 26. Bbt c3 (see below) 27. Qc5 Qe2 28. RY Qd2 29. Qe5 c2 30. Qe6+ Kh8 31. Bxc2 Qxc? 32. Qe7 Qc8 33. h3 BdS 34. a3 O15 35. Qa7 Bod 36. Rdt Bd3 37. Ret QI6 38. Qe3 a5 39. Rc7 b4 40. axb axb4 41. Rb7 Qc3 42, Qb6 Qat + 43. Kh2 Qe5+ 44. Kgt Qc3 45. Kha. 6 46, h4 Qe5 + 47. Kg! Re8 48. g3 Qat + 49.Kh2 Act 50, Kha Rht 51. Kg Qes 0-1 TASC R30 PATRICK WOLFF (2585) 1. e4 65 2. ¢3 NIG 3. 05 Nd5 4. d4 cxd4 5. cxd4 d6 6. Nig No6 7. Nc3 dxe5 8. dxe5 Nxc3 9. Qxd8+ Nxd8 10. bxc3 Bd7 11. Bd3 @6 12. 0-0 Rc8 13, Bd2 hé 14. Rfet BcS 16. Nd4 0-0 16, Nb3 Be7 17. Bad Nc6 18. f4 Rfd8 19, Be3 Bo8 20. BI2 Ki8 21. Re3 bé 22. Rh3 Nbé 23. Bd4 Bad 24. 15 oxf5 25. Bxi5 Bd7 26. RIG Kg8 27. Bxd7 Axd7 28. Ret Nc6 (see bottom of last page) 29. Fxt7 Nxd4 30. Nxd4 Kxd7 31. 06+ Ke8 32. exd7 + Kxd7 33. Rd1 BI6 34. NbS + Ke6 35. Nxa7 Rxc3 36. NbS Ro6 37. Ret + Kd5 38. Rd1 + Kod 99. Ret + Kd5 40, Rd1 + Kod 41. a4 Ke3 42. Ret + kd3 43. kf2 Ke2 44. Ki3 Kb3 45. Red Rod 46. Rxc4 Kxod 47. Nc7 Kb4 48. Nd5+ KaS 49. Ked Bd8 50. No3 Kod 51 Kd kbs 52. g3 Be7 53. Nd5 Bc5 54. No3 Bb4 55. Nd5 Bab ‘56. Ne7hS 57. Nd Kxad 58. Nid bS 59. NxhS Ka 60. Nxg7_ 4 61. NIS b3 62. Ne3 Bc7 63. Nd5 b2 64. Kc2 Ka2 65. Ne3 + Kat 1/2-1/2 White: Grandmaster Michael Rohde Black; Socrates Experimental on Pentium 60 MHz 11NIS NIG 2.04 b6 3 Nc3 Bb7 4.d4 dS (| think this line does not give black full equality) 5 cxd5 Nxd5 6 Qc2 e6 7 @4Nxc3 8 bxc3 Nd7 9 Bd3 Be7 100-00-0 11 Bid cS | 1205 exdS 13 exd5 Bxd5 14 Bxh7+ Kh8 15 BIS Bxi3 16 5(p49 Bg5 17 Bd6 Be7 18 Bg3 NIG 19 Afe1?! (I think that white should have played 19 Radi first, since this move ‘order allows black to avald getting pinned on the e fl. Despite white's awful pawn structure, his dominant piece placement and bishop pair in open position would have given him the edge.) Rg8 20 Radi Qf6 21 Qe2 eB 22 Bo2?! (22 Qa6!?) Bds 23 Qt Rxet 24 Qxet Bxg3 25 xg?! (hoping for a chance to check on the h-fle, but 25, ‘hig would heve kept the game fairly balanced) QaB 26 Qe7 Oxi3 27 Ret? Qxc3 28 Bb1 Rab 29Rd1 Reb RSE Oct + 31 Kg2Rxd8 32 Qxd8 + Ngé 33 Bed Qh6 34 BdS ‘Qhs 35 Qd7 Ni6 36 Qc8-+ Kh7 37 BIS Qe5 38.Qb7 Qe6 39 Qxa7 c4 40.a4¢3 41 Qc7 Qb3 42 Qc8? (42 Qf! would prolong the game) c2 43 Qh3 + Kg6 44 Qc8 Qb2 45 Be2 eQ) 46 Bd3+ KgS 47 QI + Khé 48 Qho-+ Nhs 49 f5 Qf6 and white resigned. This game iiustrates how dificult is to outplay a strong computer in an open middlegame position. OVIEDO TOURNAMENT December 1993, Oviedo Spain, in one of the largest chess tournaments in the world, the Chess Machine (R- 30) with the King 2.0 program (an earlier, weaker version of the new 2.2), Genius 2.0, and $57 human participants fought for victory. More than 0 grandmasters competed in the tournament. The Chess Machine (R-30) gained an ‘overwhelming 1Sth rank behind 13 Grandmasters and 1 International Master. The King 2.0 program beat 4 grandmasters- GM Salov, WGM Galliamova, GM Strikovick, and GM Makarickev. King also drew against two Grandmasters and two International Masters- GM Conquest, GM Svesnikov, IM Gomez, and IM Wells. Here are a few of the games from the Oviedo Tour- nament: THE KING ‘GM VALERY SALOV (ELO 2685) 1.04 NI6 2. d4 e6 3. NIB Bb4+ 4. Bd2 Qe? 5. 93 NOS 6. Bg2 Bxd2 + 7. Noxd2 d6 8. 0-0 €5 9, d5 Nb8 10. Qa + Nbd7 11. Qc2 a5 12. Ng5 Nc5 13. Nde4 Nixed 14. Nxed 6 15. Nxc5 bxo5 16. Bes g6 17. Bg2 0-0 18. e4 Rb8 19. 3 Bd7 20. Qo3 a4 21. bxad Rb4 22, a5 RaB 23. 14 RxaS 24. a Rbad 25. raz RaB 26. Qd3 R4a6 27. Qed 16 26. 15 195 29. Act RD6 30. Qd3 Qd8 31. Abt QbE.S2. Rubs Oxdé 33. hd gxh4 34. gxh4 Kh8 35. R2 Rg896. Kh2 Ba4 97. Qd2 ‘Qbs 38. Qhé Qd8 39. Rb2 Rg7 40. Ges Rod 41. Kho Ris 42, BIG Qa8 43. Qd3 Kg7 44. Rg2+ Kh8 45. Rg3 Qbs 46. (Qe2 QI6 47. h5 06 48. Qd3 oxds 49. Qxd5 Bc2 50. Ka? h4 51. Kgt QbB 1-0 THE KING GM SERGEY MAKARICHEV (ELO 2540) 4, dé @6 2. e4 d5 3. Nod NI6 4. 65 NId7 5. 14 05 6. NB No6 7. Bed cxd4 8, Nxd4 Bc5 9. Od2 Nxd4 10. Bxd4 Bxd4 11. Qxd4 Qb6 12. Qxb6 Nxbé 19. NDS Ke7 14. Kd2 Bd7 15. Nd4 Rac8 16. b3 15 (see digram #1) 17. exi6ep+ gxi6 18. Ret Kd6 19. Bd3 @6 20. NIS+ BxfS 24, Bxf6 Re7 22, 03 NCB 23. fxe5 + fxe5 24. Rh RI7 25, Bg Rg7 26. Bh3 Re8 27. 93 Kc7 28. AIS Age? 29. RhS NOG (600 diagram #2) 30. Bg2 Kc6 31. 04 KoS 92. Bxd5 bs 33, Ko3 a5 94. a4 ba + 35. Kd3 NcB 36, Ko2 Nd6 97. Bg2 Kb6 98. Rd1 Re6 39, Ad5 Ko7 40. Axh7 + 1-0 IM PETER WELLS (Elo 2455) THE KING 1. NG d5 2. d4 NIG 3. c4 06 4. BGS h6 5. Bxi6 Qxf6 6. {Qb3 06 7. No3 aS 8. 63 Nd7 9, Bd3 dxc4 10. Gxos Bas 11. 0-0 0-0 12. Nes Qe7 13. Nxd6 Qxd6 14, Qc2 b6 18. Ract BD? 16, Aidt Rac8 17, a4 QbB 19. BbS cxd4 20. Qxd4 Nos 21, Qd6 Bd 22. Qxb8 Rxb8 23. Nd4 RidB 24. 13 2625: Nos Bxc6 26. Bxo6 Nb3 27. bt Nd? 28. Rbct Nb3 29. Abt Nd2.90. Rlsct Nb3 1/2-1/2 Welser ‘The 1994 Welser tournament in Austria was a round robin played at 60 moves in 2 1/2 hours among 12 strong programs. Seven ofthe programs ran on 486 computers, two on RISC computers, one on a 68090, and two on Pentium 60 MHz machines. Oddly enough, the two Pen- tium entrants finished in the last two places. Since they ‘were the only amateur entrants, perhaps they ran on Pentiums in an attempt to equalize the contest a bit. Richard Lang's "Mephisto Genius 2" on 486/66 won the event, scoring 8.5 out of 11. Hiarcs 2.1 on same was second at 7.5, and the TASC ASO took third at 7. Next came MChess Pro 3.5 on 486/66 at 6.5, then Mephisto Vancouver 68030 at 6, and then tied at 5.5 were Chess- Master 4000 (on 486/33) and Gideon Pro (on 486/68). ‘Socrates 3 on 486/50 was next at 5, followed by Saitek ‘SPARC and Fritz 2 (on 486/33) at 4.5. The talenders on Pentiums were Nimzo X with 3 and Milobarus X with 2.5. ‘The results agreed with other recent events which suggest that given equal hardware, Genius 2 is the top program, with Hiares 2.1, TASC R90, and MCHess Pro 3.5 battling it out for the recognition as the second best Program. White: MChess Pro 3.5 on 486/66 Black: Hiares 2.1 on 486/66 1 eb eS 2Bo4 Ni6 3.04 exd4 4NiS dS. (Both this, move and acceptance of the pawn sav by 4..Nixe ara considered good enough for equality according to ECO) 5 exd5 Bb4+ 6 Bd2 (The book lings 6 c3 Qe7 + 7Kgt dxc3 BNXC3 0-0 = ) Bxd2+ 7Nbxd20-0 80-0Nxd5 SNxd4 NDE 10 €3 65 11 Nd4-b3 Nxo# 12 Nxc4 Bob 13 Ne3 Na6?! 14 Rel Rb8 15 Qh5 157! 16 Radi Qee 17 Qg5 GI7 18 Rd2 RbeB 19Redi hé 20.014 Bxbs 21 axbs Qxb3 22 Nx15 Qe6 23 94 RI7 24 Qad Refs 25 Rd7 Kh7 26 R7d6 Qe2 27 Rid2 QeS 28 Qc4 Qet+ 29 Kg2 Ges 30 Ro Qa8 31 Kg1 Nc7 32 Re3 b6 33 Ne7 96 34 Rd7 NeB 35 Qd3 N‘6 36 Qxg6-+ Kh8 37 Qxhé-+ Nh7 3614 4 3915 a6 4016 Rg8 41 No6 + Rxg6 42 Qxg6 Rxd7 43 Re 8+ and black soon resigned. Excalibur Review Excalibur is the spiritual descendant of Fidelity, as its president, Shane Samule, is the son of Fidelty’s founder and many of the employees at Excalibur have been former Fidelity employees. Moreover, Excalibur is based In Miami, as was Fidelity. However, Excalibur, unlike Fidelity, has sa far only marketed inexpensive chess computers. The only models ofinterest to most CCR readers are the Expert level machines utiizing the h-8 processor and programmed by David Levy and associates. These include the hand-held sensory Accolade and peg-sensory Comet (above), the table-top sensory Legend II (left) and Krypton Challenge. The Comet EBand Krypton Challenge have ewer and presumably stronger programs than the other two, though none of these models has been rated by any rating list or agency ex- cept for the Accolade, the oldest model of the four. itwas ‘ated 2065 in Action Chess (game/30) by the C.R.A., and received an astonishingly high blitz (gamel8) rating of 2344 from the W.B.C.A. Allowing for the overrating of computers at fast levels, the C.R.A. result suggests a tournament rating of around 2000 for the Accolade while the W.B.C.A. rating suggests a tournament rating of over £2100, 80 perhaps a rating in the low 2000s Is nearest the truth at 40/2. | don't have much data to say how much improved the other models are, but probably the new ones are in the mid-Expert range. The Krypton Challenge ‘seems to be more reliable than Legend Il. As between the Accolade and the Comet, the Comet has a cover while the ‘Accolade does not; on the other hand, the Accolade works Con battery or AC while the Comet requires batteries. Mephisto Review The big news here is the takeover of Hegener & Glaser, the manufacturers of Mephisto and Fidality chess computers, by its archrival Saitek. So far the two com- panies have maintained their separate identities, and not much is being said about future plans, 50 we will have to ignore this major event unt the next issue and review the two companies independently. Clearly this takeover will reduce the element of competition in computer chess, but at least Saitek will stil have compettion, atthe high end from TASC, in the mid-range from Novag, and in the lower price range from Excalibur. The root cause of the disap- pearance of Hegener & Glaser as an independent com- anys probably the trend towards pc software and away from dedicated chess computers, which in tunis primari- ly due to the phenomenal increase in power in personal ‘computers over the last few years. While some newdedicated chess computers (Le. Mephisto Nigel Short) stil use the same old 6502 processor as was used in ‘Novag and Fidelity chess computers circa 1982 (and at only around doutle the MHz speed), pc processors have accelerated by about a hundred to one over that period! Now fast RISC processors have kept the dedicated com- puters in the ballgame, but they no longer can be sold on the basis of superior strength over pc programs, so @ ‘good chunk of their market is gone. This appears to be what crippled Mephisto. ‘Anyway, some of the prospective models ‘written about in the last Issue have made thelr appearance, though their continued production appears to be In doubt at this writing. The Mephisto Genius 68030 in the Exclusive board (left) has arrived in the US., and sells for about '6200-$900 less than the TASC R90. Its indeed a very powerful chess computer with excellent features, but itis not competitive with the TASC R30 due to the 2-1 superiority of Its RISC processor over the 68030 used by Mephisto. For anyone ‘who wants a Senior Master level machine in a quality ‘wood autosensory board, the Genius 68090 and TASC R90 both fil the bil, with the tradeoff of perhaps 35 rating points and piece recognition in the R90 for the $300 extra. Perhaps more interesting tothe average readers the Mephisto Berlin 68020 (right), which appears to have no rival in tts price range (under $800). Mephisto apparently decided to put the Genius program (or a very similar version) into the Berlin 68020, so its chess is essentially like the 68030 model but runs about 2/3 the speed (the (68030 at 33 MHz is about equal to a 68020 at 38 MHz, but the Berlin 68020 runs at between 24 and 25 MHz). The difference in strength between the models should be about 35 rating points. This should put it in the mid- 2400s, a huge gain over the Berlin 68000 and ahead ofall rivals except is “big brother" 68030 and the TASC R30. If 1s available, it deserves a "best buy" tag in terms of maximuin strength per dollar, but it s a plastic pressure board, not a wood autosensory like the 030 and the R90. ‘The revised "Louguet" test puts the new Berlin at USCF 2432 and the 68030 at USCF 2471, both of which are probably conservative figures based on extrapolation from the ratings of the pe version of Genlus2. ‘Most of the other Mephisto models are elther out of production or are not available in the U.S. at prices which ‘would be competitive with other companies’ models. ‘The Mephisto Nigel Short is essentially an upgraded program for the Mephisto Milano (righ). On my newest problem test it rated 2240, and in early testing by CONS it's at USCF 2253, but | doubt that the ratings wil remain that high, since "Nigel Short" Is only a software upgrade from the Expert rated Milano. If it holds the ‘master rating and can be purchased for under $300, it might be a decent buy. Novag Review ‘Although nonew Novag products rave reached market since ourlastissue, some informationis now avaiable about fone planned new model. The "Sapphire" wil reportedly be ‘an upgrade of the hand-held Ruby (righ). Its sald to have twice the program ROM, a much larger’ ‘opening book (64,000 plies), and enough RAM (128k) forhashtablos. ‘While no data Is available yet on playing strength, these specs should produce at least a hundred point improve- ‘ment, perhaps enough to putthe Sapphire into the Master lass, even allowing for the fact that the Ruby gota much lower "Piy" rating than expected from early results. It Is expected to sell for just around $200. If allthis is correct, it ‘may be the first hand-held Master level program, and the only Master level machine currently available for around ‘$200. Like all other Novag models, the Sapphire will probab- ly be even stronger (relative to other companies’ models) at {aster time limits like Action chess than at standard touma- ‘ment level (40/2). Since hand-held models are normally Used for faster games, this is a plus for Sapphire. Delivery dates & prices are unavalable at the time of this writing. Novag also plans a new table model called "Diamond, but ack information on it at this time. Despite the drop in rating from our last issue, the Emerald remains a good buy for those interested in Expert level table top models around $150. Although the Saltek GK 2000 Is now rated a bit higher and is also around $150, the Emerald is recommended for those players who demand a reasonably large opening book. ‘As for whether one should spend the extra $40 for the hand-held Ruby over the competing Saitek Travel Cham- pion, again the need for a large opening book would be the deciding factor. Saitek Review The main news here is the takeover of Mephisto by Saitek, discussed elsewhere. No new Saitek models have come out since our last Issue, although Saltek has ‘announced plans for several new ones in 1995, The Sparc module in the Renaissance board, al- though of (iow) Senior Master strength, Is more than a hundred points below the TASC R30 in strength, and is not even in production now, while the RISC 2500 has been surpassed by the Mephisto Berfin 68020 and has been plagued by production proslems and retums. So at the moment, Saitek is relying on its takeover of Mephisto for mastor level machines. The less expensive models have already been fully reviewed in past CCR issues. The GK 2000 (above) has been marked down to around $160 and the Travel Cham- pion (left) to under $100, at which prices they are both excellent buys. At these ices, they are probably the best values in their respective categories, namely Expert level table top and hand-held machines. ‘The current COR rating list puts the competing Novag machines (Emerald and Ruby) in betwee.) the Travel Champ and the GK 2UW0. The GK 2000 thus offers a slightly higher rating and more features than the Emerald {for around $10 more, while the Travel Champ is nearly as strong as the Ruby and quite a bit less expensive. These Saitek models are extremely strong intactical play, surely ‘of master strength in that arene, but the opening book is small and the endgame knowledge Is somewhat limited. TASC Revi iew Since our last review, the only new development is that the TASC ChessSystem R20 is now available in the US. with the new version 2.2 of the Koning program, which appears to be somewhat stronger than the original 2.0 version, at least according to problom tests, | have been ableto get a unitfor evaluation, and | fecl confident In calling it the best dedicated chess computer on the market today, considering both strength and all other aspects. Itis.a bt unusual inthat the processor, memory, and all the keys and information displays are housed ina ‘separate module, so that the chess board itself is just that ~ an elegant wood board (about midway in size between the Mephisto Munich and Exclusive boards) with no dis tractions other than the lights on all 81 intersections used to communicate the computer's move. One advantage of this separation of module from board is that i allows for 2 huge display window, large enough to allow all relevant data to be shown at once, large enough for a display of the entire chess position, and large enough to display simulated “real" (meaning analog, not digital) chess clocks. The R90 Incorporates automatic piece recognition, ‘making it possible to set-up problems in a fraction of the time of any machine without this feature. Just select position’ from the menu, set up the problem on the board, land press ‘enter’. {f all computers had this feature, one could use much larger problem sets to test them in a reasonable time. ‘As for playing strength, the R90 was listed at 2505 in the last CCR, but those results were nearly all from the older 2.0 program. The current figure comparably calcu- lated would be 2513, stil including the 2.0 results. Ifthe results for 2.2 were segregated, they would probably be inthe 2520 -2530 range. Results from problem tests show an even greater increase from 2.0 to 2.2 than the 20 point ‘gain indicated here. These numbers look about right, because the Saltek RISC "2500" uses an older (pre 2.0) Koning program at 14 MHz with less RAM, and was rated 2422 in the last CCR. The increase to 30 MHz (the R30 speed) should add 65 points, the larger RAM (512k for hash tables + 128k vs. 128k total) another 20, and the program improvements perhaps another 25 for atotal gain of 110. Considering all ofthis, a rating of 2530 USCF looks to be realistic. The R30 performed very well in both the \iorid Microcomputer Championships and in the Harvard (Cup, and just defeated the legendary grandmaster David Bronstein at the Aegon tourney within this past week. How does the R30 compare in strength to its only rival in the dedicated market, the Mephisto Genius 68030? ‘Although both Lang and Koning have superb programs, most observers, including myself, stil consider Lang's programs to be the best, even if only by a small margin, Moreover, the Genius runs at 33 MHz vs. "only" 30 for the 90. Despite these points, I consider the R30 to beclearly stronger than the Genius 68030, for the simple reason that the "ARM2" RISC processor in the R90 is roughly twice as fast as a 68030 at the same MHz. Even allowing for the Genius program having some edge over the Koning pro- ‘gram, the 68090 would probably have to run about 50% ‘aster than It does for Mephisto Genius to equal the R30 instrength. Asitis, the R90 should enjoy about a 35 point toad over the Mephisto model. Since the R30 also offers a slighty larger board (than the Exclusive board inwhich the Genius is now sold), automatic piece recognition, and a much larger display, and costs only alittle more, it seems