WEAK ENGINES????

[Fernando]

Each time we get or play or test an old engine, we tend to presume that it is “strong” or weak” according how it was evaluated in the time it appeared in the market. That is still truth of dedicated units by the simple reason they still operates on the ground of the same hardware, but any of us that has been playing old DOS engines face a very different situation: that “weak” program that was such due to a weak engine running in a 8088 processor, perhaps a 286 or at most 386, NOW, running in very fast processor and even making use of almost unlimited memory, let me say, it is not anymore weak. There are not anymore weak engines! Let me put it in extreme terms: even a lousy program that only take into account material is now strong IF can goes beyond 12 ply in its search. So simple as that. I have experienced that with every “weak” engine these days. Let me add I have good memory and I remember well how I was in conditions to beat even fritz 3 when running in my 386 with 32 Mg RAM. Not anymore. I challenge you to honestly play any of those despicable weak engines and tell me if you still win every game or, on the contrary, with great surprise you find yourself surpassed more than once because simply the “weak” engine can see 20 moves ahead and we not…..

Honest regards
Fern

[Martin Hertz]

I can confirm your observation as really true. I've made many tests with these programs agaist human
players at different speeds, 8088 @ 4.77 MHz vs 3.3 GHz, what roughly means 2 hours thinking time in
only 1 second. All the better chess players confirmed, that the programs are much stronger especialy
in tactical positions and making much less mistakes. An extreme exampe is the CP2150, because it finds
very often good or better moves in ply 6 or 7, thats only possible with very high speed in a reasonable
time. I think the reason is, that CP2150 is one of the rare real Shannon B type programs.

[spacious_mind]

I can confirm your observation as really true. I've made many tests with these programs agaist human
players at different speeds, 8088 @ 4.77 MHz vs 3.3 GHz, what roughly means 2 hours thinking time in
only 1 second. All the better chess players confirmed, that the programs are much stronger especialy
in tactical positions and making much less mistakes. An extreme exampe is the CP2150, because it finds
very often good or better moves in ply 6 or 7, thats only possible with very high speed in a reasonable
time. I think the reason is, that CP2150 is one of the rare real Shannon B type programs.

Yes, an engine that surprises me even more than CP2150 is Sargon 3 on Amiga. I have not included it in my Division 4 because I had played already against some good opponents such as Millennium Chess Genius and it is as good. Therefore you will be seeing Sargon 3 in Division 3. I have played it against Amiga CP2150, 2175, Complete Chess System, Psion on Atari ST at 32 MHz (can't do faster) and it mostly comes out ahead. I am amazed that Sargon 3 is so good on the Amiga and ST and so much worse on PC. Somehow it was never written as well for PC. Probably whoever did the PC conversion was not so good. But of course it also reaches 9-10 ply at 30 seconds search which helps it a lot and even more ply's in endgames.

A lack of ability to use hash is probably the only thing that would stop it from beating a V11 and I would expect it to win the odd game against V11 too. I have not tried it but will have to do that some time.

Well I guess from this write up you can surmise that Sargon 3 has become one of my favorite programs

Some Examples:

[Event "Computer Test Match"]
[Site "Alabama"]
[Date "2016.04.16"]
[Round "2"]
[White "Amiga 68060 Sargon 3, 30S AV.."]
[Black "Millennium Chessgenius, 30S AV.."]
[Result "1-0"]
[ECO "B11"]
[WhiteElo "1900"]
[BlackElo "2216"]
[Annotator "SM"]
[PlyCount "95"]
[EventDate "2016.04.16"]
[EventType "match"]
[EventRounds "2"]
[EventCountry "USA"]

1. e4 c6 2. Nf3 d5 3. Nc3 Bg4 4. h3 Bxf3 5. Qxf3 Nf6 6. d4 dxe4 7. Qe3 Nbd7 8. Nxe4 {Millennium Chessgenius out of book} Nxe4 9. Qxe4 e6 {Amiga 68060 Sargon III out of book} 10. Bd2 Nf6 11. Qe3 Be7 12. Bd3 O-O 13. O-O Qb6 14. c4 Rfd8 15. Bc3 Bb4 16. Qg3 Bxc3 17. bxc3 c5 18. Rab1 Qa6 19. dxc5 Rac8 20. Rfd1 Qc6 21. Qe3 Rd7 22. Bc2 Rxd1+ 23. Bxd1 Qa6 24. Qd4 Qa5 25. Rb5 Qa6 26. Bf3 Qxa2 27. Rxb7 Qa1+ 28. Kh2 a5 29. c6 Qa3 30. Ra7 h6 31. c5 Qb3 32. Rb7 e5 33. Rxb3 exd4 34. cxd4 Kf8 35. Rb5 a4 36. d5 a3 37. d6 a2 38. Ra5 Ke8 39. Rxa2 Kd8 40. Ra7 Ke8 41. d7+ Kd8 42. dxc8=Q+ Kxc8 43. Rxf7 Kd8 44. c7+ Kc8 45. c6 Nd7 46. Rxd7
g6 47. Be2 g5 48. Ba6# 1-0

[Event "Computer Test Match"]
[Site "?"]
[Date "2016.07.30"]
[Round "1"]
[White "AM68060 32MHz Sargon 3, 30S AV.."]
[Black "AT68060 32MHz Psion 2.01, 30S AV.."]
[Result "1-0"]
[ECO "A09"]
[WhiteElo "2100"]
[BlackElo "2107"]
[Annotator "SM"]
[PlyCount "145"]
[EventDate "2016.07.30"]
[EventType "match"]
[EventRounds "2"]

1. Nf3 d5 2. c4 d4 3. g3 Nc6 4. Bg2 e5 5. d3 Nf6 6. O-O {AT68060 32MHz Psion 2. 01 out of book} Bd6 {AM68060 32MHz Sargon 3 out of book} 7. Bg5 O-O 8. Nbd2 h6 9. Bxf6 Qxf6 10. Ne4 Qf5 11. Nxd6 cxd6 12. Qb3 Rd8 13. e3 dxe3 14. fxe3 Qg4 15. Ne1 Rd7 16. Nc2 Re7 17. Qa3 Rd7 18. Bd5 Nd8 19. Bf3 Qg5 20. Bg2 Nc6 21. Bd5 Qg6 22. Rf2 Kh8 23. Raf1 f6 24. Be4 Qg5 25. Bg2 Qg6 26. d4 Rd8 27. d5 Ne7 28. c5 dxc5 29. Qxc5 Qe8 30. Rd2 Be6 31. e4 Bd7 32. Ne3 Rdc8 33. Qb4 Bb5 34. Ra1 a5 35. Qb3 Qd7 36. Bh3 Qxh3 37. Qxb5 Ra7 38. Qb6 Rca8 39. Rc1 Ra6 40. Qxb7 R6a7 41. Qb5 Qd7 42. Qxd7 Rxd7 43. d6 Kg8 44. Rc7 Rxc7 45. dxc7 Nc8 46. Rd8+ Kf7 47. Nf5 Ra6 48. Rxc8 Rc6 49. Nd6+ Kg6 50. Nb5 a4 51. Rd8 Rc1+ 52. Kg2 Rc2+ 53. Kf3 Rc4 54. c8=Q Rxc8 55. Rxc8 Kf7 56. Nd6+ Ke6 57. Nf5 g6 58. Nxh6 Kd6 59. Ra8 Kc5 60. Ra6 Kb4 61. h4 g5 62. h5 Kc4 63. Rxa4+ Kd3 64. Nf5 Kc2 65. b4 Kc3 66. b5 Kc2 67. h6 Kc3 68. h7 Kc2 69. h8=Q Kb2 70. Ke3 Kc2 71. Qh2+ Kc1 72. Rb4 g4 73. Qd2# 1-0

[Event "Computer Test Match"]
[Site "?"]
[Date "2016.07.31"]
[Round "2"]
[White "AT68060 32MHz Psion 2.01, 30S AV.."]
[Black "AM68060 32MHz Sargon 3, 30S AV.."]
[Result "0-1"]
[ECO "A06"]
[WhiteElo "2107"]
[BlackElo "2100"]
[Annotator "SM"]
[PlyCount "104"]
[EventDate "2016.07.31"]
[EventType "match"]
[EventRounds "2"]

1. Nf3 d5 2. e3 {AM68060 32MHz Sargon 3 out of book} Nf6 {AT68060 32MHz Psion 2.01 out of book} 3. Nc3 e6 4. Be2 Bd6 5. O-O Nc6 6. d3 a6 7. d4 O-O 8. a3 Bd7 9. b3 Ne4 10. Nxe4 dxe4 11. Nd2 f5 12. Nc4 Qf6 13. Bb2 Qh6 14. Nxd6 cxd6 15. d5 Ne5 16. dxe6 Qxe6 17. Qd4 Rf6 18. c4 Raf8 19. Rad1 Rg6 20. Bh5 Rgf6 21. Bc3 Bc6 22. a4 Rh6 23. Be2 Rg6 24. Bb4 Rd8 25. g3 Rf6 26. Ba5 Rd7 27. Bc3 Qf7 28. Rc1 Rh6 29. Rfd1 Re7 30. Rc2 Ree6 31. Qb6 Nd7 32. Qc7 Qf8 33. Rcd2 Nc5 34. Rb1 Re8 35. Qb6 Nd7 36. Qd4 Ree6 37. Bb2 Ne5 38. Re1 Rh3 39. Kg2 Nf3 40. Bxf3 exf3+ 41. Kg1 Reh6 42. Qxg7+ Qxg7 43. Bxg7 Kxg7 44. Rd4 Be4 45. Rxd6 Rxd6 46. c5 Rdh6 47.
Rd1 Rxh2 48. Rd7+ Kf8 49. Rd8+ Ke7 50. Rd7+ Kxd7 51. c6+ Bxc6 52. a5 Rh1# 0-1

[Event "Computer Test Match"]
[Site "?"]
[Date "2016.07.30"]
[Round "2"]
[White "AM68060 32MHz CC System, 30S AV.."]
[Black "AM68060 32MHz Sargon 3, 30S AV.."]
[Result "1/2-1/2"]
[ECO "A67"]
[WhiteElo "2100"]
[BlackElo "2107"]
[Annotator ""]
[PlyCount "173"]
[EventDate "2016.07.30"]
[EventType "match"]
[EventRounds "2"]

1. d4 Nf6 2. c4 c5 3. d5 e6 4. Nc3 exd5 5. cxd5 d6 6. e4 g6 7. f4 Bg7 8. Bb5+ Nfd7 9. Nf3 a6 {AM68060 32MHz CC System out of book} 10. Bd3 {AM68060 32MHz Sargon 3 out of book} O-O 11. O-O Qa5 12. Qb3 b5 13. Qc2 c4 14. Be2 f5 15. exf5 Rxf5 16. Rd1 Qb6+ 17. Kh1 Bb7 18. Nh4 Rf8 19. f5 Bf6 20. fxg6 Bxh4 21. Bh6 Rf6 22. Qe4 Qd8 23. gxh7+ Kh8 24. Be3 Rf8 25. a4 Qe8 26. Qxh4 Qxe3 27. Rf1 Rc8 28. Rf7 b4 29. Re7 Qd2 30. Qg4 Qh6 31. Ne4 Bxd5 32. Nxd6 Rf8 33. Bf3 Qf6 34. Re8 Ne5 35. Qh4 Qxd6 36. Rxf8+ Qxf8 37. Bxd5 Nbc6 38. Bxc6 Nxc6 39. Qxc4 Qd6 40. Rf1 Ne5 41. Qe2 Nd3 42. Qg4 Rd8 43. Qf5 Nxb2 44. Qf7 Nxa4 45. Qa2 Nc3 46. Qf2
Rd7 47. Qf5 Nb5 48. Qh5 Nc7 49. Qf5 Ne6 50. h3 Rxh7 51. Qg6 Rg7 52. Qe8+ Kh7 53. Qh5+ Kg8 54. Qe8+ Nf8 55. Qc8 a5 56. Qc4+ Kh8 57. Rf5 Ra7 58. Qc1 a4 59. Qe3 Ra8 60. Qe4 Rb8 61. Rh5+ Kg7 62. Rg5+ Ng6 63. Rxg6+ Qxg6 64. Qe5+ Qf6 65. Qxb8 Qa1+ 66. Kh2 b3 67. Qb7+ Kf6 68. Qc6+ Ke5 69. Qe8+ Kd6 70. Qd8+ Kc6 71. Qc8+ Kd5 72. Qa8+ Kc5 73. Qf8+ Kc4 74. Qc8+ Kd3 75. Qf5+ Ke3 76. Qf3+ Kd2 77. Qd5+ Ke1 78. Qe6+ Kd1 79. Qd5+ Kc2 80. Qe4+ Kb2 81. Qd4+ Kb1 82. Qe4+ Kc1 83.
Qc4+ Kb2 84. Qd4+ Kb1 85. Qe4+ Kc1 86. Qc4+ Kb2 87. Qd4+ {Draw by 3x repetition} 1/2-1/2

[Event "Computer Test Match"]
[Site "?"]
[Date "2016.07.23"]
[Round "2"]
[White "AM68060 32MHz CC2175, 30S AV.."]
[Black "AM68060 32MHz Sargon 3, 30S AV.."]
[Result "0-1"]
[ECO "D80"]
[WhiteElo "2107"]
[BlackElo "2100"]
[Annotator "SM"]
[PlyCount "56"]
[EventDate "2016.07.23"]
[EventType "match"]
[EventRounds "2"]

1. d4 Nf6 2. c4 g6 3. Nc3 d5 4. Bg5 Ne4 5. cxd5 Nxg5 6. h4 Ne4 {AM68060 32MHz Sargon 3 out of book} 7. Nxe4 Qxd5 8. Nc3 Qa5 9. Nf3 {AM68060 32MHz Sargon 3 out of book} Bg7 10. e3 O-O 11. Bc4 c5 12. h5 Nc6 13. hxg6 hxg6 14. Rc1 cxd4 15. Nxd4 Nxd4 16. exd4 Rd8 17. Qb3 e6 18. O-O Rxd4 19. Rfd1 Rh4 20. Nb5 a6 21. Nc3 Qh5 22. g3 Rh1+ 23. Kg2 Rh2+ 24. Kf1 Qf3 25. Rd8+ Kh7 26. Ke1 Qxf2+ 27. Kd1 Qg1+ 28. Bf1 Qxf1# 0-1

28 moves against CC2175

Its games like these that inspired me to create the League that I am playing.

Best regards

[Martin Hertz]

Yes, Sargon3 is not bad but somtimes it's also luck. I've analysed the last game and noticed bad moves on both sides, until this position came up:

[fen]r1b3k1/1p3pb1/p3p1p1/qN6/2B4r/1Q6/PP3PP1/2RR2K1 w - - 0 21[/fen]

The only possible playable move is Qg3, that needs ply 9 and 11 seconds on 3.3 GHz with 8MB hash
for the CC2175 to avoid that blunder in the game. CP2150 needs only ply 7 but 65 seconds. Even
Sargon3 needs ply 8 and about 10 seconds to avoid the blunder.

I've made some further tests for comparing CP2150, CC2175 and Complete Chess System. CC2175 and CCS
are very similar in the searching style, but the evalution seems very different. CP2150 is also
Shannon B, but totaly different to CC2175 or CCS. In test positions all 3 programs are very
different in finding moves and the neccessary ply.

I'm going to make a small round robin with some programs at 3.3 GHz, to see if there is a significant
difference, because maybe Shannon B scales better with more speed. I'll post the result.

[Martin Hertz]

Nick, I think you are right with Sargon3 on PC, because it seems too slow compared with your 68060 emu.
Some tests shows that it's weaker than CC2175 at 3.3 GHz. I assume, the PC version is not written in
assembly but is a compiled language, what surely means 1 ply less.

[fen]8/8/8/8/8/5N2/5R2/3k3K w - - 0 1 [/fen]

Sargon3 finds the mate in 6 in ply 11 that takes 2.5 minutes at 3.3 GHz. This should be faster for a such
simple position on a PC. If your 68060 emu is really faster, than the PC version is a very inefficient port.

[spacious_mind]

Nick, I think you are right with Sargon3 on PC, because it seems too slow compared with your 68060 emu.
Some tests shows that it's weaker than CC2175 at 3.3 GHz. I assume, the PC version is not written in
assembly but is a compiled language, what surely means 1 ply less.

[fen]8/8/8/8/8/5N2/5R2/3k3K w - - 0 1 [/fen]

Sargon3 finds the mate in 6 in ply 11 that takes 2.5 minutes at 3.3 GHz. This should be faster for a such
simple position on a PC. If your 68060 emu is really faster, than the PC version is a very inefficient port.

Hi Martin,

I will try and test this position with both Amiga and Atari tomorrow and let you know.

Thanks

[spacious_mind]

Just to show you that even a spectrum program can beat Novag Scorpio. Here is Spectrum Clock Chess which is a predecessor to CP2150 on the Sinclair Spectrum. In this game I simulated an estimated speed of a Spectrum 116 MHz. Clock Chess is I think the equivalent of Superchess 3.5.

[Event "Computer Test Match"]
[Site "Alabama"]
[Date "2016.04.12"]
[Round "1"]
[White "Novag Scorpio, 30S AV.."]
[Black "Spectrum 116MHz Clockchess 89, 30S AV.."]
[Result "0-1"]
[ECO "C41"]
[WhiteElo "2107"]
[BlackElo "2204"]
[Annotator "SM"]
[PlyCount "146"]
[EventDate "2016.04.12"]
[EventType "match"]
[EventRounds "2"]
[EventCountry "USA"]

1. e4 e5 2. Nf3 f5 3. exf5 {Spectrum 1024K 116.2MHz Clockchess 89 out of book} d6 {Novag Scorpio out of book} 4. d4 Bxf5 5. dxe5 Nc6 6. exd6 Bxd6 7. Nc3 Bc5 8. Bd2 Qe7+ 9. Be2 O-O-O 10. O-O Nb4 11. Rc1 Nf6 12. Bb5 a6 13. Ba4 Kb8 14. Re1 Qd6 15. a3 Ng4 16. axb4 Bxf2+ 17. Kh1 Bxe1 18. Qxe1 Nxh2 19. Nxh2 Qxd2 20. b5 a5 21. b6 Qxe1+ 22. Rxe1 cxb6 23. Re7 g6 24. Kg1 b5 25. Nxb5 Rd1+ 26. Kf2 Rd2+ 27. Re2 Rhd8 28. Nf3 Rxe2+ 29. Kxe2 Be4 30. Ke3 Bxf3 31. Kxf3 h5 32. Ke3 Re8+ 33. Kf4 Rc8 34. Nc3 Rc4+ 35. Kf3 g5 36. Bb3 Rf4+ 37. Ke3 Rb4 38. Nd5 Rg4 39. Kf3 a4 40. Ba2 Rd4 41. Ke2 Ka7 42. c3 Re4+ 43. Kd2 Rg4 44. Ne3 Rf4 45. Ke2 g4 46. Nd5 Rf5 47. Ke3 h4 48. Nf4 Rb5 49. Nd3 h3 50. gxh3 gxh3 51. Kf3 Rg5 52. Nf2
h2 53. Ke3 Rb5 54. b4 axb3 55. Bb1 Ra5 56. Kf3 Ra1 57. Bd3 h1=Q+ 58. Nxh1 Rxh1 59. Ke3 b2 60. Bc2 Kb6 61. Kd4 b1=Q 62. Bxb1 Rxb1 63. Kd3 Kc5 64. Kc2 Rh1 65. Kb3 b5 66. Kc2 Rh3 67. Kb3 b4 68. Kc2 bxc3 69. Kb3 Rd3 70. Kc2 Kc4 71. Kb1 Kb3 72. Kc1 Rd2 73. Kb1 Rd1# 0-1

Well speeding it gave it the ability to beat Novag Scorpio.

Another experiment Apple ][+ 48K Sargon with approx. 194 MHZ simulated against Millennium Chessgenius.

[Event "Computer Chess Match"]
[Site "?"]
[Date "2016.05.29"]
[Round "1"]
[White "Millennium Chessgenius, 30S AV.."]
[Black "Apple ][+ 194 MHZ 48K Sargon , 30S AV.."]
[Result "1/2-1/2"]
[ECO "A33"]
[WhiteElo "2226"]
[BlackElo "2200"]
[Annotator "SM"]
[PlyCount "100"]
[EventDate "2016.05.29"]
[EventRounds "2"]

1. c4 c5 2. Nf3 Nf6 3. d4 cxd4 4. Nxd4 e6 5. Nc3 Nc6 6. Bf4 {Apple ][+ 194 MHZ 48K Sargon 3 out of book} Bb4 7. Ndb5 e5 8. Bg5 O-O 9. a3 Bxc3+ {Millennium Chessgenius out of book} 10. Nxc3 h6 11. Bh4 g5 12. Bg3 d6 13. e3 h5 14. h4 g4 15. Bd3 Be6 16. O-O d5 17. cxd5 Nxd5 18. Ne4 Bf5 19. Qb3 Nb6 20. Rfd1 Be6 21. Qc3 Nd5 22. Qc5 Rc8 23. Ng5 Qb6 24. Nxe6 fxe6 25. Qxb6 axb6 26. Rac1 Nf6 27. Bc4 Kf7 28. Rd6 Rfe8 29. Rcd1 Ne4 30. Rd7+ Kf6 31. Bd3 Nc5 32. Rd6 Red8 33. Bb5 Rxd6 34. Rxd6 Ne4 35. Rd1 Ne7 36. Bd3 Nxg3 37. fxg3 Nd5 38. Kf2 Ke7 39. Ke2 Nf6 40. Rd2 e4 41. Bc2 b5 42. Bb1 Rc1 43. Bc2 Rg1 44. Kf2 Rc1 45. Bb3 Nd7 46. Rc2 Rb1 47. Ke2 Rg1 48. Kf2 Rb1 49. Ke2 Rg1 50. Kf2 Rb1 {Draw by 3x repetition}
1/2-1/2

As you can see Sargon 3 even on an Apple II emulator if you could speed it sufficiently could compete with some very good dedicated computers.

Best regards

[spacious_mind]

Nick, I think you are right with Sargon3 on PC, because it seems too slow compared with your 68060 emu.
Some tests shows that it's weaker than CC2175 at 3.3 GHz. I assume, the PC version is not written in
assembly but is a compiled language, what surely means 1 ply less.

[fen]8/8/8/8/8/5N2/5R2/3k3K w - - 0 1 [/fen]

Sargon3 finds the mate in 6 in ply 11 that takes 2.5 minutes at 3.3 GHz. This should be faster for a such
simple position on a PC. If your 68060 emu is really faster, than the PC version is a very inefficient port.

Hi Martin,

I just tested WINUAE Amiga at 68060. Sargon 3 set at infinite level takes 10 minutes and 20 seconds to find the right move at 11 Ply.

Therefore just approximately Amiga on my computer in WINUAE at 68060 plays at something like your computer and setting and speed of 800-825 MHz.

Therefore can you do me a favor please and set up Chessmaster 2100 in D-Fend with exact settings that I use which you can see here:






Please set up Chessmaster 2100 in the war room to play move 30 seconds per move and start white with move f3. Chessmaster 2100 will move instantly therefore follow with next move g3. Chessmaster 2100 will calculate g3 and move at 30 seconds.

Then also after you have tried D-Fend, please do the same with your DOS 3.3 GHz setting.

We can then compare amount of positions calculated by your Chessmaster 2100 to Amiga.

Below you can see that WINUAE with my 68060 setting searched 6,790,608 positions



My DOSBox D-Fend setting that you will be testing searched 756,678 positions, which you can see below.



It would be interesting to exactly compare your results.

Thanks and best regards

[Tibono2]

Hi Nick,
if I may suggest: try again your DFend test with Chessmaster 2100 with emulation core set to 'Dynamic'. You should get far better performance.
This means the code will be dynamically compiled rather than interpreted step by step by the DosBox emulation. This is very efficient with cycling parts of the code.
Most programs behave correctly with this feature. Some do not: set back core to 'auto' programs that fail to launch or abort whith 'Dynamic' setting.
kind regards,
Tibono

[spacious_mind]

Hi Nick,
if I may suggest: try again your DFend test with Chessmaster 2100 with emulation core set to 'Dynamic'. You should get far better performance.
This means the code will be dynamically compiled rather than interpreted step by step by the DosBox emulation. This is very efficient with cycling parts of the code.
Most programs behave correctly with this feature. Some do not: set back core to 'auto' programs that fail to launch or abort whith 'Dynamic' setting.
kind regards,
Tibono

Hi Tibono,

Thanks I will give it a go. I know I will need to do something with D-Fend as I later play the higher leagues with stronger programs. Ed has his Rebels set to good speeds and the other programs will have to match it. Probably once I get to Division 2 I might switch to a favorable frames value that favors all the programs. In Division 4 I have found for example that 63,000 frames works for all the Division 4 programs and has them searching at close to what they do at their 100% Max setting.

Best regards

[Martin Hertz]

Unfortunately I cannot test the CM2000, CM2100 or CM3000 at 3.3 GHz, because after starting the calculation, the programs
show abnormal values. The problem is, that not all DOS programs are executable on that speed any more. A good program for
speed comparison should be Colossus, because it's availabe on many systems and always written in assembly. As mentioned in
another thread, Colossus X with Lev Infinite tye 1 (selective) needs 9 seconds to play the move at 3.3 GHz. If you want to
test another program, I can try it.

[spacious_mind]

Unfortunately I cannot test the CM2000, CM2100 or CM3000 at 3.3 GHz, because after starting the calculation, the programs
show abnormal values. The problem is, that not all DOS programs are executable on that speed any more. A good program for
speed comparison should be Colossus, because it's availabe on many systems and always written in assembly. As mentioned in
another thread, Colossus X with Lev Infinite tye 1 (selective) needs 9 seconds to play the move at 3.3 GHz. If you want to
test another program, I can try it.

Hi Martin,

Yes the problem is that Colossus will not work on Amiga or Atari at 68060 so I cannot compare it to the best that is possible on these systems. Complete Chess System shows positions but it is not the same PC program I don't think.

Pity CM2100 shows positions so it was a natural choice to compare. I can't think of another that could do this. Psion yes for St but it doesn't exist on Amiga. ST is restricted to 32 MHz.

Best regards

[Martin Hertz]

I was able to measure the speed of the CM2100 under real DOS at 3.3 GHz for your test position: 2,350,000 pos/sec
thats around 10 times faster than the 68060 emulation. In the endgame CM2100 reaches up to 5,000,000 pos/sec.

[spacious_mind]

I was able to measure the speed of the CM2100 under real DOS at 3.3 GHz for your test position: 2,350,000 pos/sec
thats around 10 times faster than the 68060 emulation. In the endgame CM2100 reaches up to 5,000,000 pos/sec.

Ok that is different to the 2.5 minutes for the 11 ply finding mate test you did with Sargon 3. With that the difference was just a little over 4 times.

So it varies maybe by program? CM2100 I would trust more since they seem to have all of the same chess behaviors. PC Sargon 3 I don't trust, I think it is much worse a program on PC, it is not a very good conversion.

Thanks

[spacious_mind]

Martin,

I just tried Complete Chess Systems. I take it back they do seem to be quite similar.

The Test Position D-Fend at 100% auto takes about 30 seconds to find the mate.

Amiga 68060 takes about 10 seconds to find the mate.

Based on this your computer at 3.3 GHz should be very quick. Please let me know how it does Complete Chess Systems.

Best regards