lkaufman wrote: ↑Thu Sep 16, 2021 9:04 pm
Chessqueen wrote: ↑Wed Sep 15, 2021 10:23 pm
Well Mephisto Amsterdam and Roma etc were unit that calculated either 1 or 2 plies at the most with dedicated chips either at 12Mhz or 20 Mhz at the most, if you match them against low rated engines like Cicada rated 1536 at the most and plat at fast time control 2'+1" Mephisto will beat it since Cicada will not be able to calculate more than 8 plies at that fast speed, but it does not matter if you give Mephisto more time it will still only calculate up depth 1 on my intel i7 3.40Ghz and once it reaches that which is within 1 to 2 seconds it will keep that move before it releases it. Mephisto might still beat either of this low rated engines at TC 2'+1" .
463‑464 Storm 0.6 1571 +21 −21 38.3% +94.8 15.0% 925 74.3%
465 Damas 9 1559 +25 −25 47.8% +17.6 17.6% 626 76.1%
466 IQ23.003 1547 +21 −21 33.6% +118.2 32.6% 854 72.6%
467 Cicada 0.1 64-bit 1536 +24 −25 44.1% +48.0 19.7% 636
1: Amsterdam-UCI 11.0/20 ···················· 11110=00000=1=11101= 99.00
2: Cicada 9.0/20 00001=11111=0=00010= ···················· 99.00
20 games played / Tournament is finished
Name of the tournament: Fire8
Site/ Country: MININT-UB2PIMJ, United States
Level: Blitz 1/1
Hardware: Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz with 15.9 GB Memory
Operating system: Windows 10 Enterprise Professional (Build 9200) 64 bit
PGN-File: C:\Program Files (x86)\Arena\Tournaments\IGel.pgn
Website:
E-Mail Address:
Very interesting. Some questions and points:
1. Where do you get these Mephisto engines to run on PC?
2. If they report depth 1 or 2, that is just full ply depth, it's like saying that if you run Dragon on Stockfish to 10 plies it will only be doing 1 or 2 plies full depth, the rest are pruned to various degrees. Mephisto was also selective, though it worked differently. I think they had about 6 or 7 selective plies.
3. Something is very strange here. Mephisto Amsterdam, running on a 68000 processor initially at perhaps 12 MHz, played at somewhere around USCF 2150 level (supposedly FIDE 2050 level) against humans at classical time controls, although much lower on engine lists. Now it would presumably be running more than a hundred times faster, should be something like 2400 FIDE or so vs. humans. Yet you are getting a 1570 CCRL result! Now I know CCRL ratings are lower than human ones, especially at lower levels, but not like 800 elo! Is there something wrong with the version, or am I misunderstanding something? This seems ridiculous.
4. Have you tried playing any rapid games yourself with this pc version of Mephisto Amsterdam? Even the original dedicated model should be a tough match for you at 1950 FIDE. Perhaps that would give us a clue.
Yes, I just played against the Mephisto Roma Uci using my Intel i7 3.40, but I noticed that it always play depth 1 no matter if I set it to play game in 10 minutes or game in 5 seconds per move, which I decided it was fair for me to make a move every 5 seconds
[pgn][Event "Testing Mephisto Roma uci"]
[Date "2021.09.16"]
[Round "1"]
[White "Roma32-UCI"]
[Black "ChessQuuen"]
[Result "0-1"]
[BlackElo "1950"]
[ECO "E32"]
[Opening "Nimzo-Indian"]
[Variation "Classical, 4...O-O 5.Nf3"]
[WhiteElo "1950"]
[TimeControl "0+5"]
[Termination "adjudication"]
[PlyCount "102"]
[WhiteType "program"]
[BlackType "human"]
1. d4 e6 2. c4 Nf6 3. Nc3 Bb4 4. Qc2 O-O 5. Nf3 d5 6. e3 Nc6 7. Bd3 Bd6 8.
a3 dxc4 9. Bxc4 e5 10. O-O exd4 11. Nxd4 Nxd4 12. exd4 Ng4 13. g3 Bd7 14.
Qe4 c6 15. Bd3 Nf6 16. Qe2 Bh3 17. Re1 Bg4 18. Qc2 Bc7 19. Bg5 h6 20. Be3
Bb6 21. Ne2 Bxe2 22. Bxe2 Nd5 23. Bd3 Nxe3 24. fxe3 Qd6 25. Rf1 c5 26. Bh7+
Kh8 27. Be4 cxd4 28. Rad1 Rae8 29. Bxb7 Qe7 30. Bg2 dxe3 31. Qe2 Kg8 32. b4
Rd8 33. Rd5 Rxd5 34. Bxd5 Qe5 35. Bb3 Qc3 36. Rb1 Rd8 37. Kh1 Qc6+ 38. Qg2
Qxg2+ 39. Kxg2 Rd2+ 40. Kh3 e2 41. Kg4 Bf2 42. Kf3 e1=Q 43. Rxe1 Bxe1 44.
Bc4 Rxh2 45. g4 Bh4 46. Ke4 Bg5 47. Bd5 Rf2 48. Ke5 Rf4 49. Bc6 Rxg4 50.
Kf5 Rh4 51. Be4 Bf4[/pgn]
