Colossus 2025a now available

[MartinBryant]

I have just released an updated version of my Colossus engine on my website at http://www.colossusgames.somee.com

There are a couple of bug fixes (in particular lazy-SMP now works properly again) and tweaks from the previous version 2024b.
My tests indicate an improvement of about 15-20 ELO.

[peter]

Thanks for the new version!

Single thread 3'+1", UHO 2024 900-990cp 6mvs:

Score of lc0CPU0.32 vs Colossus2025a: 231 - 163 - 106 [0.568]
Elo difference: 47.5 +/- 27.2, LOS: 100.0 %, DrawRatio: 21.2 %
500 of 500 games finished.

Regards

[chessica]

Analyse von C:\Users\Privat\Desktop\Chess-Tools\Datenbanken pgn epd\epd\ERET Eigenmann Rapid Engine Chess.epd
Analysemotor: Colossus2025a
13.02.2025 10:28:50 Spielstufe: 15 Sekunden
Quelle : Liste von EPD/PGN Dateien
C:\Users\Privat\Desktop\Chess-Tools\Datenbanken pgn epd\epd\ERET Eigenmann Rapid Engine Chess.epd
Farben : Weiße Züge=True, Schwarze Züge=True
Richtung : vorwärts
Partien (PGN) : Von 1 bis 999999
Züge : Von 1 bis 250
Stellungen (EPD) : Alle Stellungen
Verwende Motor(en) : Geladene Motoren=Motor 1
Sekunden pro Zug verwenden : 15 Sekunden pro Zug
Analyse&zeilen : Speichern ab Such&tiefe=10
Analysen abbrechen aktivieren: False
1) Tf1xf4 Dh5-h7
2) Sf8-g6 Sf8xe6
3) f4-f5 Ld3-e2
4) Sf3xg5 Kd3-c2
5) Lb3xf7 Sc3-d5
6) .. Sd7-c5 De4-c6
7) .. Sd7-e5 Lc5xd4
8) Db3xd5 Db3xb7
9) Se5-g6 h2-h3
10) .. d5-d4 Df6-h4
11) d4xe5, Sg1-f3 Sg1-f3 * 7 Sekunden
12) Sc3xd5 Sc3xd5 * 0 Sekunden
13) De2-d2, Ld3-c2 De2-d2 * 12 Sekunden
14) Sd2-f3 Sc3xa4
15) Dh3-d3 Dh3-f1
16) Sa5-c6 h2-h4
17) .. Tb8xb2 Df6-f5
18) .. Da5-g5 h4-h3
19) Se3-f5 Sf3xd4
20) Ld3xh7 Sa4-c3
21) .. Td8-c8 d6-d5
22) h2-h4 e4-e5
23) Sh4-g6 Se3-f5
24) b2-b3(?) Lf1-e2 * 0 Sekunden
25) Lg5xd8 Lg5-d2
26) Sh4xg6 f4xe5
27) g2-g4 Kg6-g5
28) Ta7xb7(?) Ta7xb7
29) Tc1xc5 Tc1xc5 * 3 Sekunden
30) Sf3-h4 Dc2-d1
31) b3-b4 Kc1-b2
32) Sf1-g3 a2-a3
33) Sf3xe5 Se4-g5
34) Tc1-c8 Tc1-c8 * 0 Sekunden
35) Sa3xb5 Lc4xb5
36) b6-b7 Tf5-f8
37) Sc7-a8 Tc2-f2
38) e4xd5 e4xd5 * 0 Sekunden
39) a2-a3 e2-e4
40) .. Lf8-c5 Sd7-c5
41) Sf3xg5 Dh4-g3
42) .. Kc7-b7 Kc7-b7 * 0 Sekunden
43) Sd4xf5 g4xf5
44) h2-h4 h2-h4 * 0 Sekunden
45) Tb3-b4 Sg3-f1
46) Sf4-d3 h4xg5
47) Sc5xb7 Sc5xb7 * 3 Sekunden
48) c4xd5 c4xd5 * 0 Sekunden
49) f2-f4 Se4-c3
50) .. Se5-f3 Kh8-g8
51) .. Lg7-h6 Lg7-h6 * 1 Sekunde
52) b2-b4 Tf1-e1
53) .. c5-c4 a7-a6
54) .. c4-c3 Te8-d8
55) Lf4xh6 Dd2-e3
56) Lf1-e2 Sg5-e4
57) d5-d6 d5-d6 * 0 Sekunden
58) h2-h4 Ta6-a3
59) De3-d3 h4xg5
60) .. Da8xd8 e4xf3
61) .. Tf8-f6 d4-d3
62) g5-g6 h5-h6
63) .. Ke8-f8 Ke8-f8 * 0 Sekunden
64) .. Tf1xf2(?) Tf1xf2
65) .. Dc7-b6 Tf8-c8
66) Kf2-e1 Td5xd4
67) .. f4-f3 Tf8-a8
68) a2-a3 Sd2-e4
69) h4-h5 h4-h5 * 10 Sekunden
70) a2-a3 e4-e5
71) f4-f5 f4-f5 * 0 Sekunden
72) .. Da8-d5 Da8-d5 * 3 Sekunden
73) Dd1xe2 Te1xe2
74) Kd3-c4 Kd3-e3
75) .. Dh5-f5 g4xh3
76) Lg5xf6 Lg5xf6 * 3 Sekunden
77) g5xf6 Lb2xc3
78) Tf1xf6 Tf1xf6 * 8 Sekunden
79) De2xe8 Ld4xf6
80) .. Lg5-c1 Lg5-c1 * 10 Sekunden
81) Lh6-g7 Sf5-h4
82) g3-g4 g3-g4 * 3 Sekunden
83) Sc3-b5 Sc3-b5 * 8 Sekunden
84) Lh6-g7 Lh6-f4
85) Se5-g4 Se5xf7
86) c4-c5 f2xe3
87) e4-e5 Sd2-f1
88) g3-g4 g3-g4 * 2 Sekunden
89) e7-e8S Td5-d1
90) f3-f4 Ke8-d7
91) .. g7-g6, Kd6-c6 g7-g6 * 5 Sekunden
92) .. Dg5xg2(?) c7-c6 * 11 Sekunden
93) Sf7-d6 Tg3-f3
94) .. Td6-f6 Lg5-e3
95) Sf3xg5 Lc1xg5
96) Ke5-f6 Df3-f8
97) c4-c5 Lf2-e3
98) Kf1-e2(?) Kf1-e2
99) Te7-e8 c4xd5
100) Sd4-f5 Sd4-f5 * 7 Sekunden
101) .. Sd7-e5 Sd7-e5 * 3 Sekunden
102) .. Sa6xb4 Tf8xa8
103) Kb7-c8 a5-a6
104) .. Sc6-e7 Lc8-e6
105) e4-e5 d4-d5
106) .. Le7xf6 Kg8-h8
107) .. Tb8-b5 Tb8-b5 * 5 Sekunden
108) Lb3-a4 Lb3-c2
109) .. Dc4xd4 e6-e5
110) c4-c5 a4-a5
111) a2-a3 a2-a3 * 0 Sekunden
29 von 111 Zügen gefunden
13.02.2025 10:58:22, Gesamt Dauer: 00:29:31 Wertungszeit: 22:14 = 1334 Sekunden

[chessica]

[d]7k/5pp1/8/8/8/8/1K6/R7 w - - 0 1

Colossus 2025a:

Code: Select all

 12	00:00,087	 619k	7.109k	+3,78	Ta1-a7 Kh8-g8 Kb2-c3 g7-g5 Ta7-a5 f7-f6 Ta5-a7 Kg8-f8 Ta7-c7 Kf8-e8 Kc3-d4 g5-g4
 13	00:00,174	 1.264k	7.266k	+3,49	Ta1-a7 Kh8-g8 Kb2-c2 g7-g6 Kc2-c3 Kg8-g7 Ta7-a6 g6-g5 Kc3-c4 g5-g4 Ta6-d6 f7-f5 Kc4-c5
 13	00:00,356	 2.555k	7.178k	+3,63	Ta1-h1+ Kh8-g8 Kb2-c3 f7-f6 Th1-d1 Kg8-h7 Td1-d7 Kh7-g6 Kc3-d4 Kg6-h6 Kd4-d5 g7-g5 Td7-d8 Kh6-g6 Kd5-c6
 14	00:00,823	 5.820k	7.072k	+3,81	Ta1-h1+ Kh8-g8 Kb2-c3 f7-f6 Th1-d1 Kg8-h7 Kc3-d4 Kh7-g6 Kd4-e4 Kg6-f7 Td1-d7+ Kf7-g6 Td7-b7 Kg6-h6 Ke4-f5 Kh6-h7 Tb7-b8
 14	00:00,860	 6.010k	6.988k	+4,28	Ta1-a7 Kh8-g8 Kb2-c3 g7-g5 Kc3-d4 Kg8-g7 Kd4-e4 Kg7-g6 Ta7-a6+ Kg6-h5 Ta6-f6 Kh5-g4 Ke4-e3 Kg4-g3 Tf6-f3+ Kg3-g4
 15	00:01,007	 6.698k	6.651k	+4,47	Ta1-a7 Kh8-g8 Kb2-c3 g7-g5 Kc3-d4 Kg8-g7 Kd4-e5 Kg7-g6 Ta7-a6+ Kg6-h5 Ke5-f5 g5-g4 Ta6-f6 Kh5-h4 Tf6xf7
 16	00:02,072	 11.156k	5.384k	+4,42	Ta1-a7 Kh8-g8 Kb2-c3 g7-g5 Kc3-d4 Kg8-g7 Kd4-e4 Kg7-g6 Ta7-a6+ f7-f6 Ta6-c6 g5-g4 Ke4-f4 g4-g3 Kf4xg3 Kg6-f5 Tc6-c8
 17	00:04,553	 21.637k	4.752k	+4,50	Ta1-a7 Kh8-g8 Kb2-c3 g7-g5 Ta7-a5 f7-f6 Ta5-a7 g5-g4 Kc3-d4 g4-g3 Ta7-a3 g3-g2 Ta3-g3+ Kg8-f7 Tg3xg2 Kf7-e6 Tg2-g7 Ke6-f5
 18/30	00:08,099	 36.719k	4.534k	+4,45	Ta1-a7 Kh8-g8 Kb2-c3 g7-g5 Ta7-a5 f7-f6 Kc3-d4 Kg8-f7 Kd4-e4 Kf7-g6 Ta5-c5 g5-g4 Ke4-f4 Kg6-f7 Kf4xg4 Kf7-e6 Kg4-f4
 18/30	00:17,188	 71.849k	4.180k	+4,50	Ta1-h1+ Kh8-g8 Kb2-c3 f7-f6 Th1-a1 Kg8-h7 Kc3-d4 g7-g5 Kd4-e4 Kh7-g6 Ta1-f1 Kg6-f7 Ke4-f5 Kf7-e7 Tf1-e1+ Ke7-d7 Kf5xf6
 19/32	00:22,013	 89.558k	4.068k	+5,51	Ta1-h1+ Kh8-g8 Kb2-c3 f7-f6 Th1-a1 Kg8-f7 Ta1-a7+ Kf7-e6 Ta7xg7 Ke6-e5 Kc3-d3 Ke5-f4 Kd3-d4 Kf4-f3 Tg7-f7 Kf3-e2 Tf7xf6 Ke2-d2 Tf6-f2+ Kd2-c1
 19/32	00:23,292	 93.601k	4.019k	+5,63	Ta1-a7 Kh8-g8 Kb2-c3 f7-f6 Kc3-d3 Kg8-h7 Kd3-e4 Kh7-g6 Ta7-c7 f6-f5+ Ke4-f4 Kg6-f6 Tc7-c6+ Kf6-e7 Tc6-g6 Ke7-f8 Kf4xf5 Kf8-e7 Tg6xg7+ Ke7-d6 Kf5-f6
 19/32	00:28,104	 108.409k	3.857k	+M198	Kb2-b3 Kh8-g8 Kb3-c4 Kg8-h7 Kc4-d5 Kh7-h6 Kd5-e5 Kh6-g6 Ta1-g1+ Kg6-h7 Ke5-e4 Kh7-g8 Ke4-d4 Kg8-f8 Tg1-a1 g7-g5 Ta1-a7 f7-f5 Ta7-a5 f5-f4
 20/34	00:28,339	 109.096k	3.850k	+M206	Kb2-b3 Kh8-g8 Kb3-c4 Kg8-h7 Kc4-d5 Kh7-h6 Kd5-e5 Kh6-g6 Ta1-g1+ Kg6-h7 Ke5-d4 Kh7-g8 Tg1-a1 Kg8-f8 Ta1-a7 g7-g6 Ta7-a8+ Kf8-e7 Ta8-a6 f7-f6
 21/34	00:31,785	 118.812k	3.738k	+M209	Kb2-b3 Kh8-g8 Kb3-c4 Kg8-h7 Kc4-d5 Kh7-h6 Kd5-e5 Kh6-g6 Ta1-g1+ Kg6-h7 Ke5-d4 f7-f6 Tg1-h1+ Kh7-g8 Kd4-d5 g7-g6 Th1-b1 f6-f5 Tb1-a1 Kg8-f7 Ta1-a7+ Kf7-f8
 22/34	00:36,699	 133.302k	3.632k	+M280	Kb2-b3 Kh8-g8 Kb3-c4 g7-g5 Ta1-f1 f7-f6 Tf1xf6 Kg8-g7 Tf6-f3 g5-g4 Tf3-f1 g4-g3 Tf1-f3 Kg7-g6 Tf3xg3+ Kg6-f5 Kc4-d4 Kf5-e6 Tg3-g2 Ke6-e7 Kd4-c5 Ke7-f6
 23/34	00:54,744	 186.138k	3.400k	+151,90	Kb2-b3 Kh8-g8 Kb3-c4 g7-g5 Kc4-d5 Kg8-f8 Kd5-e4 Kf8-e7 Ta1-a6 g5-g4 Ke4-f4 g4-g3 Kf4xg3 f7-f6 Kg3-f4 Ke7-f7 Kf4-g4 Kf7-g6 Ta6-e6 Kg6-f7 Te6-d6 Kf7-e7
 24/35	01:26,498	 284.370k	3.288k	+6,10	Kb2-b3 Kh8-g8 Kb3-c3 Kg8-h7 Kc3-d4 g7-g5 Ta1-a6 Kh7-g7 Kd4-e5 Kg7-f8 Ta6-a4 f7-f5 Ta4-a8+ Kf8-f7 Ke5xf5 Kf7-e7 Ta8-a6 Ke7-d7 Ta6-f6 Kd7-c7 Kf5xg5 Kc7-b8 Tf6-f7 Kb8-a8
 24/35	01:35,189	 311.489k	3.272k	+151,86	Kb2-c3 Kh8-h7 Kc3-d4 Kh7-g6 Kd4-e5 f7-f6+ Ke5-e4 f6-f5+ Ke4-e5 Kg6-f7 Ta1-a6 f5-f4 Ke5xf4 Kf7-e7 Kf4-e4 g7-g5 Ke4-e5 Ke7-d7 Ta6-d6+ Kd7-e7 Td6-h6 Ke7-f7 Th6-f6+ Kf7-e7 Ke5-f5 g5-g4 Tf6-a6 g4-g3
 25/37	02:43,830	 519.198k	3.169k	+151,86	Kb2-c3 Kh8-h7 Kc3-d4 Kh7-g6 Kd4-e5 f7-f6+ Ke5-e4 f6-f5+ Ke4-e5 f5-f4 Ke5xf4 Kg6-f7 Kf4-e5 g7-g5 Ta1-f1+ Kf7-e7 Tf1-g1 Ke7-d7 Tg1xg5 Kd7-c7 Ke5-d5 Kc7-b7 Tg5-g7+ Kb7-b8 Kd5-d4 Kb8-c8
 26/37	07:15,264	 1.357.337k	3.118k	+151,86	Kb2-c3 Kh8-h7 Ta1-a6 g7-g5 Kc3-d4 g5-g4 Kd4-e4 g4-g3 Ta6-a3 Kh7-g6 Ta3xg3+ Kg6-f6 Ke4-d5 Kf6-e7 Tg3-d3 f7-f6 Kd5-e4 f6-f5+ Ke4-e5 f5-f4 Ke5xf4 Ke7-f6 Td3-d6+ Kf6-f7 Td6-d1 Kf7-g6 Kf4-g4 Kg6-g7
 27/40	16:12,344	 3.074.073k	3.162k	+151,86	Kb2-c3 Kh8-h7 Ta1-a6 g7-g5 Kc3-d4 Kh7-g7 Kd4-e5 Kg7-f8 Ta6-a8+ Kf8-g7 Ke5-f5 g5-g4 Kf5xg4 Kg7-f6 Ta8-e8 Kf6-g6 Te8-e1 Kg6-f6 Kg4-f4 Kf6-g6 Te1-g1+ Kg6-f6 Tg1-d1 Kf6-e6 Kf4-e4 Ke6-f6 Td1-d3 Kf6-e6

[Gabor Szots]

In the game below Colossus resigns in a drawn position, using 6-man syzygy tablebases.

[Event "Colossus 2025a 64-bit - Feb 22"]
[Site "Budapest"]
[Date "2025.02.25"]
[Round "32"]
[White "Yakka 1.2 64-bit"]
[Black "Colossus 2025a 64-bit"]
[Result "1-0"]
[ECO "A36"]
[Opening "English"]
[Time "12:43:53"]
[Variation "Symmetrical, 5.a3"]
[TimeControl "100+1"]
[PlyCount "214"]

1. c4 c5 2. g3 g6 3. Bg2 Bg7 4. Nc3 Nc6 5. a3 Nf6 6. Rb1 a5 7. b3 {-0.07/13
2} O-O {+0.03/20 18} 8. Bb2 {-0.07/14 2} e5 {+0.01/19 6} 9. e3 {+0.02/15 3}
d6 {+0.05/17 3} 10. Nge2 {+0.23/15 2} Bf5 {-0.01/17 4} 11. d3 {+0.32/14 2}
Qd7 {-0.03/18 3} 12. h3 {+0.25/14 3} Rad8 {-0.04/17 3} 13. g4 {+0.36/14 3}
Be6 {-0.01/17 2} 14. Ng3 {+0.26/13 3} h6 {0.00/18 3} 15. Nd5 {+0.62/15 4}
Ne7 {0.00/18 4} 16. e4 {+0.55/15 3} Nc6 {-0.08/19 7} 17. O-O {+0.36/14 2}
Nd4 {-0.07/18 2} 18. b4 {+0.34/15 3} axb4 {-0.03/15 2} 19. axb4 {+0.09/16
3} Rfe8 {-0.08/16 3} 20. bxc5 {+0.74/15 3} Bxd5 {-0.28/17 3} 21. exd5
{+1.32/17 2} dxc5 {-0.55/19 3} 22. Bxd4 {+0.96/17 4} exd4 {-0.58/19 3} 23.
Qb3 {+0.96/16 4} Rb8 {-0.58/19 2} 24. Qb6 {+0.86/16 2} Qc8 {-0.56/20 3} 25.
Ne4 {+1.18/15 2} Nxe4 {-0.59/17 2} 26. Bxe4 {+0.90/16 3} h5 {-0.11/18 1}
27. Bf3 {+1.04/15 2} Rd8 {-0.27/17 4} 28. Rfe1 {+1.52/15 3} Bf8 {-0.26/16
1} 29. Qf6 {+1.36/16 3} hxg4 {-0.29/17 0} 30. hxg4 {+1.73/15 2} Qd7
{-0.25/18 1} 31. Kg2 {+1.72/15 2} Re8 {-0.14/17 1} 32. Rxe8 {+1.35/16 4}
Rxe8 {-0.12/18 3} 33. Rb6 {+1.70/17 2} Qc7 {-0.01/20 1} 34. d6 {+2.23/19 5}
Qxb6 {-0.32/19 0} 35. Bd5 {+2.51/20 3} Re6 {-0.14/21 1} 36. Bxe6 {+0.83/21
3} fxe6 {-0.09/20 1} 37. Qxg6+ {+0.56/21 3} Kh8 {-0.09/20 1} 38. Qf6+
{+0.87/21 3} Kg8 {0.00/20 2} 39. Qxe6+ {+0.87/21 3} Kg7 {0.00/20 3} 40.
Qe5+ {+0.21/20 3} Kf7 {0.00/20 1} 41. Qf5+ {+0.48/20 3} Kg7 {+0.02/17 0}
42. d7 {+0.29/19 2} Be7 {+0.02/17 1} 43. g5 {+0.97/18 2} Bd8 {+0.02/17 1}
44. f4 {+0.59/17 3} Qd6 {+0.02/18 1} 45. Kf1 {+0.41/17 2} Bc7 {+0.02/17 1}
46. Qg4 {+0.39/16 1} Qb6 {+0.02/16 1} 47. Kf2 {+0.33/16 1} Qb2+ {+0.02/17
1} 48. Kg3 {0.00/17 1} Qd2 {+0.02/17 0} 49. Qf3 {0.00/19 2} Qe1+ {+0.45/14
0} 50. Kg4 {0.00/20 1} Qe6+ {+0.45/17 1} 51. f5 {0.00/20 1} Qxd7 {+0.62/15
1} 52. Qxb7 {0.00/22 1} Qd6 {+0.02/18 1} 53. f6+ {0.00/22 1} Kf7 {+0.02/19
1} 54. Qf3 {0.00/22 1} Qe5 {+0.02/18 2} 55. Kh5 {0.00/23 1} Qh2+ {+0.01/18
2} 56. Kg4 {0.00/23 1} Qd6 {+0.02/19 1} 57. Kh5 {0.00/23 1} Qe5 {+0.02/18
0} 58. Kh4 {0.00/22 1} Bb8 {+0.02/17 1} 59. Qh5+ {0.00/21 1} Ke6 {0.00/16
0} 60. Qg4+ {0.00/19 1} Qf5 {+0.02/19 1} 61. Qg2 {0.00/18 1} Bd6 {+0.02/18
1} 62. Qa8 {0.00/20 1} Qxd3 {+0.02/16 1} 63. Qd5+ {0.00/19 1} Kd7 {+0.02/4
0} 64. Qb7+ {0.00/19 1} Bc7 {+0.02/16 0} 65. Qd5+ {0.00/18 0} Kc8 {0.00/17
1} 66. f7 {+0.23/16 1} Qg3+ {+0.02/15 0} 67. Kh5 {0.00/17 1} Bd6 {+0.02/16
1} 68. Qc6+ {0.00/17 1} Kd8 {+0.02/16 0} 69. Qa8+ {0.00/17 1} Kd7 {+0.02/16
1} 70. Qb7+ {0.00/17 1} Kd8 {+0.02/18 2} 71. Qa8+ {0.00/19 1} Kc7 {+0.02/18
2} 72. Qa7+ {0.00/17 1} Kc8 {0.00/18 0} 73. Qa8+ {0.00/20 1} Kc7 {+0.02/100
0} 74. Qa7+ {0.00/19 1} Kc6 {+0.02/18 2} 75. Kg6 {0.00/17 0} Qd3+ {+0.60/12
0} 76. Kf6 {0.00/16 0} Qf1+ {+0.84/14 0} 77. Kg6 {0.00/16 1} Qxc4 {+1.15/15
1} 78. Qa8+ {-0.20/16 1} Kb5 {+1.37/15 0} 79. Qb7+ {-0.75/15 1} Ka5
{+1.37/16 1} 80. Qd7 {-0.01/14 0} Qd3+ {+1.32/16 1} 81. Kf6 {-0.07/14 0}
Qg3 {+1.51/12 0} 82. Qa7+ {-0.07/16 1} Kb4 {+0.98/18 0} 83. Qb6+ {-0.07/16
0} Ka3 {+0.77/18 1} 84. Qa5+ {-0.07/16 0} Kb3 {+0.89/16 0} 85. Qb5+
{-0.07/16 0} Kc2 {+0.02/19 0} 86. Qa4+ {-0.23/16 1} Kb1 {+0.02/19 0} 87.
Qb5+ {-0.13/15 0} Ka1 {+0.02/19 0} 88. Qf1+ {-0.07/16 1} Ka2 {+0.02/18 0}
89. Qa6+ {-0.07/15 0} Kb2 {+0.02/20 0} 90. Qb5+ {-0.23/16 0} Kc1 {+0.02/19
0} 91. Qc4+ {-0.41/16 1} Qc3 {+0.02/17 1} 92. Qf1+ {0.00/15 0} Kd2
{+0.02/17 0} 93. Qf2+ {0.00/17 1} Kd1 {+0.02/17 0} 94. Qg1+ {0.00/17 1} Kc2
{+0.62/15 0} 95. Qg2+ {0.00/16 1} Kb1 {+0.02/17 0} 96. Qf1+ {0.00/16 1} Kb2
{+0.02/17 1} 97. Qe2+ {0.00/16 0} Qc2 {+0.02/17 0} 98. Qb5+ {0.00/15 0} Qb3
{+0.01/17 1} 99. Qxb3+ {+4.85/16 0} Kxb3 {+0.01/31 0} 100. g6 {+4.85/16 1}
Bf8 {+0.01/33 0} 101. g7 {+7.11/17 1} Bxg7+ {+0.01/75 0} 102. Kxg7
{+7.66/16 1} c4 {-7.82/18 1} 103. f8=Q {+7.31/15 0} c3 {-7.88/18 1} 104.
Qf3 {+7.31/14 0} Kb2 {-7.90/20 1} 105. Qb7+ {+7.75/14 1} Kc1 {-7.93/21 0}
106. Qb4 {+7.29/16 1} c2 {-7.95/22 1} 107. Kh8 {+7.72/16 1} d3 {-7.49/25 0
Black resigns} 1-0

[petero2]

In the game below Colossus resigns in a drawn position, using 6-man syzygy tablebases.

No it does not. The UCI protocol does not support resigning, so it was not the engine that decided to terminate the game. A more accurate statement would be:

Colossus reports a large negative evaluation in a drawn position, which some engine testers have decided should be interpreted as the engine should lose the game, even if the game would have been drawn under standard chess rules.

[Gabor Szots]

In the game below Colossus resigns in a drawn position, using 6-man syzygy tablebases.

No it does not. The UCI protocol does not support resigning, so it was not the engine that decided to terminate the game. A more accurate statement would be:

Colossus reports a large negative evaluation in a drawn position, which some engine testers have decided should be interpreted as the engine should lose the game, even if the game would have been drawn under standard chess rules.

The game was played under Arena and resign threshold was set to -900 cp. I chose that value although some authors had suggested that even -500 was more than enough in most cases. For weaker engines I set that threshold to -1300. Maybe I should not use adjudication at all.

Anyway, the engine computed value in the game is well above -900. Also, if Arena adjudicates a game it displays 'Arena adjudication' not 'XY resigns'. So I don't know what to think and it is conceivable that Colossus did something wrong.

[MartinBryant]

In the game below Colossus resigns in a drawn position, using 6-man syzygy tablebases.

No it does not. The UCI protocol does not support resigning, so it was not the engine that decided to terminate the game. A more accurate statement would be:

Colossus reports a large negative evaluation in a drawn position, which some engine testers have decided should be interpreted as the engine should lose the game, even if the game would have been drawn under standard chess rules.

The game was played under Arena and resign threshold was set to -900 cp. I chose that value although some authors had suggested that even -500 was more than enough in most cases. For weaker engines I set that threshold to -1300. Maybe I should not use adjudication at all.

Anyway, the engine computed value in the game is well above -900. Also, if Arena adjudicates a game it displays 'Arena adjudication' not 'XY resigns'. So I don't know what to think and it is conceivable that Colossus did something wrong.

As Peter states, the UCI protocol simply does not have a command that the engine can send to the UI to declare "I resign", so it certainly seems to be a UI issue.
Colossus' output is certainly what I would expect. Once it reaches a position within the endgame tablebases it simply filters its root movelist to include only those moves which maintain the known result (in this case a draw). And you can see that it makes a number of 'only' moves before the game is terminated.
I rarely use Arena so cannot speak to its output or integrity.
However, I know from watching Graham Banks' broadcasts that he has his UI set to declare a draw immediately if the game reaches a drawn 5-man tablebase position.

Anyway, this issue may have occurred unnoticed multiple times so I think it is prudent for me to modify Colossus' evaluation output such that when it knows it is in a drawn tablebase position it uses a narrower range, perhaps simply dividing the score by 8.