Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Houdini 6.03 - seldom any early move PV's ?
- - By MrKris (*) Date 2018-01-16 03:09
I have been a customer since 2.0.

From TCEC 10, Yorkman's and limited tries with my modest equipment at short TC's it seems to me Houdini 6.03 rarely has any PV plies for early after-book moves.

I mean significantly less than Komodo and Stockfish.

Does this seem correct? It is very tedious to manually compile data from .pgn's to be sure.
AMD Phenom II X6, G/5'
,_ = PV # of plies after played move
Houdini 6.03 - Komodo 11.2.2
(1.e4 e5
2.f4)
     2...exf4 ,2
3.Nf3 ,0
     3...g5 ,2
4.Nc3 ,0
     4...Nc6 ,23
5.h3 ,0
     5...h6 ,2
6.d4 ,0
     6...Bg7 ,2
7.d5 ,0
     7... ,30
8.d6 ,20

Komodo 11.2.2 - Houdini 6.03
(1.e4 e5
2.f4)
     2...exf4 ,0
3.Nf3 ,27
     3...g5 ,0
4.h4 ,27
     4...g4 ,0
5.Ne5 ,2
     5...Nf6 ,0
6.Bc4 ,2
     6...d5 ,0
7.exd5 ,2
     7...Bd6 ,0
8.d4 ,2
     8...Nh5 ,29
Parent - - By Dragon Mist (****) [hr] Date 2018-01-16 18:57
Most strong engines adopt some sort of technique of cutting the search in certain conditions and playing the move instead of calculating the full pv - according to Robert Houdart, this saves time. So not only Houdini, but Komodo and Stockfish and quite some others, too, as observed throughout TCEC S10.
Parent - By MrKris (*) Date 2018-01-17 05:41
Thank you very much for your interesting answer, particularly "- according to Robert Houdart, this saves time."

I have to disagree with "So not only Houdini, but Komodo and Stockfish" "as observed throughout TCEC S10.", personally I am not concerned with the others.

It seems to me Houdini 6.03 uses much more 'time saving', just repeating its played move in its PV's, than its closest competitors in the early after book moves.

For example in Round 4 Houdini gave no clue or idea as to how Komodo might have done better than 19. Nd2 {0.60}.
TCEC 10 Superfinal Round 4
1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 Qc7 5. Nc3 e6 6. Be3 a6 7. Qd2 Nf6 8. O-O-O Bb4 
9. f3 Ne5 10. Nb3 b5 11. Kb1 Be7 12. Qf2 d6 end-of-book 
13. g4 h1 Nfd7 k2 14. Rg1 h0 O-O k40 15. g5 h0 Bb7 k46 16. a3 h0 Rac8 k44 
17. f4 ho Nc4 k2  18. Bxc4 h0 bxc4 k38 19. Nd2 h41 {+0.60} d5 k37 
20. f5 h42 Bc5 k42 21. exd5 Bxe3 22. Qxe3 Bxd5 23. Nde4 Rb8 24. Nxd5 exd5 
25. Nc3 Qb7 26. Qc1 Nc5 27. Rge1 Rfd8 28. Rd4 Qc6 29. Re5 Na4 30. Nxa4 Qxa4 31. Ka1 Qc6 32. c3 g6 
33. Qd2 gxf5 34. Rxf5 Rb5 35. h4 Qe6 36. Rf2 Re8 37. Ka2 Qe3 38. Qxe3 Rxe3 39. a4 Rc5 40. Rf5 Rd3 
41. Rf6 a5 42. Rdf4 Rc7 43. Ra6 Kg7 44. Rxa5 h6 45. gxh6+ Kxh6 46. Rb5 Rd1 47. a5 Kg6 48. Ka3 Kh7 
49. Rd4 Ra1+ 50. Kb4 Rb1 51. Rd2 d4 52. Rg2 Kh6 53. cxd4 c3 54. Rc2 Ra7 55. Kxc3 Rh1 56. h5 Ra8 
57. b4 Rf1 58. Ra2 Rc1+ 1-0
Houdini 6.03 1-0 Komodo 1970.00
Moves 13 through 18 #plies after move played in PV:
Houdini total 1 ply, 0.20 plies average
Komodo total 172 plies, 34.4 plies average
Parent - - By MrKris (*) Date 2018-01-18 05:38
I might pass on Houdini 7 (after buying 2 through 6) if it continues to be a moves-only utility to compete with other authors/contributors with.
Sad, because it has added polyglot, its new variable 0-to-all threads Tactical Mode looks interesting, and it increased its Standard to 8 threads/cores.
Yorkman's Live Chess Engine Tournament - Season 1 Final - Round 6.2
Stockfish 020118 64 BMI2 1-0 Houdini 6.03 - PV plies after played move
1. d4 Nf6 2. c4 e5 3. dxe5 Ng4 4. Nf3 Bc5 5. e3 Nc6 6. Be2 Ngxe5 7. Nxe5
Nxe5 8. f4 Nc6 9. a3 d6 10. O-O Bf5 11. Nc3 a5 12. Bf3 O-O end-of-book
13.Ng5 Sf=2 Ne5 H=1  14.f4 Sf-0- Nf7 H=1  15.Nxf7 Sf=2 Bd4+ H*0*  16.Kh2 Sf=2 Rxf7 H*0*
17.e5 Sf=2 Nc7 H*0*  18.Ne4 Sf=2 Nd5 H=1  19.a4 Sf=2 a5 H*0*  20.Re1 Sf=2 Qc7 H*0*
21.Qb3 Sf=2 b6 H*0* 22.Bd2 Sf=2 Re7 H*0*  23.Rad1 Sf-0- Bb7 H=2  24.c3 Sf=2 H*0*
25.Nf6+ Sf=2 Kh8 H=2  26.Bxd4 Sf=1 {1.07} cxd4 H*0*
14 moves 13 through 26 #plies after move played in PV:
Stockfish: 2/14 -0- 14% of moves, total 23, ave. 1.64/move
Houdini: 9/14 *0* 64% of moves, total 7, ave. 0.50/move

Yorkman's Live Chess Engine Tournament - Season 1 Final - Round 6.2
Houdini 6.03 1-0 Stockfish 020118 64 BMI2 - PV plies after played move
1.d4 Nf6 2.c4 e5 3.dxe5 Ng4 end-of-book  
4.Nf3 H*0* Bc5 Sf=2  5.e3 H*0* Nc6 Sf-0-  6.Be2 H*0* Ngxe5 Sf=1
7.Nxe5 H=1 Nxe5 Sf=2  8.f4 H=2 Nc6 Sf-0-  9.a3 H=1 d6 Sf=2  10.O-O H*0* Bf5 Sf=2
11.Nc3 H*0* a5 Sf=2  12.Bf3 H*0* O-O Sf=2  13.Kh1 H*0* Re8 Sf=1  14.Re1 H*0* Bd7 Sf=2
15.b3 H*0* Rc8 Sf-0-  16.Ra2 *0* Ba7 Sf-0-  17.Rae2 *0* {1.18} Qh4 Sf-0- 
14 moves 4 through 17 #plies after move played in PV:
Houdini: 11/14 *0* 79%, total 4, ave. 0.29/move
Stockfish: 5/14 -0- 36%, total 16, ave. 1.14/move

TCEC 10 Superfinal Round 6
Houdini 6.03 1-0 Komodo 1970.00 - PV plies after played move
1. d4 Nf6 2. c4 g6 3. Nf3 Bg7 4. g3 O-O 5. Bg2 d6 6. O-O Nbd7 
7. Nc3 e5 8. e4 c6 9. h3 Qb6 10. c5 dxc5 11. dxe5 Ne8 12. e6 fxe6 end-of-book
13.Ng5 H*0* Ne5 K=33  14.f4 H*0* Nf7 K=2  15.Nxf7 H=38 Bd4 K=2  16.Kh2 H*0* Rxf7 K=2
17.e5 H=32 Ng7 K=2  18.g4 H*0* Qc7 K=2  19.Ne4 H*0* b6 K=2  20.Ng5 H*0* Rf8 K=37
21.Qd3 H*0* Qe7 K=2  22.a4 H*0* Bb7 K=2  23.a5 H=42 Rad8 K=2  24.Qd3 H*0* c4 K=35
25.axb6 H*0* Bxb6 K=2  26.Ra3 H*0* K=38  27.Qc3 H*0* {1.12} Rd4 K=2
15 moves 13 through 27 #plies after move played in PV:
Houdini 12/15 *0* 80%, total 112 in the only 3 moves >0
Komodo 0/15 none 0%, total 183 in all 15 moves >0
Parent - By MrKris (*) Date 2018-02-08 08:17
YLCET, Rd. 35.2

HOUDINI 6.03
PV for 11. Bh3 : 1st after book move, 30 plies, EVAL. 0.80
@ 11. Bh3
PV for 12. f4
12. f4 @
PV for 13. e6
13. e6 @
PV for 14. Qc2
14. Qc2 @
PV for 15. f5
15. f5 @
PV for 16. Ne3
16. Ne3 @
PV for 17. O-O
@ 17. O-O
PV for 18. Na2
18. Na2 @
PV for 20. Qb2
@ 20. Qb2
PV for 20. Qb2
@ 20. Qb2
PV for 21. Rxa2
21. Rxa2 @
PV for 22. Qxa2 EVALUATION 0.90
22. Qxa2 Bb4 23. Bg2 Rd8 24. g4 Be8 25. Nc2 Bd2 26. Ba3 b5 27. Kh1 Bc3 28. Bf3 @ 28... Be5 29. Nb4 Qc7 30. Qc2 Bd6 31. Na2 Nb7 32. Bxd6 Qxd6 33. Nc3 Qd2 34. Qxd2

STOCKFISH 9
PV for 11... Qe7 : 1st after book move, 37 plies, EVAL. 0.69
@ 11... Qe7 12. f4 f6 13. e6 Rd8 14. Qc2 Nc5 15. f5 Rd4 16. Ne3 Ba5 17. O-O Nba6 18. Na2 Rxa4 19. b3 Nb4 20. Qb1 Rxa2 21. Rxa2 Nxa2 22. Qxa2 Bb6 23. Kh1 Ne4
PV for 12... f6
@ 12... f6 13. e6
PV for 13... Rd8
@ 13... Rd8 14. Qc2 Nc5 15. f5 Rd4 16. Ne3 Nba6 17. O-O Ba5 18. Na2 Rxa4 19. b3 Nb4 20. Qb1 Nxa2 21. Rxa2 Rxa2 22. Qxa2 Bb4 23. Bg2 Rd8 24. g4 Be8 25. Nc2 Bd2
PV for 14... Nc5
@ 14... Nc5 15. f5 Rd4 16. Ne3 Nba6 17. O-O Ba5 18. Na2 Rxa4 19. b3 Nb4 20. Qb2 Rxa2 21. Rxa2 Nxa2 22. Qxa2 Bb4 23. Bg2 Rd8 24. g4 Be8 25. Nc2 Bd2 26. Qxa7 Bxc1
PV for 15... Rd4
@ 15... Rd4 16. Ne3 Ba5 17. Ra3 Nba6 18. O-O Bb4 19. Ra1 Ba5 20. Na2 Rxa4 21. b3 Nb4 22. Qb2 Rxa2 23. Rxa2 Nxa2 24. Qxa2 Bb4 25. Bg2 Rd8 26. g4 Be8 27. Nc2 Bd2
PV for 16... Nba6
@ 16... Nba6
PV for 17... Ba5
@ 17... Ba5 18. Na2
PV for 18... Rxa4
@ 18... Rxa4 19. b3 Nb4 20. Qb1 Rxa2 21. Rxa2 Nxa2 22. Qxa2 Bb4 23. Bg2 h6 24. g4 Be8 25. Nc2 Bc3 26. Be3 Na6 27. b4 Bxb4 28. Nxb4 Nxb4 29. Qa5 Nd5 30. Bxd5 cxd5
PV for 19... Nb4
@ 19... Nb4 20. Qb1 Rxa2 21. Rxa2 Nxa2 22. Qxa2 Bb4 23. Bg2 Rd8 24. g4 Be8 25. Nc2 Bd2 26. Ba3 b5 27. Kh1 Bc3 28. Bf3 Rd2 29. Bc1 Rd8 30. Be3 a5 31. Rc1 Nb7
PV for 20... Rxa2
@ 20... Rxa2 21. Rxa2 Nxa2 22. Qxa2 Bb4 23. Bg2 Rd8 24. g4 Be8 25. Nc2 Bd2 26. b4 Na6 27. b5 cxb5 28. Ne3 Nb4 29. Qxa7 Bc3 30. Qxb7 Qxb7 31. Bxb7 Bc6 32. Bxc6 Nxc6
PV for 21... Nxa2
@ 21... Nxa2 22. Qxa2 Bb4 23. Bg2 Rd8 24. g4 Be8 25. Nc2 Bd2 26. Ba3 b5 27. Kh1 Bc3 28. Bf3 Rd2 29. Bc1 Rd8 30. Be3 a5 31. Rc1 Nb7 32. Ne1 Bd4 33. Bf4 Bb6
PV for 22... Bb4 EVALUATION 0.60
22... Bb4 23. Bg2 @ 23... h6 24. Kh1 a5 25. g4 Be8 26. Nc2 Na6 27. Nxb4 Nxb4 28. Qa3 c5 29. Be3 Kh7 30. Bf2 Bb5 31. Bf3 Qf8 32. Rd1 b6 33. h3 Rd8 34. Qc1 Nc6
Parent - - By MrKris (*) Date 2018-02-09 01:42
YLCET  http://chess.my2dollars.ca/live.html
Round 40.2: Book 1. d4 d5 2. c4 Nc6 3. Nf3 Bg4 4. cxd5 Bxf3 5. gxf3 Qxd5 6. e3 e5 7. Nc3 Bb4 8. Bd2 Bxc3 9. bxc3 Qd6

HOUDINI 6.03
PV for 10... Nge7 @ 10... Nge7  <0> just its played move
PV for 11... O-O 11... O-O@  <0>
PV for 12... Rab8 @ 12... Rab8  <0>
PV for 13... h6 @ 13... h6  <0>
PV for 14... Rfd8 14... Rfd8 @  <0>
PV for 15... Na5 15... Na5 @  <0>
PV for 16... Nb7 @ 16... Nb7  <0>
PV for 17... exd4 @ 17... exd4  <0>
PV for 18... Qf6 18... Qf6 19. Ke2@ <1
PV for 19... c5 19... c5 20. Bc3 Nd5 21. Ba1 Qe6 22. Be4 @ 22... cxd4 23. Rxd4 Nc5 24. Bxd5 Rxd5 25. Qxa7 Rbd8 26. Rc1 Rxd4 27. Bxd4 Rxd4 28. Rxc5 Rb4 29. Qa3 Rb1 30. Qa8 Kh7 31. Qe4 Qxe4

KOMODO 2005.00
PV for 10. Qb3 @ 10. Qb3 Nf6 11. Bc1 O-O-O 12. Rb1 b6 13. Ba3 Qd5 14. Ba6 Kb8 15. Qxd5 Nxd5 16. Kd2 Rhe8 17. Bb5 exd4 18. cxd4 Re6 19. h4 h5 20. Bc4 Rf6 21. Rb5 Nce7 22. f4
PV for 11. Qxb7 @ 11. Qxb7 O-O 12. Rd1 Rab8 13. Qa6 Nd5 14. Be2 Qg6 15. Qc4 Rfd8 16. h4 Rb2 17. h5 Qd6 18. Qd3 h6 19. Qf5 exd4 20. cxd4 Nf6 21. Rg1 Qh2 22. Rg3 Nxh5 23. Rh3
PV for 12. Qa6 @ 12. Qa6 Rfb8 13. Rd1 Rb6 14. Qc4 Rab8 15. Bd3 exd4 16. cxd4 Nb4 17. Bxb4 Rxb4 18. Qc5 Rb2 19. O-O Qxc5 20. dxc5 Rxa2 21. Rb1 Rxb1 22. Rxb1 g6 23. Rb7 Rd2 24. Rxc7
PV for 13. Rd1 @ 13. Rd1 Rb6 14. Qc4 Rfb8 15. Bd3 exd4 16. cxd4 Nb4 17. Bxb4 Rxb4 18. Qc5 Rb2 19. O-O Rxa2 20. Rb1 Rxb1 21. Rxb1 g6 22. Rb8 Kg7 23. Rb7 Qxc5 24. dxc5 Ra5 25. Rxc7
PV for 14. Qc4 @ 14. Qc4 Rfd8 15. h4 <<2
PV for 15. Bd3 @ 15. Bd3 Na5 16. Qa4 <<2
PV for 16. Qa4 @ 16. Qa4 Nac6 17. Rg1 <<2
PV for 17. Bc2 @ 17. Bc2 Qf6 18. Ke2 <<2
PV for 18. cxd4 @ 18. cxd4 Qf6 19. Ke2 c5 20. Bc3 Nd5 21. Ba1 Qe6 22. Be4 Rbc8 23. Rc1 c4 24. Qb5 Rc7 25. Rhg1 f5 26. Bxd5 Rxd5 27. Qb1 Qe8 28. h4 Nd6 29. Qc2 Nb5 30. Rgd1
PV for 19. Ke2 19. Ke2 @ 19... c5 20. Bc3 Nd5 21. Ba1 Qe6 22. Be4 Rbc8 23. Rc1 c4 24. Qb5 Rc7 25. Rhg1 f5 26. Bxd5 Rxd5 27. Qb1 Qe8 28. Bc3 Nd6 29. Qb2 Nb5 30. Kf1 Nxc3 31. Qxc3
Parent - - By Peter Grayson (****) [gb] Date 2018-02-09 14:11
Houdini 6.03 gives a full depth output on completion of its multiple iteration analysis of the position at the same ply depth. In the game it suggests at the time Houdini 6.03 moved the engine had not completed its full analysis for the ply depth. In infinite analysis mode it gives a full continuation of the expected moves at the ply depth before incrementing to the next ply depth.

Unclear why you would expect an engine to give a full continuation if its analysis has not completed? Time constraints appear to be the governing influence here and of course the full continuation of the 27 ply depth output are not included in the game record.

80: Komodo 2005.00 64-bit - Houdini 6.03 Pro x64-pext, YLCET Super X1 2018
r3k1nr/ppp2ppp/2nq4/4p3/3P4/1QP1PP2/P2B1P1P/R3KB1R b KQkq - 0 1


Analysis by Houdini 6.03 Pro x64-popc:

10...Nge7 11.Bc1 Rd8 12.Ba3 Qf6 13.Bxe7 Nxe7 14.Qxb7 c6 15.Be2 Qg6
  +/=  (0.32)   Depth: 11/24   00:00:00  4616kN
10...Nge7 11.Bc1 exd4 12.cxd4 0-0 13.Bd3 Rfe8 14.Ba3 Qf6 15.Be4 Qg5 16.0-0-0
  +/=  (0.33)   Depth: 12/29   00:00:00  9100kN
10...Nge7 11.Bc1 Qf6 12.Rg1 0-0 13.Bd3 Qxf3 14.h4 exd4 15.cxd4 g6 16.Ba3 Rae8
  =  (0.13)   Depth: 13/29   00:00:00  13190kN
10...Nge7 11.Bc1 Rd8 12.Ba3 Qf6 13.Qxb7 0-0 14.Bg2 Qg6 15.0-0 Rfe8 16.Kh1 Rb8 17.Qxc7
  +/=  (0.51)   Depth: 14/32   00:00:01  28601kN
10...Nf6 11.Bc1 Qd5 12.Rg1 0-0 13.Qxd5 Nxd5 14.Kd2 g6 15.Rb1 Nb6 16.Ba3 Rfd8 17.Bd3 Rab8
  +/=  (0.46)   Depth: 15/32   00:00:01  41367kN
10...Nge7 11.Bc1 exd4 12.cxd4 a5 13.Ba3 Nb4 14.Bd3 0-0 15.Rb1 Ned5 16.0-0 c6 17.f4 Rfe8 18.Kh1 g6
  =  (0.13)   Depth: 16/32   00:00:02  63049kN
10...Nge7 11.Bc1 exd4 12.cxd4 a5 13.Bd2 0-0 14.Rc1 a4 15.Qb1 f5 16.Bd3 Rab8 17.Qb5 Kh8 18.Rc5 Rfe8 19.Qxa4 Ng6
  =  (0.23)   Depth: 17/35   00:00:03  85330kN
10...Nge7 11.Bc1 exd4 12.cxd4 a5 13.Bd2 0-0 14.Rc1 Rfe8 15.Bd3 Nxd4 16.exd4 Nf5+ 17.Kf1 Nxd4 18.Qd1 Rad8 19.Bg5
  =  (0.23)   Depth: 18/36   00:00:03  98781kN
10...Nge7 11.Bc1
  +/=  (0.33 --)   Depth: 19/46   00:00:04  122MN
10...Nge7
  =  (0.24 ++)   Depth: 19/46   00:00:04  128MN
10...Nge7 11.Bc1
  +/=  (0.36 --)   Depth: 19/46   00:00:04  131MN
10...Nge7 11.Bc1 exd4 12.cxd4 a5 13.Bd2 0-0 14.Rc1 a4 15.Qb1 Rab8 16.Bd3 Kh8 17.f4 f5 18.Qb5 Nd5 19.Bc4 Nce7 20.Qxa4 Qg6 21.Rf1 c6
  =  (0.28)   Depth: 19/46   00:00:05  135MN
10...Nge7 11.Bc1 exd4 12.cxd4 a5 13.Bd2 0-0 14.Rc1 a4 15.Qb1 Rab8 16.Bd3 Kh8 17.f4 f5 18.Qb5 Nd5 19.Bc4 Nce7 20.Qxa4 Qg6 21.Rf1 c6
  =  (0.28)   Depth: 19/46   00:00:05  136MN
10...Nge7 11.Bc1
  +/=  (0.35 --)   Depth: 20/47   00:00:05  146MN
10...Nge7
  =  (0.28 ++)   Depth: 20/47   00:00:05  155MN
10...Nge7 11.Bc1 exd4 12.cxd4 0-0 13.Ba3 Qf6 14.Bg2 Nf5 15.Bxf8 Nfxd4 16.exd4 Nxd4 17.Qxb7 Nxf3+ 18.Bxf3 Qxa1+ 19.Ke2 Re8+ 20.Be7 Rxe7+ 21.Kd3 Rd7+ 22.Bd5 Qf6 23.Kc4 Qh4+ 24.Kc3 Qf6+ 25.Kb3 Rd8
  =  (0.25)   Depth: 20/47   00:00:05  157MN
10...Nge7 11.Bc1 exd4 12.cxd4 0-0 13.Ba3 Qf6 14.Bg2 Nf5 15.Bxf8 Nfxd4 16.exd4 Nxd4 17.Qxb7 Nxf3+ 18.Bxf3 Qxa1+ 19.Ke2 Re8+ 20.Be7 Rxe7+ 21.Kd3 Rd7+ 22.Bd5 Qf6 23.Kc4 Qh4+ 24.Kc3 Qf6+ 25.Kb3 Rd8
  =  (0.25)   Depth: 20/47   00:00:05  157MN
10...Nge7 11.Bc1
  +/=  (0.32 --)   Depth: 21/47   00:00:05  161MN
10...Nge7 11.Bc1 exd4 12.cxd4 a5 13.Bd2 0-0 14.Rc1 a4 15.Qb1 Rab8 16.Bd3 Kh8 17.f4 f5 18.Be2 Nd5 19.0-0 Nxf4 20.exf4 Nxd4
  +/=  (0.37)   Depth: 21/47   00:00:06  180MN
10...Nge7 11.Bc1 exd4 12.cxd4 a5 13.Bd2 0-0 14.Rc1 a4 15.Qb1 Rab8 16.Bd3 Kh8 17.f4 f5 18.Be2 Nd5 19.0-0 Nxf4 20.exf4 Nxd4
  +/=  (0.37)   Depth: 21/47   00:00:06  183MN
10...Nge7
  =  (0.30 ++)   Depth: 22/47   00:00:07  197MN
10...Nge7 11.Bc1
  +/=  (0.37 --)   Depth: 22/47   00:00:07  218MN
10...Nge7
  =  (0.30 ++)   Depth: 22/47   00:00:09  266MN
10...Nge7 11.Bc1
  +/=  (0.39 --)   Depth: 22/47   00:00:13  378MN
10...Nge7 11.Qxb7
  +/=  (0.59 --)   Depth: 22/47   00:00:14  423MN
10...Nge7 11.Qxb7 0-0 12.Be2 Rfe8 13.Rd1 a5 14.dxe5 Nxe5 15.Qe4 Qf6 16.f4 N5g6 17.Qd4 Qe6 18.Bc1 Qh3 19.Qd7 Qg2 20.Rf1 Nh4 21.Rd3 Rab8 22.Qg4 Nef5 23.Qxg2 Nxg2+
  +/-  (0.78)   Depth: 22/47   00:00:25  726MN
10...Nge7 11.Qxb7 0-0 12.Be2 Rfe8 13.Rd1 a5 14.dxe5 Nxe5 15.Qe4 Qf6 16.f4 N5g6 17.Qd4 Qe6 18.Bc1 Qh3 19.Qd7 Qg2 20.Rf1 Nh4 21.Rd3 Rab8 22.Qg4 Nef5 23.Qxg2 Nxg2+
  +/-  (0.78)   Depth: 22/47   00:00:25  750MN
10...Nge7
  +/-  (0.71 ++)   Depth: 23/47   00:00:27  791MN
10...Nge7
  +/=  (0.64 ++)   Depth: 23/47   00:00:27  793MN
10...Nge7
  +/=  (0.54 ++)   Depth: 23/47   00:00:27  801MN
10...Nge7 11.Qxb7 0-0 12.Be2 a5 13.Qb3 Rab8 14.Qc2 Rfe8 15.h4 h5 16.Rb1 Rxb1+ 17.Qxb1 Qe6 18.Rg1 Rb8 19.Qc2 Qh3 20.f4 exf4 21.exf4 Qh2 22.Rf1 Qxh4
  +/=  (0.59)   Depth: 23/47   00:00:29  864MN
10...Nge7 11.Qxb7 0-0 12.Be2 a5 13.Qb3 Rab8 14.Qc2 Rfe8 15.h4 h5 16.Rb1 Rxb1+ 17.Qxb1 Qe6 18.Rg1 Rb8 19.Qc2 Qh3 20.f4 exf4 21.exf4 Qh2 22.Rf1 Qxh4
  +/=  (0.59)   Depth: 23/47   00:00:29  864MN
10...Nge7
  +/=  (0.52 ++)   Depth: 24/47   00:00:33  977MN
10...Nge7 11.Qxb7
  +/=  (0.59 --)   Depth: 24/47   00:00:34  1002MN
10...Nge7 11.Qxb7
  +/=  (0.69 --)   Depth: 24/47   00:00:35  1035MN
10...Nge7 11.Qxb7 0-0 12.Be2 a5 13.Qb5 Rab8 14.Qd3 Rb2 15.Bc1 Rb6 16.Rb1 exd4 17.Rxb6 cxb6 18.a3 Qc7 19.exd4 Re8 20.0-0 Nd5 21.Qc4 Qd7 22.Bd3 Rc8 23.Qb3 Nxc3 24.Bc4 Nxd4
  +/-  (0.81)   Depth: 24/47   00:00:45  1332MN
10...Nf6
  +/=  (0.60 ++)   Depth: 24/49   00:00:47  1410MN
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Bg2 Rhe8 14.0-0 Qe6 15.Qxe6+ fxe6 16.f4 e4 17.f3 exf3 18.Rxf3 Kd7 19.Rg3 Rg8 20.e4 Ne7 21.Bf3 g6 22.Rh3 h5 23.Rh4 Rg7
  +/=  (0.58)   Depth: 24/49   00:00:52  1532MN
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Bg2 Rhe8 14.0-0 Qe6 15.Qxe6+ fxe6 16.f4 e4 17.f3 exf3 18.Rxf3 Kd7 19.Rg3 Rg8 20.e4 Ne7 21.Bf3 g6 22.Rh3 h5 23.Rh4 Rg7
  +/=  (0.58)   Depth: 24/49   00:00:52  1546MN
10...Nf6 11.Bc1
  +/=  (0.65 --)   Depth: 25/51   00:00:55  1638MN
10...Nf6
  +/=  (0.58 ++)   Depth: 25/55   00:01:01  1819MN
10...Nf6 11.Bc1
  +/=  (0.65 --)   Depth: 25/55   00:01:02  1838MN
10...Nf6
  +/=  (0.56 ++)   Depth: 25/55   00:01:13  2159MN, tb=2
10...Nf6 11.Bc1
  +/=  (0.68 --)   Depth: 25/55   00:01:23  2476MN, tb=39
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Ba3 Qd5 14.Qxd5 Nxd5 15.Kd2 exd4 16.cxd4 Kb7 17.h4 Rhe8 18.h5 Nf6 19.Bd3 a6 20.Rbg1 g6 21.hxg6 fxg6 22.Rc1 Ne7 23.Rb1 Ned5 24.Be2 Ne7
  +/=  (0.58)   Depth: 25/55   00:02:24  4280MN, tb=60
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Ba3 Qd5 14.Qxd5 Nxd5 15.Kd2 exd4 16.cxd4 Kb7 17.h4 Rhe8 18.h5 Nf6 19.Bd3 a6 20.Rbg1 g6 21.hxg6 fxg6 22.Rc1 Ne7 23.Rb1 Ned5 24.Be2 Ne7
  +/=  (0.58)   Depth: 25/55   00:02:25  4313MN, tb=60
10...Nf6 11.Bc1
  +/=  (0.65 --)   Depth: 26/55   00:02:25  4315MN, tb=60
10...Nf6
  +/=  (0.58 ++)   Depth: 26/55   00:02:30  4452MN, tb=60
10...Nf6 11.Bc1
  +/=  (0.65 --)   Depth: 26/55   00:02:30  4456MN, tb=60
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Bg2 Kb8 14.0-0 h5 15.Bb2 Qe6 16.Qxe6 fxe6 17.Rfc1 Rh6 18.c4 exd4 19.f4 Rg6 20.Kf1 Rxg2 21.Kxg2 dxe3 22.Rd1 Rxd1 23.Rxd1 exf2 24.Kxf2 Na5
  +/=  (0.64)   Depth: 26/55   00:02:34  4573MN, tb=71
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Bg2 Kb8 14.0-0 h5 15.Bb2 Qe6 16.Qxe6 fxe6 17.Rfc1 Rh6 18.c4 exd4 19.f4 Rg6 20.Kf1 Rxg2 21.Kxg2 dxe3 22.Rd1 Rxd1 23.Rxd1 exf2 24.Kxf2 Na5
  +/=  (0.64)   Depth: 26/55   00:02:34  4578MN, tb=71
10...Nf6
  +/=  (0.57 ++)   Depth: 27/55   00:02:39  4746MN, tb=71
10...Nf6 11.Bc1
  +/=  (0.64 --)   Depth: 27/55   00:02:39  4749MN, tb=71
10...Nf6
  +/=  (0.57 ++)   Depth: 27/55   00:02:53  5163MN, tb=72
10...Nf6 11.Bc1
  +/=  (0.65 --)   Depth: 27/55   00:02:58  5306MN, tb=72
10...Nf6 11.Bc1 exd4 12.Ba3 Qd5 13.Qxd5 Nxd5 14.cxd4 0-0-0 15.Kd2 Rhe8 16.Rc1 g6 17.h4 f5 18.Bb5 Re6 19.Bxc6 bxc6 20.Bc5 a5 21.Rb1 Ree8 22.h5 g5 23.h6 Kd7 24.Rbg1 Rg8
  +/=  (0.64)   Depth: 27/55   00:03:06  5536MN, tb=75
10...Nf6 11.Bc1 exd4 12.Ba3 Qd5 13.Qxd5 Nxd5 14.cxd4 0-0-0 15.Kd2 Rhe8 16.Rc1 g6 17.h4 f5 18.Bb5 Re6 19.Bxc6 bxc6 20.Bc5 a5 21.Rb1 Ree8 22.h5 g5 23.h6 Kd7 24.Rbg1 Rg8
  +/=  (0.64)   Depth: 27/55   00:03:06  5536MN, tb=75
10...Nf6 11.Bc1
  +/-  (0.71 --)   Depth: 28/55   00:03:22  6038MN, tb=76
10...Nf6
  +/=  (0.64 ++)   Depth: 28/55   00:04:01  7206MN, tb=87
10...Nf6 11.Bc1
  +/-  (0.71 --)   Depth: 28/55   00:04:02  7253MN, tb=87
10...Nf6
  +/=  (0.62 ++)   Depth: 28/55   00:04:07  7395MN, tb=87
10...Nf6 11.Bc1
  +/-  (0.74 --)   Depth: 28/55   00:04:08  7431MN, tb=87
10...Nf6
  +/=  (0.57 ++)   Depth: 28/56   00:04:50  8668MN, tb=145
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Bg2 Kb8 14.0-0 Rhe8 15.Ba3 Qd7 16.Rbd1 Qe6 17.Qxe6 fxe6 18.f4 e4 19.c4 Ne7 20.Bxe7 Rxe7 21.f3 exf3 22.Bxf3 g6 23.Rb1 Kc8 24.Rfc1 c5 25.dxc5 bxc5 26.Rc2 Rd6 27.Rb5 Rc7 28.Kf2 Rb6
  +/=  (0.60)   Depth: 28/56   00:04:54  8796MN, tb=145
10...Nf6 11.Bc1 0-0-0 12.Rb1 b6 13.Bg2 Kb8 14.0-0 Rhe8 15.Ba3 Qd7 16.Rbd1 Qe6 17.Qxe6 fxe6 18.f4 e4 19.c4 Ne7 20.Bxe7 Rxe7 21.f3 exf3 22.Bxf3 g6 23.Rb1 Kc8 24.Rfc1 c5 25.dxc5 bxc5 26.Rc2 Rd6 27.Rb5 Rc7 28.Kf2 Rb6
  +/=  (0.60)   Depth: 28/56   00:04:54  8804MN, tb=145
10...Nf6 11.Bc1
  +/=  (0.67 --)   Depth: 29/56   00:04:59  8965MN, tb=146
10...Nf6
  +/=  (0.60 ++)   Depth: 29/56   00:05:08  9222MN, tb=146

(Grayson, Newport, South Wales 09.02.2018)
Parent - - By MrKris (*) Date 2018-03-02 04:41
What is unclear is why you do not understand, or wish to deflect from with your programming blah-blah-blah, the issue of usefulness for chess enthusiasts.
Specifically engine output in the .pgn's from TCEC, YLCET, Arena's 'save PV's as variations', GUI auto-analysis etc.

Your solution seems to be for all to get there own 44 or 36 core machine and manually watch and/or save the streaming output, as you provided, per move they might be interested in - rather than have it every move, as in Komodo's apparent minimum 2 plies after its move, in the TCEC/YLCET .pgn, or in their own engine-engine games .pgn or auto-analysis .pgn.

Do you think I am a programmer, or I raised a programming issue?
No, the issue is usefulness vs. 0, null, none, to chess enthusiasts.

It is amazing to me the you (and Dragon Mist above) seem to be denying the blatantly obvious fact -evidenced in the TCEC and YLCET archives- that Houdini 6.03  almost always has for its first 8 or more moves after the book 0 after-move plies in its PV.
I say obvious because at first there where many chatters at TCEC commenting about it, though now it is quite old news for those actually interested in chess itself, engine or human.

Then, essentially only after the game has been decided, Houdini starts with long PV's roughly (though I suspect more 'blanks') as other engines.
As below, when Houdini's evaluation reached 1.32 (although Komodo managed to draw the recent example) it had its first PV with 'ahead' moves (a 1 ply 'slip' on move 13, otherwise 0).
YLCET Super X3 
http://chess.my2dollars.ca/ylcetarchivefix.html?s=4&p=YLCET-SuperX3_e1e_Pgn.pgn&c=YLCET-SuperX3_e1e_TmpTable.html&sc=YLCET-SuperX3_e1e_Schedule.html&gn=44
Game 44 - Rd. 22.2 - Houdini 6.03 1/2-1/2 Komodo 2016
Book: 1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Be2 O-O 6. Nf3 Bg4 7. Be3 Nfd7 8. Qd2 e5 9. d5 a5 10. O-O-O
Website 'Archive' PVs:

  Komodo 10... Na6 11. Kb1 Qe7 12. h3 Bxf3 13. Bxf3 f5 14. Qe2 Bf6 15. Bh6 Bg7 16. Bc1 Qh4 17. g4 f4 18. g5 Qxg5 19. h4 Qh6 20. h5 g5 21. Bg4 Ndc5 22. Rhg1 Kh8 <-many plies after move

Houdini 11. h4  <---- 0 plies after move ----

  Komodo 11... Nac5 12. Kb1 a4 <- 2 plies after move, seems Komodo minimum

Houdini 12. Ng1 <---- 0 plies after move ----

  Komodo 12... Bxe2 13. Ngxe2 h5 14. Qc2 f5 15. exf5 gxf5 16. Kb1 f4 17. Bxc5 Nxc5 18. Nc1 a4 19. Nd3 Nxd3 20. Qxd3 a3 21. Ne4 Qc8 22. b3 Qf5 23. c5 dxc5 24. Nxc5 Qxd3 <-many plies after move

Houdini 13. Ngxe2 h5 <-- 1 ply after move

  Komodo 13... Nb6 14. Bxc5 dxc5 <- 2 plies after move

Houdini 14. Bxc5  <---- 0 plies after move ----

  Komodo 14... dxc5 15. Qd3 Nc8 <- 2 plies after move

Houdini 15. Qd3 <---- 0 plies after move ----

  Komodo 15... Qf6 16. g4 a4 <- 2 plies after move

Houdini 16. g4 <---- 0 plies after move ----

  Komodo 16... Nc8 17. Kb1 Qe7 18. Rdg1 Nd6 19. Nb5 a4 20. Nec3 h6 21. h5 g5 22. Rg3 Rfd8 23. a3 Ra6 24. Nxd6 cxd6 25. Qc2 Qd7 26. Rhg1 Rf8 27. Rf3 Re8 28. Rf5 Raa8 <- many plies after move

Houdini 17. a4 <---- 0 plies after move ----

  Komodo 17... Nd6 18. Kb1 Ra6 <- 2 plies after move

Houdini 18. f3 Qe7 19. Kb1 Rab8 20. Ka2 Qd7 21. Nb5 Nxb5 22. axb5 Qe7 23. Rdg1 b6 24. g5 f6 25. Nc3 Rbd8 26. Kb1 Qf7 27. Rh3 Kh8 28. Qd2 Rd6 29. Rgh1 Kg8 30. Qe3 <- many plies after move, 1.32 evaluation.
Parent - - By Peter Grayson (****) [gb] Date 2018-03-03 00:33

>What is unclear is why you do not understand, or wish to deflect from with your programming blah-blah-blah, the issue of usefulness for chess enthusiasts.
>Specifically engine output in the .pgn's from TCEC, YLCET, Arena's 'save PV's as variations', GUI auto-analysis etc.


I fully understand the point raised but it is you who do not understand the answer. You criticise the engine for a failing of the GUI to record the full engine output. Whether in game play or analysis, the engine output gives a full line of expected moves on completion of its analysis at the ply depth but may not do so if it moves before completion of its analysis at the ply depth. That surely is not too difficult to understand? If the GUI recorded the engine utput in its entirety it would show something rather different to what you see as a snapshot of the output at the time the move is made. If you have the engine you will see it, if you do not have the engine then your criticism is based on erroneous information because of the limitations of the GUI record.

>Your solution seems to be for all to get there own 44 or 36 core machine and manually watch and/or save the streaming output, as you provided, per move they might be interested in - rather than have it every move, as in Komodo's apparent minimum 2 plies after its move, in the TCEC/YLCET .pgn, or in their own engine-engine games .pgn or auto-analysis .pgn.


Where did I say that? The pasted engine output is exactly what you would see in game or analysis mode and was used as an example to illustrate the actual engine output and not the attenuated limited output presented in the GUI record.

>It is amazing to me the you (and Dragon Mist above) seem to be denying the blatantly obvious fact -evidenced in the TCEC and YLCET archives- that Houdini 6.03  almost always has for its first 8 or more moves after the book 0 after-move plies in its PV.
>I say obvious because at first there where many chatters at TCEC commenting about it, though now it is quite old news for those actually interested in chess itself, engine or human.


What is amazing to me is you have no clue as to what you are talking about. If you had the engine you would clearly see that was the case.

>Then, essentially only after the game has been decided, Houdini starts with long PV's roughly (though I suspect more 'blanks') as other engines.


Houdini 6.03 gives full PV's during and always on the completion of the ply depth search. The attenuated PV is seen when at some point of analysis there may be multiple moves for its opponent and it is unclear which is the best line, thus its analysis is incomplete. I have looked at some of the TCEC and YLCET games and the recorded output for Houdini is not what I see in terms of length of PV. For example, in a game against Stockfish, the GUI records a single move where a book is not used giving 1.d4 at 30 ply depth. Run on my 10 year old Q9550 quad I see something very different ...

Houdini 6.03 x64
1.d4 d5 2.c4 e6 3.Nf3 Nf6 4.Nc3 Bb4 5.Bg5 Nbd7 6.cxd5 exd5 7.e3 c5 8.dxc5 Qa5 9.Nd2 Bxc3 10.bxc3 0-0 11.Be2 h6 12.Bh4 Qxc3 13.0-0 Qb4 14.Bxf6 Nxf6 15.Nb3 Rb8 16.Qd4
  =  (0.26)   Depth: 30/53   00:13:07  3226MN

Your crticism of the engine is unfounded and unjustified.
Parent - - By gsgs (***) [de] Date 2018-03-03 07:58 Edited 2018-03-03 08:03
TCEC goes have the full log files for download, the pvs of the last completed depth should be in it
(I didn't check now, but looked at it earlier and : afair)

It should be easy to change the program that it outputs the last pv (completed last depth
or current pv or both or/and fail high/low pvs)

Houdart said this was the task of the GUI , and yes, the GUI should also be able to
easily do that (presumably only a small change in the code)

it should be better to schedule meetings to watch a replay of the games for the fans with full pvs
and Stockfish and Komodo and Houdini analysis ... but they want it "live"
Parent - - By Peter Grayson (****) [us] Date 2018-03-05 16:02 Upvotes 1
Thank you for information.

I run matches mostly using the Fritz GUI, occasionally the Shredder GUI when the game pgn's only record the move made but the engine output window gives a full record of engine's PV length. A pity this cannot be transmitted during the game because it would avoid these types of criticism when it is not the engine at fault but the interface method of recording and displaying engine output.

Personally, I admire the enthusiasm of those who provide their time and resource to make these games and matches publicly available but as usual there are always people who find something to complain about!

PeterG
Parent - By Antares (****) [de] Date 2018-03-05 16:08

> I admire the enthusiasm of those who provide their time and resource to make these games and matches publicly available but as usual there are always people who find something to complain about!


+1, welcome to 2018.
Parent - - By gsgs (***) [de] Date 2018-03-08 04:12 Upvotes 1
I mean, it's basically the GUI's fault, but I think the engine could and should also
print the full last pv at fail-high/low - maybe with a warning sign like "#" so to
indicate that the >=,<= eval doesn't correspond to that old pv

It would be nice to see the actual move that is being revised, what the engine
had been overlooking/misevaluated

And I wouldn't consider it as unwelcome criticism, more like an improvement suggestion
Parent - - By Peter Grayson (****) [us] Date 2018-03-08 09:39

> And I wouldn't consider it as unwelcome criticism, more like an improvement suggestion


I agree but the criticism was focused against the engine rather than where the problem was.

> It would be nice to see the actual move that is being revised, what the engine
> had been overlooking/misevaluated


If the engine reconsiders its initial move going through multiple iterations at the same ply depth, very often the score shows (!) indicating an improvement or (--) indicating a worsening of assessment, but of course the extent of this is not seen when watching the Website transmissions. When a single or two ply PV is shown it is likely the engine has not completed its assessment at the ply depth and may have found a number of moves giving an equal score in which case it would be difficult for it to display a single multi-move PV string at that ply depth.

That is my interpretation of it because the engine always gives a full PV before moving on to the next ply depth but I may be wrong.

PeterG
Parent - By gsgs (***) [de] Date 2018-03-10 00:43
whenever the engine displays a fail high/low it could just display the current variation
upto the current move as well, although the eval will be revised and the variation
may no longer be the calculated best one.
Parent - - By MrKris (*) Date 2018-03-05 19:20 Edited 2018-03-05 19:26
Thank you, interesting!

Do you have a link? Was it about H6 specifically, and its PVs?

I will look again but all I found so far was very old, from 2010?! : http://www.open-chess.org/viewtopic.php?f=7&t=873

TCEC uses their version of cutechess-cli, linked to at the bottom of their live page.
YLCET uses ChessGUI, 'Info' button on their live page, the same as Graham Banks uses on talkchess.com.
I have Arena and Windows CMD window I could try compare to the TCEC & YLCET .pgn's  but it would be so much slower and would SMP vary so much - AMD Phenom II X6.
Parent - - By gsgs (***) [de] Date 2018-03-08 04:04 Edited 2018-03-08 04:32 Upvotes 1
a link to the log file ?  http://tcec.chessdom.com/dl.php?season=11&division=P&log=1&zip=1

ok, I'm downloading it now and checking the missing H6-pvs ...

--------edit-----------------------
I checked the season 10 superfinal log-file instead (459MB uncompressed, 100 games)

1st game , 2nd move for black=Houdini 6  has no pv

in the log-file is the full pv at depth 29 , while 1 upperbound and 2 lowerbound displays at depth 30
had only short pvs


825510   <Houdini 6.03(1): info multipv 1 depth 29 seldepth 53 score cp -11 time 155077 nodes 9862389333 nps 63596000 tbhits 0 hashfull 980 pv e7e6 g1f3 d7d5 b1c3 f8b4 c1g5 b8d7 c4d5 e6d5 e2e3 c7c5 d4c5 d8a5 f3d2 b4c3 b2c3 e8g8 f1e2 h7h6 g5h4 a5c5 e1g1 b7b6 h2h3 c8b7 d2b3 c5a3 d1d4 f8c8 f1d1 b7a6
...
825949   <Houdini 6.03(1): info multipv 1 depth 30 seldepth 53 score cp -18 upperbound time 155517 nodes 9890139410 nps 63595000 tbhits 0 hashfull 980 pv e7e6 g1f3
...
863617   <Houdini 6.03(1): info multipv 1 depth 30 seldepth 54 score cp -11 lowerbound time 193185 nodes 12278386341 nps 63557000 tbhits 0 hashfull 988 pv e7e6
...
863617   <Houdini 6.03(1): info multipv 1 depth 30 seldepth 54 score cp -11 lowerbound time 193185 nodes 12278389100 nps 63557000 tbhits 0 hashfull 988 pv e7e6
863617   <Houdini 6.03(1): bestmove e7e6 ponder g1f3

Parent - - By MrKris (*) Date 2018-03-09 03:08
a link to the log file ? Thank you for your link and idea of checking the logs! (As you note TCEC Season 10's is complete.)

Sorry, it is unclear, what I meant by "Do you have a link? Was it about H6 specifically, and its PVs?" was:
...
Houdart said this was the task of the GUI , and yes, the GUI should also be able to
easily do that (presumably only a small change in the code)
...
from your post above http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=578095#pid578095

gsgs, I am sure you are well aware of the UCI/GUI situation, you seem quite knowledgeable about it.
However, I would like to summarize it just for clarity, potentially for other readers of this thread.
Thanks for your ideas!
Again, just for other/new readers, my point is about Houdini with the current system: although it needs more data! - H6's higher frequency of short final PV's than Stockfish or Komodo. (A move's final PV shows in the TCEC/YLCET PV windows and .pgn's, also in Arena's .pgn's, if the option is selected, for 'home' games.)

- - -
The Universal Chess Interface (UCI) protocol is for both engines and GUI's so that both understand each other.
UCI description: http://wbec-ridderkerk.nl/html/UCIProtocol.html
It is extremely flexible, clearly by intention.

(A side note: this UCI link is dated 2004, however I read elsewhere that 'The King' inside the old ChessMaster, which has it own dedicated GUI, is UCI but with DRM, so I believe UCI is probably older than 2004.)

All GUI's that support UCI (Winboard/Xboard needs an adapter) follow the above 'universal' standard and expect all "UCI engines" to also follow it, the (singular, only one, the same one) UCI protocol.
- - -

From the above UCI description link: "...
* refutation   ...
     move  is refuted by the line  ... , i can be any number >= 1.
     Example: after move d1h5 is searched, the engine can send
     "info refutation d1h5 g6h5"
     if g6h5 is the best answer after d1h5 or if g6h5 refutes the move d1h5.
     if there is norefutation for d1h5 found, the engine should just send
     "info refutation d1h5"
    The engine should only send this if the option "UCI_ShowRefutations" is set to true.
..."
I assume all "UCI GUI's" have been ready for this -Arena has a checkbox for it for example-, although I am not aware of an engine that uses it.

gsgs " ... but I think the engine could and should also
print the full last pv at fail-high/low - maybe with a warning sign like "#" so to
indicate that the >=,<= eval doesn't correspond to that old pv ...
"

From the above UCI description link: " ...
* info
...
  * string
    any string str which will be displayed be the engine,
    if there is a string command the rest of the line will be interpreted as .
... "
Houdini 3 used the UCI 'info string' and warned with text (watching the current lines in Arena) when this occured: "Houdini 3 Released October 15, 2012 ...
The Accelerated Principal Variation Search or “Smart Fail-High” is especially useful in very deep analysis when a different move becomes best at very high search depth. Houdini 3 will apply an automatic depth reduction that often speeds up finding the Principal Variation by a factor of 5 to 10. ...
" http://www.cruxis.com/chess/manual/index.html?version_history.htm
Arena displays info strings, and was fine with this feature (the .pgn I think that Arena only will output the last evaluation, depth and PV, not a string before it).
Parent - - By gsgs (***) [de] Date 2018-03-10 01:03 Upvotes 1
Mr Kris, there was a discussion in TCEC-chat AFAIR (so no link) , probably during superfinale 10 ,
where Houdart said this.
My interpretation :
He saw no reason to change anything in Houdini, since those missing pvs
where an issue of the GUI, they should rather change the GUI. (cutechess)

I know little about Arena or other GUIs , except Droidfish, and a little cutechess-cli.
I just run analysis from batch-files and scripts.

But I assume the GUI just receives what we see in the log files and "interprets" it and decides
what of it to display and how.

Now, why has Houdini apparently more fail high/lows than e.g. Komodo ?
I don't know. Maybe the built-in-contempt. Maybe the time-management that
rarer waits for completion of a depth.
Parent - By MrKris (*) Date 2018-03-11 05:10
Thank you again!! Very interesting!
Yes, the log is just the engines' stdin (from the GUI) and stdout (to the GUI) but labled. I can be re-created, although with SMP variance ect., with the Command Window/Terminal with UCI commands 'go movetime [milliseconds]' etc.

With your clues I searched and the chat was saved and posted on TalkChess:

Post subject: Re: Why PVs aren't displayed -- Robert Houdart in TCEC chat
http://talkchess.com/forum/viewtopic.php?t=65874

Extracts:
Marco Belli   Posted: Mon Dec 04, 2017 8:12 am
I understood that it's houdini that send an empty pv. I don't think that latest cutechess can solve the problem

Reply to: Robert Houdart [Houdini author] Posted: Mon Dec 04, 2017 8:52 am " ... It's not a fix of a bug, it's a new feature. [proposed for GUI's]...":
Marco Belli [Valjolet author]  Posted: Mon Dec 04, 2017 9:23 am 
Another solution is that the engine will not print the pv of a not finished iteration.i think it's the solution adopted by most engines.
This way the last pv string sent to cutechess is the full pv of the last finished iteration.
[unanswered]

and

Peter Österlund [Droidfish author] Posted: Tue Dec 05, 2017 5:43 pm
...
"marklefler[Komodo co-author]:Robert thanks. Are all Houdini shortened PV due to fail high/low unresolved searches? I ask because the number of one move PVs is very high. For example, just one full PV this game so far. {MrKris: "one move PV" = just the move played, Marco Belli above "empty pv"}

RobertHoudart:@marklefler just interrupted searches, the PV is simply not yet available at this iteration"
-TCEC Chat

That is not the only case where a short PV is sent to the GUI though. From the S10_Stage_1_debug.txt file:
Code:
36228525   <Houdini 6.02(11): info multipv 1 depth 27 seldepth 46 score cp -23 time 67342 nodes 2346491893 nps 34844000 tbhits 0 hashfull 715 pv b7b6 a2a3 c7c6 c3c4 b8d7 f1e2 c8b7 e1g1 f8g7 c4d5 c6d5 b1d2 f6e4 a1c1 e8g8 d1c2 a8c8 c2a4 e4d2 f3d2 c8a8 a4c2 a8c8 c2a4
36228525   <Houdini 6.02(11): info multipv 1 depth 27 seldepth 46 score cp -19 time 67343 nodes 2346492776 nps 34843000 tbhits 0 hashfull 715 pv b7b6 a2a3
36228525   <Houdini 6.02(11): bestmove b7b6 ponder a2a3

Houdini first produces a long PV for depth 27 but then immediately replaces it with a short PV before the bestmove is announced.
[unanswered]
Parent - By Antares (****) [de] Date 2018-03-08 10:01 Upvotes 1

> but I think the engine could and should also print the full last pv at fail-high/low


Indeed it would be great when engines would [optionally] provide additional information, for example why a root move "fails" and is exchanged for another.
Parent - By MrKris (*) Date 2018-03-05 19:44
gsgs mentioned something about TCEC's GUI this thread: By gsgs (**) [de] Date 2018-03-02 23:58 Edited 2018-03-03 00:03
I missed that, I will try to look into it.
What I tallied so far from their 'Archive' viewer:

YLCET Super X3 http://chess.my2dollars.ca/live.html
Houdini 6.03: 70% of 1 through 10 moves after the book have PV's with 0 after-move plies,
Stockfish dev: 15% of 1 through 10 moves after the book have PV's with 0 after-move plies:

Rd. 1.1  1.2  3.1  3.2  4.1  4.2  6.1  6.2
eng k-H  H-k  H-s  s-H  k-H  H-k  H-s  s-H
 1. m m  0 m  0 m  m m  m 0  0 m  m m  m 0 
 2. m m  0 m  m m  0 0  m m  0 m  m m  m 0
 3. 2 m  0 2  0 m  m m  2 0  0 m  1 0  m 0
 4. 2 m  0 2  0 m  0 m  2 0  m m  0 m  0 0
 5. m m  0 2  0 m  m m  m 0  0 2  0 m  0 0
 6. m m  0 m  0 m  m 0  m 0  0 2  1 m  m 0
 7. m m  1 m  0 m  m m  2 m  0 m  0 m  m 0
 8. 2 0  0 m  0 m  0 m  m 0  0 2  1 m  m 0
 9. m 0  m 2  0 m  m 0  2 0  0 2  0 1  m 0
10. m 0  0 2  0 m  m 0  m 0  0 m  1 m  m 0
plies after move:                                     
<10 3 0  1 5  0 0  0 0  4 0  0 4  4 1  0 0   H= 5/80  s= 1/40  k=16/40
  m 7 7  1 5  1 10 7 5  6 2  1 6  2 8  8 0   H=19/80  s=33/40  k=24/40
  0 0 3  8 0  9 0  3 5  0 8  9 0  4 1  2 10  H=56/80  s= 6/40  k =0/40 
    H<k  H<k  H<<s H<s  H<k  H<k  H<s  H<<s
<10 = 1 to 10 plies after played move
m = > 10 ; 0 = played move in PV only
Parent - By Akbarfan (***) Date 2018-02-09 19:58
I see you are using Komodo 2005.00 and therefore you are betatester for Komodo, right?

How high is your estimate Elo gain K 2005.00 compared to K 11.2.2 ?
Parent - By MrKris (*) Date 2018-03-09 03:12 Edited 2018-03-09 03:15
What I tallied so far from their 'Archive' viewer:

YLCET Super X3 http://chess.my2dollars.ca/live.html
1 through 10 moves after the book:

Houdini 6.03     : 67% of moves have PV's with 0 after-move plies,
Stockfish vs Ho : 21% of moves have PV's with 0 after-move plies,
Komodo vs Ho   : 39% 1 to 10 plies after move played, 61% >10 plies after move played

* 1/16 games Houdini more after move PV plies than opponent
Rd. 1.1  1.2  3.1  3.2  4.1  4.2  6.1  6.2    7.1  7.2  9.1  9.2 10.1 10.2 12.1 12.2 
eng k-H  H-k  H-s  s-H  k-H  H-k  H-s  s-H    k-H  H-k  H-s  s-H  k-H  H-k  H-s  s-H
 1. m m  0 m  0 m  m m  m 0  0 m  m m  m 0    2 0  0 m  0 0  0 0  2 0  0 2  0 m  0 0
 2. m m  0 m  m m  0 0  m m  0 m  m m  m 0    m m  0 2  1 m  m 0  m 0  0 m  m m  m 0
 3. 2 m  0 2  0 m  m m  2 0  0 m  1 0  m 0    m 0  0 2  0 m  m 0  2 0  0 m  0 m  0 8
 4. 2 m  0 2  0 m  0 m  2 0  m m  0 m  0 0    2 m  0 2  0 m  m 1  2 0  0 m  0 m  m 0
 5. m m  0 2  0 m  m m  m 0  0 2  0 m  0 0    m 0  m 2  1 m  m 0  m 0  0 m  0 m  0 m
 6. m m  0 m  0 m  m 0  m 0  0 2  1 m  m 0    2 0  m m  0 0  m m  2 0  m m  1 m  m m
 7. m m  1 m  0 m  m m  2 m  0 m  0 m  m 0    m m  m m  0 0  m m  m m  0 m  0 0  0 m
 8. 2 0  0 m  0 m  0 m  m 0  0 2  1 m  m 0    m 0  m m  0 m  m m  2 0  m m  0 m  m 0
 9. m 0  m 2  0 m  m 0  2 0  0 2  0 1  m 0    m m  m m  0 m  m 0  m 0  0 m  m m  0 m
10. m 0  0 2  0 m  m 0  m 0  0 m  1 m  m 0    m 0  m m  0 m  m 0  2 0  m 2  0 m  0 m
plies after move:                                   
<10 3 0  1 5  0 0  0 0  4 0  0 4  4 1  0 0    3 0  0 4  2 0  0 1  6 0  0 2  1 0  0 1
  m 7 7  1 5  1 10 7 5  6 2  1 6  2 8  8 0    7 4  6 6  0 7  9 3  4 1  3 8  2 9  4 5
  0 0 3  8 0  9 0  3 5  0 8  9 0  4 1  2 10   0 6  4 0  8 3  1 6  0 9  7 0  7 1  6 4
    H<k  H<k  H<<s H<s  H<k  H<k  H<s  H<<s   k>H  H<k  H<s  s>H k>>H  H<k  H<s *s<H
<10 = 1 to 10 plies after played move
m = > 10 ; 0 = played move in PV only                                        Totals    Ho     Sf    Ko
<10  H= 5/80  s= 1/40  k=16/40                <10  H= 5/80  s= 0/40  k=15/40    <10  10/160  1/80 31/80
  m  H=19/80  s=33/40  k=24/40                  m  H=24/80  s=29/40  k=25/40    >10  43/160 62/80 49/80
  0  H=56/80  s= 6/40  k =0/40                  0  H=51/80  s=11/40  k= 0/40      0 107/160 17/80  0/80
Up Topic The Rybka Lounge / Computer Chess / Houdini 6.03 - seldom any early move PV's ?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill