Moderator: Ras
All that it shows (again) is, that a position test sooner or later has problems to rank a new engine - as all position tests showed in the past ...swami wrote: My guess is that it has improved by over 60 Elo. STS Elo Estimate certainly showed more than 60 elo change between STS Total scores of Stockfish 1.8 (3124) and Stockfish 1.9 (3210). It could be anywhere between 40 and 60 elo... It could even be more than 60 though. We could say 40-50 just to be on safe side.
Thanks Paul,NATIONAL12 wrote:I think Paul that your 9775 is not overclocked .Why?
my Skully is overclocked to 3.6,latest 17 chips run faster due to a bottle-neck caused by FB-DIMMS having to be used in Skully.
i have just started a 40/40 engine game R4-Fire 1.31,both run on 12cpu.will post later when finished so you can see speeds.
[Event ", 40m/40+40m/40+40m"]Theodor wrote:Thanks Paul,NATIONAL12 wrote:I think Paul that your 9775 is not overclocked .Why?
my Skully is overclocked to 3.6,latest 17 chips run faster due to a bottle-neck caused by FB-DIMMS having to be used in Skully.
i have just started a 40/40 engine game R4-Fire 1.31,both run on 12cpu.will post later when finished so you can see speeds.
It will be very interesting.
WOW !!!!!!!!!NATIONAL12 wrote:[Event ", 40m/40+40m/40+40m"]Theodor wrote:Thanks Paul,NATIONAL12 wrote:I think Paul that your 9775 is not overclocked .Why?
my Skully is overclocked to 3.6,latest 17 chips run faster due to a bottle-neck caused by FB-DIMMS having to be used in Skully.
i have just started a 40/40 engine game R4-Fire 1.31,both run on 12cpu.will post later when finished so you can see speeds.
It will be very interesting.
[Site "PAULWATSON-PC"]
[Date "2010.10.04"]
[Round "1"]
[White "Deep Rybka 4 SSE42 x64"]
[Black "Fire 1.31 x64"]
[Result "1-0"]
[ECO "B17"]
[Annotator "0.38;0.39"]
[PlyCount "123"]
[TimeControl "40/2400:40/2400:2400"]
{Intel(R) Xeon(R) CPU X5680 @ 3.33GHz 4186 MHz W=22.0 plies; 1,
080kN/s; 91,465 TBAs; Masterbook 3.0.ctg B=24.9 plies; 21,824kN/s; Masterbook
3.0.ctg} 1. e4 {B 0} c6 {B 0} 2. d4 {B 0} d5 {B 0} 3. Nc3 {B 0} dxe4 {B 0} 4.
Nxe4 {B 0} Nd7 {B 0} 5. Nf3 {B 0} Ngf6 {B 0} 6. Nxf6+ {B 0} Nxf6 {B 0} 7. Bc4 {
B 0} e6 {B 0} 8. O-O {B 0} Be7 {B 0} 9. Qe2 {B 0} O-O {B 0} 10. c3 {B 0} b6 {
B 0 Both last book move} 11. Bf4 {0.38/19 169} Bb7 {0.39/23 68 (Nd5)} 12. Be5 {
0.34/18 127 (Rad1)} Qc8 {0.29/22 65 (Bd6)} 13. Rfe1 {0.30/18 79 (Rad1)} c5 {0.
29/20 21 (h6)} 14. Rad1 {0.30/18 66} Rd8 {0.26/19 31} 15. Bd3 {0.30/18 78 (h3)}
h6 {0.24/23 199} 16. dxc5 {0.34/18 43 (h3)} Qxc5 {0.30/23 70} 17. Nd4 {0.34/18
74} a5 {0.30/22 111 (a6)} 18. a3 {0.35/18 62 (a4)} Nd7 {0.33/23 274 (Bd5)} 19.
Bg3 {0.44/19 114} Bf6 {0.35/22 194 (Kh8)} 20. Bb5 {0.33/21 127 (Bc2)} Bd5 {0.
36/25 61 (Nf8)} 21. c4 {0.60/20 43} Bxd4 {0.36/23 1} 22. cxd5 {0.60/21 25} e5 {
0.34/25 72} 23. Rc1 {0.60/20 52 (Bc6)} Qxd5 {0.35/25 77} 24. Bc6 {0.59/20 18}
Qb3 {0.35/25 81} 25. Bxa8 {0.66/20 39} Rxa8 {0.35/23 0} 26. Rcd1 {0.78/20 75}
f6 {0.43/25 91} 27. Kh1 {0.85/21 48 (Rd2)} Nc5 {0.45/26 66 (Rf8)} 28. f4 {0.97/
22 74} a4 {0.58/26 104 (Rf8)} 29. Rd2 {0.96/22 70} Rf8 {0.62/25 55} 30. Rf1 {
0.96/22 88} Rf7 {0.65/26 71 (Qd5)} 31. fxe5 {1.25/21 56} fxe5 {0.65/24 0} 32.
Bf2 {1.13/21 65} Ne6 {0.79/26 93} 33. Bxd4 {1.28/22 68} Rxf1+ {0.90/27 69} 34.
Qxf1 {1.36/21 14} exd4 {0.90/27 43} 35. Kg1 {1.50/23 104 (g3)} Qe3+ {0.85/27
72 (Qd5)} 36. Qf2 {1.38/23 52} Qg5 {0.83/27 193 (Qe4)} 37. Re2 {2.18/21 54
(Qe2)} Nf4 {1.25/25 72 (Qd5)} 38. Re8+ {2.15/20 24} Kh7 {1.09/23 0} 39. Qf3 {
2.06/21 21} d3 {1.56/25 107} 40. Qe4+ {2.06/22 61} Ng6 {1.56/24 0} 41. Qxd3 {
2.17/22 13} Qc1+ {1.65/26 66} 42. Kf2 {2.34/24 137} Qxb2+ {1.65/26 91} 43. Re2
{2.39/24 37} Qf6+ {1.63/26 96} 44. Qf3 {2.39/24 19} Qh4+ {1.65/26 85} 45. Kg1 {
2.42/24 25} Qd4+ {1.75/27 113} 46. Qe3 {2.60/20 23} Qd1+ {1.75/27 73} 47. Kf2 {
2.63/20 30 (Re1)} b5 {2.09/23 62 (Qd5)} 48. g3 {2.91/21 30} Qd7 {2.22/25 71
(Qd6)} 49. h4 {3.40/23 65 (Kg2)} Nf8 {2.58/25 80 (Qd6)} 50. Qe8 {3.50/25 110
(Kg2)} Qd6 {2.72/26 66} 51. Qxb5 {3.51/24 55 (Kg2)} Qxa3 {2.14/23 1} 52. Kg2 {
3.58/25 30} Qb3 {2.34/26 45 (Qd6)} 53. Qxb3 {3.73/28 118} axb3 {2.34/24 0} 54.
Rb2 {3.73/29 40 (Re3)} Ne6 {2.64/26 67 (Ng6)} 55. Rxb3 {3.85/30 45} Kg6 {2.75/
27 98 (Nc5)} 56. Rd3 {4.37/30 80 (Rb4)} Kf6 {2.83/27 74 (Nf8)} 57. Kf3 {5.04/
29 63} Kg6 {3.23/27 163 (Nc7)} 58. Rd6 {10.68/28 37} Kf7 {3.23/28 84 (Kf5)} 59.
Ke4 {10.68/24 13} Ke7 {6.71/28 300 (h5)} 60. Ke5 {10.20/20 14} Nc5 {7.84/28
183 (Nf8)} 61. Rc6 {11.32/19 7} Nd3+ {6.71/25 0} 62. Kf5 {11.63/24 4} 1-0
good job i am not into speedboats.It certainly is difficult to rank an engine by just using 12 set of positions test but this one is doing good thus far IMO (as it gives nearly a good rough idea but not necessarily exactly an accurate)IWB wrote:Hello Swami,
All that it shows (again) is, that a position test sooner or later has problems to rank a new engine - as all position tests showed in the past ...swami wrote: My guess is that it has improved by over 60 Elo. STS Elo Estimate certainly showed more than 60 elo change between STS Total scores of Stockfish 1.8 (3124) and Stockfish 1.9 (3210). It could be anywhere between 40 and 60 elo... It could even be more than 60 though. We could say 40-50 just to be on safe side.
Regards and bye
Ingo
My feeling based on looking at games is that the stockfish team removed some productive king safety evaluation at long time controlDon wrote:It appears from our own testing of Komodo that most evaluation improvements can be proved at very fast time controls or even fixed depth games. But there are some notable exceptions. I think highly dynamic things such as king safety need more depth.mcostalba wrote:Extrapolating to long time controls is a difficult and tricky exercise. Regarding evaluation I have experienced that most of the cases (but not all) can be safely proved at fast TC and they will hold also at longer TC.Uri Blass wrote:Is it evaluation stuff that is almost useless because it caused stockfish to be slower or simply because it is not clear if the evaluation is better even with fixed number of nodes?Jim Ablett wrote:
Stockfish 1.9 JA by the Stockfish team.
We have mainly removed evaluation stuff that proved to be almost
useless, we have now access to a better hardware facility(*) and we
are able to test with a better resolution our evaluation code so to
remove old stuff that we never dared to touch.
In the second case it is a good idea to remove it.
In the first case I am not sure because I suspect that evaluation knowledge (that is productive assuming no price in speed) can help more at longer time control.
Anyhow to answer your question, is the first case. We never test at fixed depth because is a very artificial condition and _could_ lead to artifacts.
stockfish(including stockfish1.9) clearly outsearch Komodo so it seems that you still use very safe pruning relative to stockfish.Don wrote: In my own testing it seems like every version that is a little safer plays worse at longer time controls but I have never tried to measure this with low enough error margins. As you point out this requires a LOT of CPU power to do well.
Hello Uri, what about the time used to reach this depths ? Is one version faster ?Uri Blass wrote:stockfish(including stockfish1.9) clearly outsearch Komodo so it seems that you still use very safe pruning relative to stockfish.Don wrote: In my own testing it seems like every version that is a little safer plays worse at longer time controls but I have never tried to measure this with low enough error margins. As you point out this requires a LOT of CPU power to do well.
It is possible that more safe pruning is better for stockfish when less safe pruning is better for komodo.
based on testing at fixed depth the difference in pruning between stockfish1.9 and stockfish1.8 is very small at small depths
It is possible that Stockfish1.9 does not like the noomen test but here are my result in these positions at fixed depth
depth 2:
Stockfish1.8-stockfish1.9 55:45
depth 7:
Stockfish1.8-stockfish1.9 52:48
depth 12:
Stockfish1.9-Stockfish1.8 50.5:49.5
