![[fi]](/mwf/v12/flags/fi.png "Finland")
When I tested Rybka with PH enabled against Glaurung in 1.(!) game Rybka loses this way:
[Event "Nunn"]
[Site "?"]
[Date "2008.09.05"]
[Round "1"]
[White "Rybka 3 1-cpu 32-bit PH"]
[Black "Glaurung 2.1"]
[Result "0-1"]
[ECO "B33"]
[Annotator "0.16;-0.07"]
[PlyCount "66"]
[TimeControl "60/300:0/0:0/0"]
1. e4 c5 2. Nf3 e6 3. d4 cxd4 4. Nxd4 Nc6 5. Nb5 Nf6 6. N1c3 d6 7. Bf4
e5 8. Bg5 a6 9. Na3 b5 10. Bxf6 gxf6 11. Nd5 f5 12. c3 Bg7 13. exf5 Bxf5 {
Both last book move} 14. Qf3 {0.16/10 13} Be6 {-0.07/14 14} 15. O-O-O {
(Bd3) 0.10/10 5} O-O {-0.62/12 10} 16. Nc2 {(Kb1) 0.11/10 4} f5 {
(Ne7) -1.01/13 14} 17. Kb1 {0.03/10 24} Ne7 {(Rc8) -1.05/14 14} 18. Nce3 {
(Nxe7+) 0.06/10 15} Rc8 {-1.45/13 8} 19. g3 {(Rd2) 0.00/10 6} Rc5 {
(Nxd5) -1.35/13 14} 20. Nxe7+ {(Nb4) -0.10/10 10} Qxe7 {-1.62/10 0} 21. Bg2 {
(Bh3) -0.17/10 9} e4 {(b4) -1.96/13 11} 22. Qf4 {(Qe2) -0.68/9 14} Be5 {
-2.76/13 11} 23. Qh4 {-1.04/10 19} Qf7 {(Qxh4) -2.76/13 10} 24. g4 {
(a3) -0.87/8 7} Bxa2+ {(b4) -3.92/12 10} 25. Ka1 {-2.01/8 11} Qb3 {
(Bb3) -4.80/12 9} 26. Qg5+ {(gxf5) -4.24/8 51} Kh8 {-7.31/12 9} 27. Rd5 {
(Qe7) -4.24/8 35} Qa4 {(Rxd5) -23.50/11 5} 28. Qf6+ {(Nc2) -25.28/8 24} Bxf6 {
-#10/11 2} 29. Rd4 {(Nc2) -25.29/7 2} Bxd4 {-#6/10 1} 30. Rd1 {(Bf1) -#10/5 0}
Bxe3 {-#4/6 0} 31. Rd4 {-#8/3 0} Bc4+ {(Bxd4) -#3/6 0} 32. Kb1 {-#3/3 0} Qa2+ {
-#2/6 0} 33. Kc2 {-#2/3 0} Bb3# {-#1/6 0} 0-1
In second and succeeding tries Rybka loses same way, only first moves after book are executed instantly :-) Note, that
default Rybka won from this opening. So is anybody really using PH feature?
Another thing I don't understand = meaning of save/load hash. When analysing some positions longer it doesn't change anything
to automatic persistent hash feature. And there is problem with main line: only one move is displayed = almost useless analysis.
So is anybody really using load/save hash feature?
Jouni
[Event "Nunn"]
[Site "?"]
[Date "2008.09.05"]
[Round "1"]
[White "Rybka 3 1-cpu 32-bit PH"]
[Black "Glaurung 2.1"]
[Result "0-1"]
[ECO "B33"]
[Annotator "0.16;-0.07"]
[PlyCount "66"]
[TimeControl "60/300:0/0:0/0"]
1. e4 c5 2. Nf3 e6 3. d4 cxd4 4. Nxd4 Nc6 5. Nb5 Nf6 6. N1c3 d6 7. Bf4
e5 8. Bg5 a6 9. Na3 b5 10. Bxf6 gxf6 11. Nd5 f5 12. c3 Bg7 13. exf5 Bxf5 {
Both last book move} 14. Qf3 {0.16/10 13} Be6 {-0.07/14 14} 15. O-O-O {
(Bd3) 0.10/10 5} O-O {-0.62/12 10} 16. Nc2 {(Kb1) 0.11/10 4} f5 {
(Ne7) -1.01/13 14} 17. Kb1 {0.03/10 24} Ne7 {(Rc8) -1.05/14 14} 18. Nce3 {
(Nxe7+) 0.06/10 15} Rc8 {-1.45/13 8} 19. g3 {(Rd2) 0.00/10 6} Rc5 {
(Nxd5) -1.35/13 14} 20. Nxe7+ {(Nb4) -0.10/10 10} Qxe7 {-1.62/10 0} 21. Bg2 {
(Bh3) -0.17/10 9} e4 {(b4) -1.96/13 11} 22. Qf4 {(Qe2) -0.68/9 14} Be5 {
-2.76/13 11} 23. Qh4 {-1.04/10 19} Qf7 {(Qxh4) -2.76/13 10} 24. g4 {
(a3) -0.87/8 7} Bxa2+ {(b4) -3.92/12 10} 25. Ka1 {-2.01/8 11} Qb3 {
(Bb3) -4.80/12 9} 26. Qg5+ {(gxf5) -4.24/8 51} Kh8 {-7.31/12 9} 27. Rd5 {
(Qe7) -4.24/8 35} Qa4 {(Rxd5) -23.50/11 5} 28. Qf6+ {(Nc2) -25.28/8 24} Bxf6 {
-#10/11 2} 29. Rd4 {(Nc2) -25.29/7 2} Bxd4 {-#6/10 1} 30. Rd1 {(Bf1) -#10/5 0}
Bxe3 {-#4/6 0} 31. Rd4 {-#8/3 0} Bc4+ {(Bxd4) -#3/6 0} 32. Kb1 {-#3/3 0} Qa2+ {
-#2/6 0} 33. Kc2 {-#2/3 0} Bb3# {-#1/6 0} 0-1
In second and succeeding tries Rybka loses same way, only first moves after book are executed instantly :-) Note, that
default Rybka won from this opening. So is anybody really using PH feature?
Another thing I don't understand = meaning of save/load hash. When analysing some positions longer it doesn't change anything
to automatic persistent hash feature. And there is problem with main line: only one move is displayed = almost useless analysis.
So is anybody really using load/save hash feature?
Jouni
![[at]](/mwf/v12/flags/at.png "Austria")
There is a bug in persistant hash. You should not use it in game play. The bug will cause it to sometimes move very quickly and sometimes lose the game because of this. This bug does not affect analysis mode.
![[ru]](/mwf/v12/flags/ru.png "Russian Federation")
>Another thing I don't understand = meaning of save/load hash. When analysing some positions longer it doesn't change anything
>to automatic persistent hash feature.
Saved hash it's not Persistent Hash. Read about two different hashes here.
Hi,
Yes I printed and read it carefully without understanding much! Can You or any forum reader post example of
how to benefit from saving/loading hash, please?
Jouni
Yes I printed and read it carefully without understanding much! Can You or any forum reader post example of
how to benefit from saving/loading hash, please?
Jouni
Example: I've done analysis of initial position a week ago, and I've saved hash (512 MB file). Now I can continue:
Analysis by Rybka 3 1-cpu 32-bit:
1.e4 Nc6 2.Nf3
= (0.20) Depth: 17 00:00:00 0kN
1.e4 Nc6 2.Nf3
= (0.20) Depth: 18 00:00:00 0kN
Oh, I have Persistent Hash stored for this position too. Now I cleared hash from memory and loaded Persistent Hash:
Analysis by Rybka 3 1-cpu 32-bit:
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 6 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 7 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 8 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 9 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 10 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 11 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 12 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 13 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 14 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 15 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 16 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 17 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 18 00:00:00
What is difference between these two hashes? Well, perhaps, Saved Hash from first example along with Preserve Analysis can store more information about a position. You know how to use Preserve Analysis, right? Why don't someone try using Saved Hash with Persistent Hash and tell us if it's different. I don't like Preserve Analysis in Rybka - it's too slow.
Analysis by Rybka 3 1-cpu 32-bit:
1.e4 Nc6 2.Nf3
= (0.20) Depth: 17 00:00:00 0kN
1.e4 Nc6 2.Nf3
= (0.20) Depth: 18 00:00:00 0kN
Oh, I have Persistent Hash stored for this position too. Now I cleared hash from memory and loaded Persistent Hash:
Analysis by Rybka 3 1-cpu 32-bit:
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 6 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 7 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 8 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 9 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 10 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 11 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 12 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 13 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 14 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 15 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 16 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 17 00:00:00
1.e4 Nc6 2.Nf3 Nf6 3.e5 Nd5 4.c4
= (0.20) Depth: 18 00:00:00
What is difference between these two hashes? Well, perhaps, Saved Hash from first example along with Preserve Analysis can store more information about a position. You know how to use Preserve Analysis, right? Why don't someone try using Saved Hash with Persistent Hash and tell us if it's different. I don't like Preserve Analysis in Rybka - it's too slow.
"Set the (Saved Hash File) engine parameter to where you want hash contents to be saved. The path should point directly to the file itself (ie. not just the directory). By convention, saved hash files have an extension of .rsh.
4) Whenever you have performed a deep analysis on some specific position and expect to want Rybka to access this work later, invoke the (Save Hash) button from the engine parameter list.
5) When you want to restore the analysis from step #3, make sure that the (Preserve Analysis) box is checked and invoke the (Load Hash) button from the engine parameter list."
Jouni, I agree with you. It is not clear what one is supposed to do after completing IDeA analysis, say overnight, with the intention of using the output of IDeA in Infinite Analysis. Also, this "explanation" does not make clear what is the difference between "Saved Hash" and "Persistent Hash" other than that they are different files.
Does "deep analysis" mean IDeA? In 5) we are told to check Preserve Analysis, then click Load Hash. Does one then uncheck "Preserve Analysis"? Note that Vas has told us never to use Preserve Analysis, so it is too confusing to be useful.
4) Whenever you have performed a deep analysis on some specific position and expect to want Rybka to access this work later, invoke the (Save Hash) button from the engine parameter list.
5) When you want to restore the analysis from step #3, make sure that the (Preserve Analysis) box is checked and invoke the (Load Hash) button from the engine parameter list."
Jouni, I agree with you. It is not clear what one is supposed to do after completing IDeA analysis, say overnight, with the intention of using the output of IDeA in Infinite Analysis. Also, this "explanation" does not make clear what is the difference between "Saved Hash" and "Persistent Hash" other than that they are different files.
Does "deep analysis" mean IDeA? In 5) we are told to check Preserve Analysis, then click Load Hash. Does one then uncheck "Preserve Analysis"? Note that Vas has told us never to use Preserve Analysis, so it is too confusing to be useful.
![[de]](/mwf/v12/flags/de.png "Germany")
Besides the bug Linus mentioned there is another serious bug which will fxxx up your hash file if you use the feature "next best move" during analysing. I would say right now using hash files is generally not advised.
![[us]](/mwf/v12/flags/us.png "United States")
Do Vas and Larry still have the position that no bug fixes will be made to Rybka 3
before Rybka 4 is released?
I have not heard from either of them on this forum, since Vas said "be patient",
which says to me they are trying to have the forum users find all the bugs that
they didn't find before release. And, also assess, the seriousness of the bugs before
making a decision. I think that all the bugs mentioned so far are of a seroius nature
and should be fixed before R4. Maybe Vas and Larry disagree, but I have stated my
opinion here several times already. Where's Jessica when you need her? :)
I would like to hear from Vas and Larry on their Bug Fix position with respect to R3.
Customers have a right to getting the product that they were promised and were sold!
If anyone agrees with me then let them speak up as I do! If everyone wants to wait
a year thats OK too. Maybe Rybka will not be in the lead on the rating charts by that time.
before Rybka 4 is released?
I have not heard from either of them on this forum, since Vas said "be patient",
which says to me they are trying to have the forum users find all the bugs that
they didn't find before release. And, also assess, the seriousness of the bugs before
making a decision. I think that all the bugs mentioned so far are of a seroius nature
and should be fixed before R4. Maybe Vas and Larry disagree, but I have stated my
opinion here several times already. Where's Jessica when you need her? :)
I would like to hear from Vas and Larry on their Bug Fix position with respect to R3.
Customers have a right to getting the product that they were promised and were sold!
If anyone agrees with me then let them speak up as I do! If everyone wants to wait
a year thats OK too. Maybe Rybka will not be in the lead on the rating charts by that time.
![[ch]](/mwf/v12/flags/ch.png "Switzerland")
I cant remember the post but Vas said somewhere here that we should be patient, indicating that there will surely be a bug fix.
Now the question is when, and the answer is be patient. :-)
regards
PS I am less patient to see Rybka on my iPhone or PPC
Now the question is when, and the answer is be patient. :-)
regards
PS I am less patient to see Rybka on my iPhone or PPC
So this much touted persistent hash-feature is not recommended at all! OK I give up with that feature.
Jouni
Jouni
Persistent hash os ok in analysis mode as long as you don't use this "next best move" thing, which I never do anyway.
Who knows? Do you know how exactly IDeA in Aquarium does get alternative moves? Just as an example...
So I have also totally abandoned persistent hash for the time being.
So I have also totally abandoned persistent hash for the time being.
You are right. I did not think of that. Maybe it is safer to deactivate it for now and wait for the Rybka 3.01 bugfix update.
Vas was talking about "Preserve Analysis". Vas has told us never to use Preserve Analysis.
> Vas has told us never to use Preserve Analysis.
Ah, where has he said that? To my knowledge he just pointed out the PH bug in game play.
Anyway, this is quite disappointing. PH was one of the key features of R3.
I tried to Load the Saved Hash file (with "Preserve Analysis" enabled and also tried without it enabled) but no luck either way. Has anyone been able to Load Saved Hash file? Any hints to give me?
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill