Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Support / Rybka 4.1 Release Notes
1 2 3 Previous Next  
Parent - By Somnus (**) [ca] Date 2011-03-07 21:03
Thanks Vas. Should be interesting to test out.

Peter
Parent - - By Marcos Alexandre (**) Date 2011-03-07 23:18
Thx, Vas work very well for me.:wink:
Parent - By Maxou93 (**) [fr] Date 2011-03-08 11:51
Mr Rajlich,
Thank you very much. I'm under Aquarium and it work also very well for me. No path problem, program found the exe by itself.
Best regards
Parent - - By Wayne Lowrance (***) Date 2011-03-08 01:55
Howdy. As per usual I have difficulty with up dates. I had purchased Rybka 4 from chessbase (fritz interface)
I downloaded 4.1  and pointed it to where I feel is right, which was:

program files (x86\common files\chessbase\Engines.uci\rybka 4
did not work. I pointed it to many other locations etc. no work. What up ?
thank you
Wayne
Parent - - By Uly (Gold) [mx] Date 2011-03-08 03:09
Point it to where you have the .exe files.
Parent - By Wayne Lowrance (***) Date 2011-03-08 04:44
Thank you. Strange. I do a search "rybka*.exe" and only thing that pops out is the 4.1 update.
I cannot find any chess program *.exe. Yet all my programs work ? any advise ? I hate Computers.
I should go back to OTB, but at 81, I think not :)
Wayne
Parent - - By Wayne Lowrance (***) Date 2011-03-08 04:52
Well praise the lord. I did it. With the new installation I just needed to point one step deeper.
My thank Uly
p.s. good luck in remainder of Jimmy's tournament.
Wayne
Parent - By TheHug (Bronze) [us] Date 2011-03-08 04:53
Good that you got it fix. I was about to try and give you a hand :)
Parent - By Uly (Gold) [mx] Date 2011-03-08 05:31

> good luck in remainder of Jimmy's tournament.


Thanks, I'm looking forward to playing you at some point. Good luck as well!
Parent - By Quapsel (****) [de] Date 2011-03-08 07:52

> Rybka 4.1 is the latest UCI Rybka chess engine and is a free update for Rybka 4 customers.


Thanx, Vas!
We'll have an  interested look...
Quap
Parent - - By Pascal Pflaum [de] Date 2011-03-08 07:59
Rybka 4.1 doesn't support "go nodes x" anymore. Can anyone confirm this problem?
Parent - By Vempele (Silver) [fi] Date 2011-03-08 08:12
Confirmed.
Parent - - By EvanKatz [us] Date 2011-03-08 19:35
Just a small correction to the on-line instructions for the Rybka 4.1 upgrade.  Specifically, in the instructions (which are set forth at http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=21098), it says:

Note 1: The updater should be pointed to the directory in which the Rybka 4 executables reside. If you have trouble finding the Rybka 4 executables, you can perform a system file search for "Rybka*.exe".

However, this is incorrect.  Instead of “Rybka*.exe”, the correct search term should be “*Rybka*4*.exe” (with the addition of an asterisk before “Rybka”, and also the insertion of a “*4” immediately after “Rybka”), for the following two reasons: 

1.  The Deep Rybka program installs executables called “Deep Rybka”, so that they will not be disclosed by the original search term suggested on your web site.  Accordingly, users need to insert the asterisk wild card before “Rybka” (as well as after it).

2.  By inserting the number 4, the many users who have many older versions of Rybka installed (i.e., Rybka 2 and 3), will not get confused with the many directories and files that contain such older versions.
Parent - By Dragon Mist (****) [hr] Date 2011-03-08 21:19

> you can perform a system file search for "Rybka*.exe".
>
> However, this is incorrect.  Instead of “Rybka*.exe”, the correct search term should be “*Rybka*4*.exe” (with the addition of an asterisk before “Rybka”, and also the insertion of a “*4” immediately after “Rybka”), for the following two reasons: 
>
> 1.  The Deep Rybka program installs executables called “Deep Rybka”, so that they will not be disclosed by the original search term suggested on your web site.  Accordingly, users need to insert the asterisk wild card before “Rybka” (as well as after it).
>


A good advice, as this is what I encountered yesterday and was shocked to see search returned no results! :-)
Parent - - By Mark Eldridge (****) [gb] Date 2011-03-13 01:00
Can someone help me please? I cannot download the below link what am i doing wrong? I have spent hours trying but no luck.

I could not place it on my c drive it says i need administrator rights, even though i am the administrator.
Parent - - By Uly (Gold) [mx] Date 2011-03-13 01:18
What happens when you right-click the link and Save As...?

http://rybkaforum.net/mwf/rybkaattach/68/320768/Rybka41Update.exe
Parent - - By Mark Eldridge (****) [gb] Date 2011-03-13 01:27
The green downloading line last for only 2-3 seconds then goes.
Parent - - By Uly (Gold) [mx] Date 2011-03-13 02:42
I blame the browser.

Try installing a third-party solution, such as Free download manager (that's what I use to download my files)

http://www.freedownloadmanager.org/download.htm
Parent - By Mark Eldridge (****) [gb] Date 2011-03-13 14:42
Ok i have now downloaded freedownloadmanager. What do i do next?
Parent - - By Mark Eldridge (****) [gb] Date 2011-03-13 14:53
Still will not download. What am i doing wrong?
Parent - - By keoki010 (Silver) [us] Date 2011-03-13 15:05
Try turning off UAC temporarily. It should let you download it then. Also try a different directory. ie.  C:\user\mark\etc. etc.
Parent - - By Mark Eldridge (****) [gb] Date 2011-03-13 15:57
Sorry but what is UAC?
Parent - - By Vempele (Silver) [fi] Date 2011-03-13 15:58
User Account Control.
Parent - - By Mark Eldridge (****) [gb] Date 2011-03-13 16:29
How do you turn that off?
Parent - - By Vempele (Silver) [fi] Date 2011-03-13 16:31
Control Panel\All Control Panel Items\User Accounts

Change User Account Control settings.
Parent - - By Mark Eldridge (****) [gb] Date 2011-03-13 16:39
You cannot turn it off in windows 7, it has a sliding scale form always notify to never notify.
Parent - - By Vempele (Silver) [fi] Date 2011-03-13 16:46
Never notify = UAC off.
Parent - By Mark Eldridge (****) [gb] Date 2011-03-13 17:04
Ok will give it a try.
Parent - By Mark Eldridge (****) [gb] Date 2011-03-13 19:06
Did it and it still will not download it.
Parent - - By Harvey Williamson (*****) Date 2011-03-15 22:14 Edited 2011-03-15 22:18
Although the 'stall' bug seems much improved I hit it big time today:

r2q1r2/3b1pbk/1p1p1n1p/1Pp1pPp1/3nP1P1/2NP4/2P1N1BP/1RBQ1R1K b - - 0 1


Analysis by Deep Rybka 4.1 SSE42 x64:

19...Ra5 20.h4 gxh4 21.g5 hxg5 22.Bxg5 Rh8 23.Nxd4 cxd4 24.Nd5 Rxb5 25.Rxb5 Bxb5 26.Rg1 Kg8 27.Bf3 Kf8 28.Qb1 Bc6 29.Nxb6 Bh6 30.Bxh6+ Rxh6 31.Nc4 Ne8 32.Qb4 Qe7 33.Na5 Bd7 34.Qb7 Rh8
  =  (-0.18)   Depth: 25   00:44:33 1387mN, tb=2
19...Ra5 20.h4 gxh4 21.g5 hxg5 22.Bxg5 Rh8 23.Nxd4 cxd4 24.Nd5 Rxb5 25.Rxb5 Bxb5 26.Rg1 Kg8[] 27.Bf3 Kf8[] 28.Qb1 Bc6 29.Nxb6 Bh6 30.Bxf6 Qxf6 31.Nc4 Qd8 32.Qb4 Rg8 33.Nxe5 Rxg1+ 34.Kxg1 Qg5+
  =  (0.00)   Depth: 26   10:13:57  19431mN, tb=254

(Williamson, Microsoft 15.03.2011)
Parent - - By Banned for Life (Gold) Date 2011-03-15 23:44
Taking ten minutes on fast hardware to resolve a fail low at depth 26 in a complex, closed position might be considered an example of an exploding branching factor, but this isn't the stalling bug...
Parent - - By Harvey Williamson (*****) Date 2011-03-16 07:16
it was 10 hours not minutes. The move that took all the time was not the pv it was the 2nd move in the iteration.
Parent - - By Uly (Gold) [mx] Date 2011-03-16 07:48
His comment is still correct:

>depth 26 in a complex, closed position might be considered an example of an exploding branching factor, but this isn't the stalling bug...

Parent - - By Dragon Mist (****) [hr] Date 2011-03-16 09:55
I understand Harvey, and second his "complaint". Although I did not encounter such drastic exploding branching factor with 4.1 YET, I've had this with R4 many times. It may be that R4.1 is better in this regard, but Harveys example leaves no doubt, it is still there. As for how we gonna call it, the stall bug or exploding branching factor, it is irrelevant, we don't want it, and we feel it is not right. The argument that the stall bug NEVER returns anything is IMHO purely academic; it may be that people that claim this just didn't have the nerve to wait for the new iterration to come up, as Harvey obviously did. In a nuttshell, we IA people feel this is a bug, and whether R4.1 returns anything after 50 hours or not is of less importance. I personaly don't have the time to test the problem in detail; it is so (proc) time consuming. What I know is that with R4, and on fail low situation (as I reported ages ago), IA on serious depths might take as much as 50 or 100 branching factor to report anything new (note: it could be that this happens also in lower depths, but didn't seem important to me). This was supposed to be avoided with "always score .." and "display upperbounds .." stuff, but my findings are (at least with R4) that leaving those 2 unchecked significantly reduces the occurence of the problem/the branching factor in these situations.
Parent - - By Uly (Gold) [mx] Date 2011-03-16 10:36

> it may be that people that claim this just didn't have the nerve to wait for the new iterration to come up


No, it was debugged with the Visualizer, and when the stall bug hit the search just dies, Rybka is using 100% of the CPU but doing nothing, she dies other than receiving further commands. Basically, with the Visualizer you could tell the difference between an exploding branching factor and the stall bug because in the former you can see what is Rybka doing all that time, and in the latter you can see that Rybka is doing nothing.

So, here Harvey could have checked with the Visualizer what was Rybka doing on those 10 hours, it could have been the unfixed wrong move ordering bug where Rybka randomly (you never know when she'll do it) orders some random move (you'll never know what move or where) first (in some deep branch) and has to resolve it first before ordering the right move, and it could easily take 10 hours to do so (unless the bug was fixed, I'm just saying the report of what happened is incomplete, we just know is not the stall bug because Rybka eventually came back).

>In a nuttshell, we IA people feel this is a bug,


Exploding branching factor is not a bug, it would be a bug if with the Vizualizer the user notices Rybka is doing it wrong. There have been reports of those branching factors since Rybka 2 or before, Vas said they're normal (in fact, the stall bug was overlooked in Rybka 4 because he thought it was exploding branching factor).

>I personaly don't have the time to test the problem in detail


Nobody has, that's why it could be some bug in there that causes the exploding branching factor, but we don't know unless we look at Vizualizer output.

It's possible the exploding branching factor is not the problem, but a symptom of an underlying bug. You'd need to check the Visualizer to see what Rybka is doing when that happens for any abnormality.
Parent - - By tomgdrums (****) Date 2011-03-16 14:50

> Basically, with the Visualizer you could tell the difference between an exploding branching factor and the stall bug because in the former you can see what is Rybka doing all that time, and in the latter you can see that Rybka is doing nothing.
>
> So, here Harvey could have checked with the Visualizer what was Rybka doing on those 10 hours, it could have been the unfixed wrong move ordering bug where Rybka randomly (you never know when she'll do it) orders some random move (you'll never know what move or where) first (in some deep branch) and has to resolve it first before ordering the right move, and it could easily take 10 hours to do so (unless the bug was fixed, I'm just saying the report of what happened is incomplete, we just know is not the stall bug because Rybka eventually came back).
>


Vas himself said the visualizer was untested and somewhat buggy.  He also recommended that most users not use it, and that it does support multi-pv.  And why should the users have to check the visualizer to see if Rybka is working or not?  Isn't that what the bug fix was for?
Parent - - By Banned for Life (Gold) Date 2011-03-16 15:29
- The stall bug was fixed in 4.1, or at least it has not been seen by a credible witness.
- The visualizer is a useful tool, but it is in alpha form. It is useful for people who are willing to put up with bugs in order to get more information out of the engine.
Parent - - By Uly (Gold) [mx] Date 2011-03-17 01:48
What Visualizer bugs (just lack of MultiPV support?)? I think (and I could be wrong) Vas is spreading misinformation if he says it has bugs. The Visualizer works just fine, but it's aimed to advanced users.
Parent - - By Banned for Life (Gold) Date 2011-03-17 02:23
You're right. Bugs was not what I was looking for here. User hostile would probably be a better description at this point (and I'm not complaining).
Parent - By Uly (Gold) [mx] Date 2011-03-17 03:18

> User hostile


Well, Aquarium is like that and many people use it, I don't see why wouldn't they use the Visualizer (or at least try it, maybe they discover that the extra information is of no use to them).
Parent - By Dragon Mist (****) [hr] Date 2011-03-16 15:38

> No, it was debugged with the Visualizer


You are right on all accounts, I totally forgot about the Visualizer.
Parent - By Uly (Gold) [mx] Date 2011-03-16 01:56
The stall bug was about the engine never coming back to live. That is, you get

19...Ra5 20.h4 gxh4 21.g5 hxg5 22.Bxg5 Rh8 23.Nxd4 cxd4 24.Nd5 Rxb5 25.Rxb5 Bxb5 26.Rg1 Kg8 27.Bf3 Kf8 28.Qb1 Bc6 29.Nxb6 Bh6 30.Bxh6+ Rxh6 31.Nc4 Ne8 32.Qb4 Qe7 33.Na5 Bd7 34.Qb7 Rh8
  =  (-0.18)   Depth: 25   00:44:33 1387mN, tb=2

Then you get nothing, ever. THAT was fixed.

In this specific case, if it was a fail low, you have to switch on "Show upperbounds" in engine settings, then, if you see a fail low, you know it could take 10 hours or more to resolve, so what you do it forcing 19...Ra5, and reach depth 25 (which is equivalent to depth 26 at the root), it'll turn it into a fail high from the white side and it'll be solved in reasonable time (if you're lucky even relative root depth 27 could be reached in another 50 minutes).
- - By Vempele (Silver) [fi] Date 2011-03-07 10:13
:eek: There was more interest in this thread when all it said was "TBA". :grin:
Parent - - By bnc (***) Date 2011-03-07 10:22
Its easy to miss, apparently it was marked already read/old when it was posted :mad:
Parent - By Vempele (Silver) [fi] Date 2011-03-07 10:24
For beta testers. Others hadn't seen the second post before.
Parent - - By siam (**) [nl] Date 2011-03-07 11:42
Well wait till the USA is awake, there will be more intrest.... and the European are at work right now:grin:
Parent - - By LordSirKnight [us] Date 2011-03-07 14:35
None of the binary's work for me ... I got my rybka 4 from chess ok but, it does not work.  I dont have a processor for the SSE version and the normal .exe dont work for 32 or 64.
Parent - - By LordSirKnight [us] Date 2011-03-07 14:38
All is well  now ... i got things to work.
Parent - - By Vasik Rajlich (Silver) [pl] Date 2011-03-07 18:08
If it's not too much trouble, can you briefly describe the problem and solution? (I got a support email about the same thing.)

Vas
Parent - - By irulats (****) [ie] Date 2011-03-07 18:35

> If it's not too much trouble, can you briefly describe the problem and solution?


I think the problem might be what happened to me. I directed the updater to the Rybka folder in C:\Aquarium\Engines

However the Rybka engines are actually in subfolders of this folder eg C:\Aquarium\Engines\Rybka\1-cpu\x64 or

C:\Aquarium\Engines\Rybka\Multi-cpu\x64

The updater should perhaps be capable of searching down the subfolders. At any rate, when it's pointed to the right folder it works okay..  :smile::smile:
Up Topic Rybka Support & Discussion / Rybka Support / Rybka 4.1 Release Notes
1 2 3 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill