Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Problems with Crafty 19.0

Author: Slater Wold

Date: 23:27:24 10/22/02

Go up one level in this thread


On October 22, 2002 at 20:00:23, Robert Hyatt wrote:

>On October 22, 2002 at 18:51:19, Slater Wold wrote:
>
>>On October 22, 2002 at 18:35:25, Dann Corbit wrote:
>>
>>>On October 22, 2002 at 18:28:15, Slater Wold wrote:
>>>
>>>>I have tried to analyze the same position twice now with Crafty 19.0.  Both time
>>>>it has locked, both in the same place.  Coincedence?  Or a bug?
>>>>
>>>>I am using the exe from Hyatt's FTP site.  This output is from the screen:
>>>
>>>[snip]
>>>
>>>>               20   109:00  -2.61   1. d8=Q Nf7+ 2. Ke7 Nxd8 3. Kxd8 Ba5+
>>>>                                    4. Kd7 Bc7 5. Nxe3 Bxg3 6. Kc6 Bf2
>>>>                                    7. Nd5 Bd4 8. Kb7 Nb4 9. Nxb4 cxb4
>>>>                                    10. Bc2+ Kg7 11. Kc6 h5 12. Kd6 (s=7)
>>>>               20   163:45   5/19*  1. Nxe3
>>>>
>>>>
>>>>
>>>>This is after about 18.5 hours of running.  The time is not updating, although
>>>>Windows still reports 100% usage of the CPUs.
>>>
>>>Two distinct possibilities:
>>>
>>>I have seen two effects that both look like that...
>>>If you get a fail high or fail low at that point, it might take a very, very
>>>long time to resolve.
>>
>>I've seen both last forever.  But the time always updates.  Remember, this is
>>almost 1,000 minutes later.  Still it had not updated.
>I have not seen any infinite loops, but it can happen.  If you are running in
>console
>mode, type a "." (period) and it should give you a brief status of which moves
>are being
>searched ply by ply and so forth.  If that doesn't respond, then something is
>definitely
>wrong.  If it does, it might still be hung, so do it multiple times  look at the
>things like
>
>4: 20/41  that means at ply 4, it has searched 20 of 41 moves.  If _none_ are
>changing,
>then something is hosed...  if some are changing, it is making progress...

Cool, I'll try that.  Thanks!

>
>
>
>
>>
>>>Another thing that has happened is that the timer wraps.  However, when that
>>>happens (usually on very long analysis of several days) you will see a nonesense
>>>number in the time like
>>>
>>>               20   991834234163:45   5/19*  1. Nxe3
>>>
>>>That problem is obviously going to be very hard to reproduce.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.