To SF team / Jim Ablett: SF 2.11 crash during first game...

Discussion of anything and everything relating to chess playing software and machines.

Moderators: Harvey Williamson, Dann Corbit, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Martin Thoresen
Posts: 1833
Joined: Wed Jun 21, 2006 10:07 pm

To SF team / Jim Ablett: SF 2.11 crash during first game...

Post by Martin Thoresen » Wed May 18, 2011 10:44 pm

... of TCEC stage 2a.
Faulting application name: stockfish-211-64-ja.exe, version: 0.0.0.0, time stamp: 0x4dc7b844
Faulting module name: stockfish-211-64-ja.exe, version: 0.0.0.0, time stamp: 0x4dc7b844
Exception code: 0xc0000005
Fault offset: 0x0000000000023be7
Faulting process id: 0x908
Faulting application start time: 0x01cc159914c1f710
Faulting application path: E:\Engines\Stockfish\2.11\stockfish-211-64-ja.exe
Faulting module path: E:\Engines\Stockfish\2.11\stockfish-211-64-ja.exe
Report Id: 50f46f54-819e-11e0-b010-8188a0363a44
http://home.halden.net/mordor/sf_crash.jpg

The RAM has passed 24 hours of Memtest86+ and Memtest for Windows.

Any idea?

Best,
Martin

mhalstern
Posts: 484
Joined: Wed Nov 18, 2009 12:09 am

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by mhalstern » Thu May 19, 2011 12:19 am

I've played around 30 games with the new stockfish on my overclocked Quad without any crashes.

bob
Posts: 20920
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by bob » Thu May 19, 2011 12:31 am

kranium wrote:From Openchess.org
http://www.open-chess.org/viewtopic.php?f=5&t=1156
mcostalba wrote: Anyhow your sentence is interesting because with few words you perfectly synthesize what IMHO is the wrong way to approach engine development, in particular:

Of course everybody uses the approach he prefers, my comment simply says that we use a completely different approach in SF and we are happy with that ! :-)
A suggestion to the Marco and the Stockfish team...
perhaps you should redirect your time to re-evaluating your own flawed testing techniques, instead of dissing other engines/authors.

So many buggy versions, so often followed by urgent maintenance releases, one after another...
sheeesh... IMO this certainly wouldn't (and doesn't) quality you to criticize another's methods!

(unless of course, you have absolutely zero humility..which apparently may be the case?)
Of course I realize how easy it must be to get fat head (when lauded and praised as the Talkchess 'darling').

I do have one question for Marco and the team.
I've seen that Marco rigorously applied C++ templates, etc. and other optimizations to the Glaurung code...
and that Joona proficiently 'tuned' parameters...
but (as Bob has stated and I am confident would agree)...
this is clearly not the same thing as 'writing your own code from scratch.

Therefore, does this really qualify Marco (and Joona) as legitimate Stockfish 'authors'?
If so, it's certainly seems a double standard being applied...(i.e. Fire, and Houdini)?
Norman, it is time to either shut up or suffer the consequences. It is not _that_ uncommon to run into a compiler or O/S issue when something is released. It does _not_ mean "flawed testing" since no non-commercial programmer is going to keep several machines with various O/S types and versions, and then use all the possible compiler flags...

So stop this pro-clone nonsense. Every single thread, you have to pipe up with "rah-rah-rah-sis-boom-bah, cloners, cloners, rah-rah-rah. Give it up. NOW.

User avatar
kranium
Posts: 1909
Joined: Thu May 29, 2008 8:43 am

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by kranium » Thu May 19, 2011 12:38 am

bob wrote:
kranium wrote:From Openchess.org
http://www.open-chess.org/viewtopic.php?f=5&t=1156
mcostalba wrote: Anyhow your sentence is interesting because with few words you perfectly synthesize what IMHO is the wrong way to approach engine development, in particular:

Of course everybody uses the approach he prefers, my comment simply says that we use a completely different approach in SF and we are happy with that ! :-)
A suggestion to the Marco and the Stockfish team...
perhaps you should redirect your time to re-evaluating your own flawed testing techniques, instead of dissing other engines/authors.

So many buggy versions, so often followed by urgent maintenance releases, one after another...
sheeesh... IMO this certainly wouldn't (and doesn't) quality you to criticize another's methods!

(unless of course, you have absolutely zero humility..which apparently may be the case?)
Of course I realize how easy it must be to get fat head (when lauded and praised as the Talkchess 'darling').

I do have one question for Marco and the team.
I've seen that Marco rigorously applied C++ templates, etc. and other optimizations to the Glaurung code...
and that Joona proficiently 'tuned' parameters...
but (as Bob has stated and I am confident would agree)...
this is clearly not the same thing as 'writing your own code from scratch.

Therefore, does this really qualify Marco (and Joona) as legitimate Stockfish 'authors'?
If so, it's certainly seems a double standard being applied...(i.e. Fire, and Houdini)?
Norman, it is time to either shut up or suffer the consequences. It is not _that_ uncommon to run into a compiler or O/S issue when something is released. It does _not_ mean "flawed testing" since no non-commercial programmer is going to keep several machines with various O/S types and versions, and then use all the possible compiler flags...

So stop this pro-clone nonsense. Every single thread, you have to pipe up with "rah-rah-rah-sis-boom-bah, cloners, cloners, rah-rah-rah. Give it up. NOW.
I recognize the threat...reminds me of 'Surrender Dorothy' from the Wizard of Oz.
But, I will do as you say Bob...and shut up, as I realize my opinions must be difficult for the entrenched establishment to tolerate.

I will refrain if it means I might get banned by you again for a year without any due process (i.e. any opportunity to defend myself) as you did in the past.

bob
Posts: 20920
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by bob » Thu May 19, 2011 2:26 am

kranium wrote:
bob wrote:
kranium wrote:From Openchess.org
http://www.open-chess.org/viewtopic.php?f=5&t=1156
mcostalba wrote: Anyhow your sentence is interesting because with few words you perfectly synthesize what IMHO is the wrong way to approach engine development, in particular:

Of course everybody uses the approach he prefers, my comment simply says that we use a completely different approach in SF and we are happy with that ! :-)
A suggestion to the Marco and the Stockfish team...
perhaps you should redirect your time to re-evaluating your own flawed testing techniques, instead of dissing other engines/authors.

So many buggy versions, so often followed by urgent maintenance releases, one after another...
sheeesh... IMO this certainly wouldn't (and doesn't) quality you to criticize another's methods!

(unless of course, you have absolutely zero humility..which apparently may be the case?)
Of course I realize how easy it must be to get fat head (when lauded and praised as the Talkchess 'darling').

I do have one question for Marco and the team.
I've seen that Marco rigorously applied C++ templates, etc. and other optimizations to the Glaurung code...
and that Joona proficiently 'tuned' parameters...
but (as Bob has stated and I am confident would agree)...
this is clearly not the same thing as 'writing your own code from scratch.

Therefore, does this really qualify Marco (and Joona) as legitimate Stockfish 'authors'?
If so, it's certainly seems a double standard being applied...(i.e. Fire, and Houdini)?
Norman, it is time to either shut up or suffer the consequences. It is not _that_ uncommon to run into a compiler or O/S issue when something is released. It does _not_ mean "flawed testing" since no non-commercial programmer is going to keep several machines with various O/S types and versions, and then use all the possible compiler flags...

So stop this pro-clone nonsense. Every single thread, you have to pipe up with "rah-rah-rah-sis-boom-bah, cloners, cloners, rah-rah-rah. Give it up. NOW.
I recognize the threat...reminds me of 'Surrender Dorothy' from the Wizard of Oz.
But, I will do as you say Bob...and shut up, as I realize my opinions must be difficult for the entrenched establishment to tolerate.

I will refrain if it means I might get banned by you again for a year without any due process (i.e. any opportunity to defend myself) as you did in the past.

Nothing is difficult for the entrenched establishment to tolerate. But all you are doing is continuing to beat the same drum, long after the head has split, so that it is just making an irritating noise. However, as far as your last statement goes, I have _never_ banned you. If you were banned while I was a moderator, it was a team decision, not a unilateral one.

We do not ban without warnings. But your continued insults directed at others, without merit or justification, has to stop. A gratuitous insult tossed at the SF guys because they have some sort of issue with a new version is not reasonable. It may well be it is not their problem at all. I've had my share of such bugs where my testing is on linux and windows goes ape-shit for no good reason. It does not represent a lack of testing, necessarily. It might represent a lack of testing on a specific piece of hardware, operating system and compiler. None of us have the time, energy or interest in trying to test on every platform. We offer free programs. Tested as well as we can test 'em. No insults needed nor wanted...

mhalstern
Posts: 484
Joined: Wed Nov 18, 2009 12:09 am

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by mhalstern » Thu May 19, 2011 2:30 am

Norman,

As you are watching this thread, I have a question:

Is Fire 1.31 stronger than FireBird 1.2? - or does it just fix bugs? If so, how many elo stronger is it.

mcostalba
Posts: 2684
Joined: Sat Jun 14, 2008 7:17 pm

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by mcostalba » Thu May 19, 2011 6:07 am

Martin Thoresen wrote: Any idea?
No, sorry. These random and not reproducible crashes are almost impossible to track down (at least for me). Here the problem is that are not reproducible, I can run thousand of games on my QUAD without experiencing one....

User avatar
Dr.Wael Deeb
Posts: 9772
Joined: Wed Mar 08, 2006 7:44 pm
Location: Amman,Jordan

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by Dr.Wael Deeb » Thu May 19, 2011 7:04 am

mhalstern wrote:Norman,

As you are watching this thread, I have a question:

Is Fire 1.31 stronger than FireBird 1.2? - or does it just fix bugs? If so, how many elo stronger is it.
Fire 1.1,a personality of it and a personality of Fire 1.31 are stronger than Fire 1.31 & fire 1.2 in my rating list....
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….

User avatar
Eelco de Groot
Posts: 4277
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by Eelco de Groot » Thu May 19, 2011 11:38 am

mcostalba wrote:
Martin Thoresen wrote: Any idea?
No, sorry. These random and not reproducible crashes are almost impossible to track down (at least for me). Here the problem is that are not reproducible, I can run thousand of games on my QUAD without experiencing one....
The fact that Stockfish crashed in the first game would suggest it is not a random event though. Maybe it is best to just wait if it happens again. I'm assuming that, since both engines were calculating you could rule out the book reading code? Also, after castling White seemed to be under attack but the score of both engines is 0.00 resp. -0.00. So possibly a repetition of moves, and just maybe there is a problem with accepting TT hits in PV nodes. The error code I believe is a general one. I have had a few of those for different problems. Maybe Martin could get a version that dumps error output in case of a crash? If that is within the TCEC rules and Marco would want to try to track it down. The elo would not be affected very seriously by that I assume. That is all I can think of Martin...

Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

rbarreira
Posts: 900
Joined: Tue Apr 27, 2010 1:48 pm

Re: To SF team / Jim Ablett: SF 2.11 crash during first game

Post by rbarreira » Thu May 19, 2011 12:00 pm

If I understand it correctly TCEC uses an overclocked CPU. That makes the probability of random crashes higher.

Post Reply