Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Q&A with Feng-Hsiung Hsu

Author: Robert Hyatt

Date: 14:52:46 10/17/02

Go up one level in this thread


On October 17, 2002 at 13:35:57, Vincent Diepeveen wrote:

>On October 16, 2002 at 23:35:10, Robert Hyatt wrote:
>
>>On October 15, 2002 at 14:01:35, Johan Melin wrote:
>>
>>>On October 14, 2002 at 07:34:16, Bas Hamstra wrote:
>>>
>>>>Bob, did you read the Hsu transcript posted here? It is pretty clear to me that
>>>>Hsu himself says 12 ply fullwidth *total*. Case closed. Please read the complete
>>>>transcript.
>>>>
>>>>Best regards,
>>>>Bas.
>>>
>>>I agree. The transcript with Hsu is clear. But it would be out of character for
>>>CCC if everybody just agreed with each other, there still has to be a fight ...
>>>;)
>>>
>>>/Johan Melin
>>
>>
>>Here is the relevant part of the transcript:
>>
>>-----------------------------------------------------------------------------------------------------------------------
>>CrazyBird(DM) kibitzes: 12(6) means 12 plies of brute force (not
>>counting the search extensions & quiescence).
>>CrazyBird(DM) kibitzes: 6 means the maximum hardware search depth
>>allowed.
>>CrazyBird(DM) kibitzes: this means that the PV could be up to 6 plies
>>deeper before quiescence.
>>-----------------------------------------------------------------------------------------------------------------------
>>
>>OK, some questions:
>>
>>1.  If 12(6) means 12 plies total, with 6 done in hardware, how do you reconcile
>>that
>>_last_ sentence above (the PV could be up to 6 plies _deeper_ before
>>quiescence).
>>Deeper than what?  Only possible answer is deeper than 12 plies.
>>
>>2.  If 12(6) means 12 plies total, with 6 in hardware, what does 4(5) mean?  4
>>plies total
>>with 5 done in hardware?  Which means the SP hardware does _no_ searching and
>>that
>>_everything_ (and then some it seems) is done by the hardware chess chips?
>>
>>Neither of those matches what he says above.
>>
>>I read his explanation of 12(6) as "12 plies of brute force, up to 6 more plies
>>of hardware
>>search [which is not brute-force as he has explained elsewhere, because of the
>>forward pruning
>>they were doing]".  I don't see any _other_ way to take the above, which came
>>_directly_ from
>>the answer he gave...  So I won't say 12(6) means 18 plies, but it _clearly_
>>means "more than
>>12 plies, and up to 18 plies"...  At least that has to be true...
>>
>>Otherwise you can't reconcile _anything_ with the log files and the 4(5) type
>>notation...
>
>You are modifying the statements again.
>
>Statement 1 is: it is 12 ply and NOT 18.

I'm not modifying _anything_...


>
>The second statement is independant from the first.
>Later again explicitly gets asked whether it was 12 or 18 in total,
>and the answer was: 12.
>
>Very clear statement. Formulating it like this is not nice of course from
>Hsu. But logically spoken it is a very clear statement. Repeated again
>later in the chat when questions and kibitzes were asking whether it was 18.
>
>Some of them he has seen.
>
>Note that when asked from reinforce(IM) what '6 ply is just the hardware
>partition', a kibitz which he has seen and also later asked by Eek again,
>he didn't go into that.
>
>Any judge here will see clear proof for 12 Bob.
>
>Logically spoken.
>Mathematically spoken.
>
>And of course if it was 18, IBM would have shouted it out loud in 1997
>already. Getting 2 times deeper than any other machine. It would have
>been incredible of course.
>
>But also theoretical impossible. Hsu realizes that very well.

Maybe or maybe not.  His "up to six more plies" is an interesting change in the
definition...  Which needs explanation...



>
>The people who are capable of doing experiments to see whether it is possible
>to get beyond 12 ply much fullwidth with singular extensions and doing all
>other extensions as well, those people all believe it is 12 ply.
>
>There was even produced a crafty version by me, and of course i am willing
>to give the source code of that to any person who wants to see it, so that
>he can verify the changes.

What is the version supposed to do?  It is trivial to disable null-move
completely to see
that it will roughly search 10 plies or so in normal games...


>
>However no one was interested in running it.
>
>Despite that we fight about 18 ply here.
>
>Your own thing fullwidth without hashtable last 6 ply and without nullmove
>and with a limited form of singular extnsions even, it never gets beyond
>12 ply ever.

So?  "my thing" also won't be going over 1M nodes per second or so either.
"my thing" doesn't do _any_ forward pruning.  They claim to do at least a
couple of things, one canning bad captures and the other using the Kaissa
"method of analogies" to eliminate moves...

>
>And it DOES search at 1 million nodes a second *easily*. In fact it gets
>speeds up to 1.5 MLN easily at todays K7s.
>
>So a 15 minute search is already far more than equal to deep blue each move.
>

How do you figure that?  When deep blue is over 100x faster?



>10% efficiency from 126 MLN nodes a second = 12.6 MLN a second.

No way it only searched 12M nodes per second.  That is pure fantasy-land crap...


>
>You get 1 MLN a second with that version easily, and of course DBIIs
>126MLN is because of what happened in the far endgame, the real speed
>the first few moves of a game it got perhaps 30-60 MLN nodes a second
>and from that 10% is just 3-6MLN nodes a second. In short, a short search
>with crafty there is sufficient :)



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.