Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: best chess programmers

Author: Robert Hyatt

Date: 19:16:29 07/21/00

Go up one level in this thread


On July 21, 2000 at 19:30:04, Tom Kerrigan wrote:

>On July 21, 2000 at 16:48:29, Robert Hyatt wrote:
>
>>On July 21, 2000 at 15:00:18, Tom Kerrigan wrote:
>>
>>>This thread is going nowhere fast. I don't agree with anything you say, but it
>>>would be a waste of my time to explain why. But here are two quick comments that
>>>will hopefully make you feel incredibly stupid.
>>>-Tom
>>>
>>>
>>>On July 21, 2000 at 10:13:00, Robert Hyatt wrote:
>>>
>>>>Sure I am.  Should I tell you when the first SMP PC came along?  Of course,
>>>
>>>Yes, please do tell me when the first SMP PC came along. Because I distinctly
>>>remember reading computer magazines when I was ~10 years old, and they
>>>advertised SMP 386 computers. And that was a long, long time before Crafty.
>>>
>>>>>Just face it, you didn't invent SMP PC chess programs. If you think I'd believe
>>>>>that crap, you might as well start telling me that you invented the Internet.
>>>>I never said I did.  I did write the first shared memory parallel search that
>>>
>>>Har. Actually, you did say it. Direct quote:
>>>"With plenty of innovations from rotated bitmaps to being
>>>the first SMP (PC) program running..."
>>>
>>>Or did you mean that other SMP PC programs didn't run? They all seg faulted?
>>>
>>>-Tom
>>
>>
>>Name the SMP PC program.  I've never heard of one.  To date, I know of the
>>following:  (in order of their creation I think):
>
>I already named Zugzwang. True, Zug is a message passing program, but that
>doesn't contradict the fact that it was running on an SMP computer and utilizing
>all the processors. If by SMP program you really mean, "utilizes all the
>processors the way Bob Hyatt thinks they should be utilized," then perhaps you
>should start saying that.

SMP search is a different animal than distributed search, in any book you
care to mention.  Zug doesn't do anything on an SMP machine.  So if you want
to use that as an example, then the first program that could run on a SMP
machine must have been Sargon, because it would definitely run on one, even
if it only used 1 cpu.

If you want to talk about a program that runs on SMP hardware, and _uses_ the
SMP hardware, then Zugzwang doesn't fit.  So do we talk about _real_ SMP-aware
programs, or just any old program that could run on a SMP machine, whether it
used it or not???





>
>>BTW, there is a _big_ difference between writing the first SMP PC program
>>and "inventing SMP PC chess".  If you don't understand the difference, read
>>a book...
>
>I've read many books, some even on computer chess, and I still don't see the
>difference. Do enlighten me.



If you don't see the difference, there is not much I can do.  The computer
chess program was _invented_ _long_ before it was written.  IE it is easier to
take some idea someone else came up with and implement it, than it is to do the
complete design _and_ implementation yourself.  The PVS (principal variation
search) algorithm came from Newborn/Marsland...  I just happened to implement
it on a shared memory machine first...

>
>>There aren't many SMP pc chess programs at all.  There are more message-
>>passing programs around, dating back to the late 1970's...
>
>Again, what is it exactly in a message-passing program that excludes it from
>being Symetrically Multi Processed?


SMP machines have shared memory.  They have atomic locking facilities to provide
control over interleaved updates.  The debugging problems are absolutely 100%
different from the problems seen in debugging a distributed chess engine.  And
those bugs are 500% more difficult to find and fix.  You can share things with
no regard to cost, which means you can share anything on a SMP box, while you
are very careful on a message-passing machine because of message latency and
jitter/lag.

Other than that, they are identical of course. :)





>
>>As far as SMP 386's, I'm not aware of any that were useful, and even a hand full
>
>Great, Bob Hyatt doesn't think they're useful, therefore they don't exist? Good
>call. By the way, you still didn't tell me when SMP PCs came out.
>
>-Tom


Try middle 486 era for some that were sort of useful (2 cpus only, shared cache,
poor performing.)  The first _real_ SMP machines were the pentium pro based
machines, way later in time than the 386.  After the P6, SMP became common.



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.