Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Some Crafty 16.19 results on my XP 2.44GHz

Author: Matt Taylor

Date: 09:39:34 02/24/03

Go up one level in this thread


On February 23, 2003 at 21:30:46, Robert Hyatt wrote:

>On February 23, 2003 at 01:54:06, Jeremiah Penery wrote:
>
>>On February 23, 2003 at 00:55:00, Robert Hyatt wrote:
>>
>>>On February 22, 2003 at 19:40:44, Jeremiah Penery wrote:
>>>
>>>>On February 22, 2003 at 17:40:10, Robert Hyatt wrote:
>>>>
>>>>>Wouldn't argue.  And I'd bet it would not fail a single time either.  Until
>>>>>you push the clock beyond what the engineers set the limit at.
>>>>
>>>>Tell that to the people who ever bought a P3 1.13GHz processor. :)
>>>
>>>
>>>So? The original pentium had a horrible FP bug.  That happens.  Care to check
>>>the AMD errata sheets?  They do it to.  As did even the Crays...
>>
>>So, the Intel engineers pushed the clock beyond the limit.
>
>No they didn't.  They simply made an error in computing how fast it would
>run.  Just like the FP divide error where a table was copied but one entry
>was omitted...  That's an error in engineering, not in trying to push the
>chip to the edge and beyond, IMHO.  I'd bet they were _surprised_ when the
>failure reports came in, and they found what was causing the problem quite
>quickly, whether it was a slower gate or a longer path, or cross-coupling
>that was unexpected, how knows.  Even Hsu ran into some of that after he had
>done multiple chess chips.  I don't get too hyper about human errors.  Meat
>does make mistakes. :)
>
>>  In essence, they
>>overclocked it.  You seem to think it's ok for Intel to do it, but that anyone
>>else who does it is risking catastrophic meltdown every time they turn on their
>>machine.
>
>If you think Intel produced the chips, then started cranking up the clock to
>see how fast it would go, you are mistaken.  How do they know _now_ how fast
>the next generation will run???  The answers are found in electrical
>engineering.  And they can be wrong.  Bridges _have_ fallen.  Buildings _have_
>blown over.  Planes have lost wings.  Shuttles have lost tiles.  And none of
>it was caused by trial and error.  Just a mistake here and there.  Which is a
>big difference between using the I-beam dimensions given in an engineering text
>for a building X feet tall,  but building the thing X+N feet tall and hoping it
>works.  That doesn't happen.

Intel engineers know how fast a chip is going to run when they crank out the
design. About 3 years ago my Dad mentioned to me that the Pentium 4 would clock
to 5 GHz. I do not know where he heard about it, but I recently read the same
thing in the Intel roadmap -- they will be near 5 GHz by the end of this year
with the Pentium 4 core.

I can't say anything regarding the 1.13 GHz Pentium 3 mistake, but many mistakes
in other engineering disciplines have been marketting/management mistakes rather
than engineering mistakes.

-Matt



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.