Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The need to unmake move

Author: Robert Hyatt

Date: 11:02:44 09/03/03

Go up one level in this thread


On September 03, 2003 at 13:51:30, Sune Fischer wrote:

>On September 03, 2003 at 13:36:51, Gian-Carlo Pascutto wrote:
>
>>On September 03, 2003 at 13:17:40, Robert Hyatt wrote:
>>
>>>you can split the tree _only_ at the root, and with two processors get a 1.5x
>>>speedup.  This has been done.  I did it in 1983.  Of course, with 4 processors,
>>>the speedup is still around 1.5x.  But you said two.
>>
>>Hahahaahahahahaa!
>>
>>Joke of the month.
>
>Are you refering to Vincent or Bob here?
>
>Going <1.0 on a quad sounds funnier to me. :)
>
>-S.
>
>
>>--
>>GCP


I assume he was talking about going from 2 to 4 giving zero.  :)

However, the example was real.  In 1983 we were given two weeks to get a
parallel search of any kind working.  Actually 10 days.  Splitting at the
root was the only viable option, and some testing gave us a speedup of
1.5X roughly, which was not bad.  But 3 months later Cray had their 4-way
XMP machine running and the speedup was _still_ about 1.5X.  We then went
to the original PVS (not the PVS search we all use today, but PVS as a
parallel search algorithm published by Monty Newborn).  That got us to
2.5X to 3.0X with the average closer to 2.5X.




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.