Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: New crap statement ? Perpetuum mobile

Author: Sune Fischer

Date: 11:23:50 10/03/01

Go up one level in this thread


On October 03, 2001 at 13:17:20, Miguel A. Ballicora wrote:

>>This analogy fails because there are tasks that can be done by two persons and
>>not by one alone.
>
>Exactly, some tasks can be done by two persons and not by one alone. That is
>how the mechanics of the solution changes. That is what Bruce Moreland
>wanted to illustrate. Once you have two persons, you do not have to
>do exactly the same procedure that you do with only one person. It is
>more efficient to change the procedure and take advantage of that. That is
>a synergistic effect. The example was intended to show what a synergistic
>effect is and nothing more. For that purpose, the analogy is ok.

I wonder if not this example was produced in a attempt to make a disproof of
Hyatt's statements.
You say that "you do not have to do exactly the same procedure that you do with
only one person" as though that is also the case for two CPUs. It is not, and
thats why it is a bad analogy. I there is a faster algorithm, it would work
equally well (if not better) in serial.

>>This is not the case for CPUs.
>
>The question is: could be the case for two threads? I say "probably" not,
>Bob Hyatt says "provably" not. That is the point of disagreement.

Yes I understand that, but it is a huge point.

Assume you have an optimal serial algorithm,
and a parallel version of that same algorithm which runs >2 faster on 2 CPUs.

The physics is indisputable: 2 CPUs can at most do twice the work of 1.
Any algorithm that can run on two CPUs can also run on 1 CPU.
Just run the parallel version on the single CPU!


So we have reduced the problem to the following:
There exists a parallel algorithim, which in no way can be rewritten into a
serial one without increased "work".
This is now the point of disagreement, right?

I'm pretty sure I can also prove that wrong, but before I spend time on that, I
would like to hear that we agree so far.
If I have stated something wrong, please correct me.

-S.



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.