Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How much really does CPU power mean?

Author: Detlef Pordzik

Date: 17:38:14 08/28/00

Go up one level in this thread


On August 28, 2000 at 16:48:21, Eddie wrote:

>On August 28, 2000 at 16:32:46, Peter Skinner wrote:
>
>>In comparing a PIII 450 vs let's say an Athlon 750, what would the expected game
>>win/loss % be for the PIII 450.
>>
>>For testing purposes, let's say we are using Crafty.
>>
>>Would the 450 get absolutely killed?
>
>Thats a good question for sure ... I'd like an answer myself.
>
>Stefan won it all with an Athlon .... so is the Athlon superior over the Pentium
>in computer chess ?
>
>Experts in this field may help us ...... :))

Let me try it this way :

a) a doubling of processing power allows -at average - 1 ply ( half move )
deeper search under tournament conditions ( 40 / 120 ). And this varies, of
course, from prog to prog.
So it majorly depends, for which purpose one uses his progs / PC combination.
At analysis level the CPU ain't that important anymore - the more the technology
of using the hash tables.
besides :
CPU can't be compared with CPU - lots of technics, Eddie.
To give only one example : the former K6-3/xx isn't a real hit concerning the
speed under today's general views anymore - BUT :
the on die cache is adressed not just with the bus clock speed = 100 MHZ instead
with the full CPU speed - so, for example, 450 MHZ - this is important, as well.

b) SMK won this championship using an ATHLON - last year he used an INTEL - and
won as well. The brand seems somehow over estimated, at least so I'd say. But,
no rule without exeption :
FRITZ is, was and probably allways will be : hand optimized for INTEL codes.....

to sum this up, coming back to one of the questions :
using the very same prog + release vs, same ammount of hash, same HD speed etc.
of course the faster CPU would bring a noticeable benefit under these
conditions.
If you start this row with the other extreme :
Crafty on the fast CPU and a slow searcher like CS TAL on the weaker CPU, the
benefit isn't that clear anymore at all, because these kind of progs ( like
HIARCS, as well ) depend only half as much on top technology. They get their
kicks from the built in knowledge codes......

in the end I fear, this rather troubles more than it cleared up the question -
but so it is :
nothing is easy, in the end :-))

ELVIS



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.