Author: Albert Silver
Date: 11:00:40 02/22/02
Go up one level in this thread
On February 22, 2002 at 13:59:28, Paul Doire wrote:
>On February 22, 2002 at 13:56:43, Albert Silver wrote:
>
>>On February 22, 2002 at 13:50:20, Paul Doire wrote:
>>
>>>On February 22, 2002 at 13:46:00, Albert Silver wrote:
>>>
>>>>On February 22, 2002 at 13:24:30, Paul Doire wrote:
>>>>
>>>>>On cpu equivalent of 6.4Ghz (4x stronger than match cpu).
>>>>>
>>>>>Does that mean it could have gotten 2820 performance rating ?(based on whenever
>>>>>processor speed doubles programs increase in strength 50-70elo)
>>>>
>>>>Not necessarily. The performance is based on its results against humans and not
>>>>automatically adding points in proportion to the speed. Suppose it still weren't
>>>>able to do any better than 2-2 against Van Wely on such hardware? You see, that
>>>>extra ply might not mean it will suddenly beat Van Wely in 40/2h.
>>>
>>>
>>>What will it take to win against humans 75% of the time?
>>
>>
>>If the human is rated 2700 then the combination of hardware and software would
>>have to perform at 2900. Naturally, better hardware and better software will
>>yield results, but it isn't so simple. Take today's game for example. How many
>>moves would Rebel play differently if allowed to calculate an extra ply? How
>>decisive are those moves? Just adding a ply shouldn't derail Van Wely so easily.
>
>
>Are you saying that quadrupling the cpu speed will only add one ply to the
>search
>at the same time controls???
More or less, yes. Maybe a ply and a half.
Albert
>
>Paul
>
>
>
>>>>>Will we have off the shelf world champion caliber silicon beasts by 2005 ??
>>>>>
>>>>>What do you think?
>>>>>
>>>>>Paul
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.