Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Status of Brutus?

Author: Vincent Diepeveen

Date: 10:13:32 07/27/03

Go up one level in this thread


On July 27, 2003 at 11:32:54, Joost Buijs wrote:

>Vincent,
>
>Personally I think the whole Brutus project is a little bit outdated.

I agree.

If Brutus wins this world champs he would have won on a supercomputer with a
software program too!

>Chrilly said in one of his interviews that Brutus on a single FPGA card has
>problems with the top programs on multi processor systems, that's why they are
>developing BrutusP. In my opinion this is just being done to try to win the
>world championchip.

The project is quite cool. I might buy a card under the condition that it is the
SAME like the thing he plays with :)

>If they manage to win this event, it will be nice advertisement and a lot of
>these Brutus cards will be sold. The people won't realise that a single Brutus
>card from let's say $500 will run at the same level as the other top engines on
>moderate hardware.

A good sales organisation can sell it without problems. If they just sell a
small part of what Fritz sells then they will earn tens of millions.

>It is true that more knowledge can be added without additional slowdown, but as
>you said it needs bigger FPGA's so the price will go up.

My guess is that general purpose processors will relatively get faster than
brutus will be in hardware.

>If the clock frequency of the FPGA increases, a faster interface bus is needed.
>Fortunately the PCI bus will be replaced with PCI express in the not so far
>future, this bus runs at 8 GBps, compare this to 133 MBps for the standard PCI
>bus.

Bandwidth is not important. Latency is. I doubt latency from PCI-X is much
faster!

>If Chrilly manages to increase the playing strength of Brutus by adding more
>knowledge, this would be very interesting, but tuning all these parameters will
>be a hell of a job. I wonder if this can be done.
>
>In the mean time the level of the software only engines will go up as well,
>within three years a standard uni processor PC will easely run at 4000-6000
>knps.

I couldn't have put it better than that!

>Groeten,
>
>Joost



This page took 0.01 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.