Author: Robert Hyatt
Date: 22:09:40 01/29/99
Go up one level in this thread
On January 29, 1999 at 22:47:44, Kim Hvarre wrote: >On January 29, 1999 at 22:15:15, Robert Hyatt wrote: > >>On January 29, 1999 at 13:42:56, KarinsDad wrote: >> >>>On January 29, 1999 at 10:27:58, Peter Fendrich wrote: >>> >>>>On January 29, 1999 at 09:51:36, William H Rogers wrote: >>>> > >>> >>>Why? The Deep Blue software was written in the first place, why can you not port >>>all of the ideas (if not the identical code) from the current PC software? >>> >> >>Simple.. the DB 'chips' are _hardware_. they do an alpha/beta search, they >>evaluate positions, and they are 'cast in stone' so that they are going to >>search the same tree and evaluate the same positions the same way no matter >>what. >> >>As I said before, move a micro to DB, you end up with deep blue and _nothing_ >>else... Because the micro program has to do everything like the hardware >>dictates, has to use the hardware evaluation, etc... > >Hmmm... and once You have Deep Thought it'll be just like the successors; Deep >Blue, Deeper Blue and what ever. Once You have decided what to "hardwire" on >Your dedicated chips, it's done and over?!;)) > >This discussion has been running from time to time various places. The point >isn't about the actual formalism transforming one idear (PC-sw) to another >DB-hw), it's - as mentioned - the likelyhood of doing the "chessstuff" better >than the DB-team, and it is rather big looking at the thin outcome of all that >cabinets of hardware and speed! (In contradition to results from the better >sw-developers). > I don't want to get into a protracted argument. But the 'better software developers' are _already_ working on the deep blue project. _that_ is too often overlooked. They aren't just 'hardware designers' by a long shot. >> >>there is _no_ C compiler for the DB hardware. the chips are vlsi circuits >>and not something that is 'programmable'... > >Right - there are just given as is ... > >> >>exactly the opposite. you take out what you can't afford computationally, >>to keep your tactical speed at an acceptable level. DB has _no_ such problem >>and gives up _nothing_ they want to do, they just designed it into the hardware >>where the cost was _zero_... (speed cost). > >Se above. If they really is able to implement infinite amonts of >"chessknowledge" in hardware, then they ought to, which they obvious did not. A >bit like racing MC's - you can build a superior (regarding speed/moment(um)) MC >and you will still loose to the ones, that are more rigid, stable, better >designed, etc., etc. "obviously they did not" yet no other program has beaten any GM in a 40/2hr time control match? No other program has beaten a 'super-GM' in a 40/2hr match? And the DB guys aren't very good and didn't have 'very much chess knowledge'. We've discussed some interesting data that many don't like but which is public knowledge about DB vs a couple of micro programs where DB was handicapped _severely_ and still blew the micros out completely... So DB is doing _something_ right.. > >So the poll-question is rather sensefull, perhaps with a little refrasing as >e.g.: "if the DB-team have had access to the brilliance of the best >sw-programmers of today, do You then think, they would have come up with a >better result?" > >Yes, is my humble bet. > >kim The DB guys are _far_ sharper than you give 'em credit for. They developed _many_ of the ideas the commercial programs use (ie singular extensions, and so forth came _from_ the DB guys and ended up in chess genius, _not_ the other way around. Ditto for PVS. And other ideas... To make a statement like the above is _really_ an insult to a bunch of guys that are at least as good as _anybody_ in the world with respect to computer chess, and they are probably a lot better than anyone else... At least they have a performance record no one else has produced to date...
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.