Author: Tom Likens
Date: 10:20:58 06/05/04
Go up one level in this thread
On June 05, 2004 at 10:13:53, Uri Blass wrote: >On June 05, 2004 at 09:54:55, Marc wrote: > >>Maybe a bit off topic here ... >> >>Of course it is nice to watch a much too strong engine to wipe you of the board, >>over and over again. But eventually this gets a bit boring. >>So I wonder, has anybody programmed an engine which is not really strong, but >>fun to play against? (about ELO 1700-1900) >> >>Crippling a strong engine is somewhat dissatisfying, for some reason. > >The problem is not in the engines but in the hardware that you have. >It is too fast. > >It is only the hardware that make the impression that engines are better than >1900. > >You should ask for hardware that is 10000 times slower than the hardware that >you have. > >The real smart people are not the programmers but the people who build hardware >that is faster every year and I have no idea how they do it. > >The fact that it seems to me that most of the progress in the last 30 years were >done thanks to better hardware and not thanks to better software suggest that >we(the programmers) are relatively stupid. > >Uri Thank you Uri, As one of those "hardware" guys I accept your accolades ;-) Seriously though, hardware people are no smarter or dumber than the software types, it's just a different problem with a different set of rules and solutions. We hardware types owe an awful lot to the process guys who keep making the chips and ASICs faster every year (although that's likely to slow soon). Designing a modern microprocessor involves a *huge* number of people and the design effort is measured in *man-years* (i.e. hundreds of man-years). Also individuals are specialists and part of large teams. One group designs the floating-point unit, while another works on the MMU (the memory-management unit), an yet another group works on the ALU (which itself is brokened down into multiple sections which are farmed out). And let's not forget the extraordinary amount of money it takes to create one of these silicon beauties, multiple millions of dollars. Money to pay the engineers, money for the non-recurring engineering costs (NRE), money for the masks (500k to 1M+) etc. etc. Chess software has not had this kind of effort expended on it, because the monetary reward has not been there. Deep Blue was the only project that went down this road (and no, I'm not looking to turn this into a Deep Blue thread. I think more than enough virtual ink has been spilled over that can-of-worms). I do agree though, that a large measure of the strength of the programs these days is directly proportional to the increases in hardware speed. But I disagree that programmers are "relatively" stupid, since I'm *also* a programmer. regards, --tom
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.