Author: John Coffey
Date: 10:52:28 01/21/99
Go up one level in this thread
On January 20, 1999 at 19:58:15, William Kerr wrote: >On January 20, 1999 at 17:27:35, John Coffey wrote: > >>On January 20, 1999 at 12:54:01, William Kerr wrote: >> >>>In Feb 1979 I wrote a letter to Fidelity Electronics asking for information >>>on Chess Challenger 10. Here is a synoposis of what they wrote back with: >>> >>>The processor was a Z80A (probability running a 4 Mhz.) >>> >>>The software could analyze 35 chess positions per second. >>> >>>The following table shows how many moves it searched at each level >>> >>>where X = all moves >>> >>> level depth (plys) TIME >>> >>> 1 2 3 4 5 6 >>> >>> 1 X 5 Seconds >>> 2 20 X 15 Seconds >>> 3 16 24 X 35 Seconds >>> 4 20 8 4 1.5 Minutes >>> 5 20 8 4 2 2.5 Minutes >>> 6 X X X X 1 Hour >>> 7 X X X X 24 X 24 Hours >>> 8 20 24 4 4 X 11 Minutes >>> 9 20 12 4 4 X 6 Minutes >>> 10 16 24 4 X 3 Minutes >>> >>>For those of you that never owned any of the Chess Computers back in the >>>70's may find these postings pointless but for those that did, info >>>on old chess computers brings back memories. Though I suppose in 20 years >>>someone will be telling how strong CM6k, Fritz5.32, Craftyxx.x, Rebel10, .... >>>was and the hardware they were run on. >> >>Why would you conclude that it was running at 4mhz? If so, the rate of >>35 moves per second seems very low as some 6502 4 mhz machines searched >>900 movers per second. (The 6502 is about twice as fast per clock cycle as a >>Z80.) >> >> >>John Coffey > >The Z80A was the 4 Mhz version of the Z80. Concidering that the Z80 requires a >conciderable number of clock ticks per instruction, I'm assuming they were >running the micro at 4 Mhz just to get 35 chess positions per second in assembly >language. The letter did state that when faster micros were available they would >use them. To your point about the 6502, for a long time it was the micro of >choice for programming chess. Its instructions seemed almost tailored for >assembly language chess programming. Its instructions required few clock ticks >to execute especially on data stored in the first 256 bytes of ram. > >Bill I believe that almost anybody could write a search engine that would go faster than 35 nodes per second on a 4mhz Z80. Although the Z80 was a slower processor than the 6502, it wasn't that slow. Most equivalent instructions took about twice as long, but some instructions on the Z80 could do more. I conclude the Fidelity models must have relied so heavily on selective search that it slowed the seach.... or else it was just a very inefficient engine. John Coffey
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.