Author: Andrew Slough
Date: 14:59:42 05/07/99
Go up one level in this thread
On May 06, 1999 at 10:33:46, Rajen Gupta wrote: >Obviously the SSDF have made an informed choice in selecting an end of the line >processor (AMD) which will run on an end of the line motherboard but if I were >them , I would have chosen a Celeron 466 processor (which in fact runs >marginally faster than a P ii or Piii 450 but at i/3 the cost) monted on a BX >motherboard with a slotket. The money saved could be better spent either on >future upgrades, more RAM etc. > >Rajen Gupta I don't know anything about the SSDF's decision, but for chess, the K6 line is very nice. Chess code tends to have lots of branches and the shorter pipeline in the K6 combined with the better branch prediction (~95% compared to P6s ~90%) make it a better core. The larger L1 cache (64k compared to P6s 32k) is better for chess than the integrated L2 cache in P6 chips because L2 cache will be thrashed by the hash tables. A Celeron 466 is more expensive (~170$ compared to ~$130 for K6-2 450 [according to www.pricewatch.com]) and only runs at a 66Mhz front side bus speed, which makes it slightly worse for your hash tables. There are some interesting comparisons of chips on the Rebel homepage at http://www.rebel.nl/bench.htm - of course they're only representative of Rebel, not all chess programs. I'm not trying to sell AMD chips, I'm just interested in the architectural benefits of each for a chess program. If you want to play Quake then definitely get a Celeron :-) Andy
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.