Author: Robert Hyatt
Date: 15:13:03 10/23/97
Go up one level in this thread
On October 23, 1997 at 16:50:45, Chris Whittington wrote: > >Ok, I'll buy that. How about using the average of several programs to >establish this 'chessmark'. > >They'ld (the averaging programs) have to be available for verification >testing. > >They'ld have to be drawn from a pool of varying types, eg not all fast >ones, not all commercial ones. Mixed asm and C. Say Crafty and Ftitz as >fast representatives. Rebel and Genius as medium speed ones, CSTal and >Hiarcs as slow ones. Three use C, three use asm. Use masses of hash >table. > >That's six programs, test them on average on time taken to get to a >particular point on say the BT 2630 test .... > >Chris .asm programs are a problem... they won't run on an alpha, or if they do using the odd DEC tool for that, they run at grossly exaggerated slow speeds... I was thinking of a non-chess benchmark... something that maybe sorts, shuffles memory, randomly probes a large array and so forth. Although several chess programs could certainly be used... But a single contrived thing would work.
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.