Author: Will Singleton
Date: 15:12:02 12/13/99
Go up one level in this thread
On December 13, 1999 at 17:29:35, Roger wrote: >These green lists are cool. > >I noticed that someone posted that using a PIII coppermine chip gave speedups >for many of the chess programs over just a plain PIII. Plus, a PIII 450 is gonna >give more speed than a PII 450. > >After reading that, it occurred to me that the R/megahertz statistic might be >biased, and that the authors could run some kind of benchmark on their systems >to equalize things out. > >The R/benchmark statistic would probably give a different sorting than >R/megahertz. But what would this benchmark be? > >Roger Right, it's just an approximation. The difficulty is that chess programs will perform differently depending on the implementation (choice of data structs, etc). The compiler makes a difference, too. Even if you ran benchmarks (which was the original reason for the TSCP benchmarking thing), the various programs are sufficiently different from TSCP to cause inaccuracy. So I ended up just using mhz, which is good enough for the purpose. Will
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.