Author: Gregor Overney
Date: 20:47:12 08/31/99
Go up one level in this thread
On August 31, 1999 at 22:10:04, Andrew Slough wrote: > >I think that's supposed to be 100-200% faster isn't it? I think you're right to >be suspicious of benchmarketing though. I think the 100-200% comes from the G4 >"Velocity Engine", which (AFAIK) is just SIMD single precision floating point. >They claim a peak execution rate of 4 Gigaflops, compared to a 500Mhz P3's 2 >Gigaflops. None of this is of any use for chess programs (or real world apps) >though. > If one considers speech recognition applications and video I/O applications as "none real world apps" you might have a point. Unfortunately, the G4's much better FFT cabalilities make it much more suitable than the PIII for exactly almost everything that deals with digital signal conditioning. However, one might argue that DSP's are even faster doing 32-bit "digital stuff". So why use a general purpose CPU instead of a DSP for this kind of application? - It reduces the number of chips in your system; makes PCs cheaper. The introduction of the G4 is mainly a cost reduction thing and nothing more. You pointed already out that the G4 does not really provide a factor 2.94 in speed improvement compared to a PIII/600 when running a chess engine. This is correct, unless someone is using FFT as a tool for her/his new chess engine ;-) Gregor
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.