Author: Robert Hyatt
Date: 08:12:37 03/29/01
Go up one level in this thread
On March 28, 2001 at 16:29:22, Rajen Gupta wrote: >No one knows as yet > >rajen gupta The question is all about latency. Quite often we see effective memory speeds increase by simply fetching more data at one time. But in doing so, it is also common for the latency to slip a bit. For programs that are doing regular-pattern memory accessing, this trade-off is usually a big win. But for programs (notably chess programs) with more random access patterns, the burst speed isn't as important as the raw latency is. And there, rambus seems to be hurting. However, as a comparison point, my quads all use plain old EDO ram with 4-way interleaving to ramp up the effective speed. They are fast as hell...
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.