Author: Robert Hyatt
Date: 11:41:39 08/21/02
Go up one level in this thread
On August 21, 2002 at 14:28:07, Roy Eassa wrote: >On August 21, 2002 at 14:10:35, Gian-Carlo Pascutto wrote: > >>On August 21, 2002 at 14:06:46, Robert Hyatt wrote: >> >>>DB2 could do over 1B nps _peak_. But it rarely hit that for reasons I have >>>explained in the past, dealing with balancing the speed of the SP2 against the >>>speed of the chess chips. Hsu wrote that it averaged about 200M nps in 1992. >> >>Average: 126M >> >>Peak: 330M >> >>-- >>GCP > > >The 126M/330M numbers are from 1992, five years before the Kasparov match, >right? No, they are "typical" numbers given by Hsu. Theoretical peak was 1050M nodes per second. Easy to compute as I did earlier. Observed peak was a calculation done by keeping up with what percentage of the time each chess processor was busy (measuring its duty-cycle) and then multiplying that by the processor's speed. Sum over all processors to get a fairly accurate NPS value. They found 350M or so was the peak they sustained for a single move in 1997... Which is very fast...
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.