Author: Robert Hyatt
Date: 11:11:26 10/22/97
Go up one level in this thread
On October 22, 1997 at 13:20:08, Bruce Moreland wrote: >On October 22, 1997 at 09:55:03, Chris Whittington wrote: > >>I expected a few p2 300's. Don't really approve, but their advantage is >>not going to be 3.5 x as with liquid nitrogen cooled alphas, is it ? > >Good question. I have no idea what the difference is between a K6/233 >and a PII/300. > >I can give you some hard number for Alpha, but not for a 767 mhz Alpha. > >With my program, running an Enorex 500 mhz Alpha (1mb cache), it appears >that the Alpha is 1.5x faster than my P6/200. This is based upon a run >of the LCTI test at 210 seconds per position (98 minutes total). > >Running on a Polywell 533 mhz Alpha (2mb cache), mine goes 1.75x faster >than my P6/200, based upon the same test. > >I have heard that a K6/233 is faster than a P6/200. > >bruce we have K6/233's running on ICC. they are 15-20% faster than my P6/200. so let's call it 1.2X for simplicity. Your 500mhz alpha is 1.5X. I'm going to (most likely) run on a 500mhz alpha. Am I therefore now "within the spirit" or not??? BTW, a PII/300 can easily be 1.5X a P2/300 with the right coding style. So if the PII/300 is "on the same field" then the 500 alpha has to be too... they seem similar. *EXCEPT* that crafty runs like hell on an alpha compared to a P6, because of all the 64bit stuff I use. So how do we compare now? Crafty on a P6 vs Crafty on an Alpha? If so, I want to compare fritz on a P6 to Fritz on an alpha. Wait? Fritz can't do that right? Because of the assembly problem? So my extra 64bit burst of speed shouldn't be counted either in the performance comparison because I am not using assembly language... It is *not* an easy issue to wave off as fair/unfair...
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.