Author: Vincent Diepeveen
Date: 12:57:38 08/18/00
Go up one level in this thread
On August 18, 2000 at 13:38:28, Dann Corbit wrote: >On August 18, 2000 at 07:20:20, Vincent Diepeveen wrote: >[snip] >>First of all a 64 cpu machine is like, let's guess: 50 million dollars? > >5 million, give or take a couple million. > >>So also in 9 years of time we can't afford that. >> >>Secondly a 64 processor alpha is perhaps not having shared memory, so getting >>a good speedup is real tough then. > >It's SMP. is it SHARED or non shared memory? just that the processors are symmetrical multiprocessing is a bit little info for me! >>But i think bigger hashtables are giving your goal quicker as you think. > >I was figuring 4 Gig ram per CPU. The memory bus on that machine is >astonishing, also. aha so non shared. what is the latency of a message? i calculated that a cluster with a 1usec latency of a message is giving a speedup of less as root square. Obviously some techniques might be invented to get a better speedup at clusters. I prefer a 4 processor at 10 Ghz in 2010 however over a 256 processor of each 1 Ghz at a cluster.
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.