InvestorsHub Logo
Followers 2
Posts 834
Boards Moderated 0
Alias Born 03/11/2003

Re: CombJelly post# 8151

Monday, 07/07/2003 8:11:22 PM

Monday, July 07, 2003 8:11:22 PM

Post# of 97799
W: "4 dimensional cube or greater"

N: "So you think it'd be better to involve time travel or hyperspace? What 4th dimension would you add to the cluster?"

W: "but the design complexity goes through the roof"

N: "I'll say it would, it even makes me laugh a little."

Combjelly: "He is talking about the topology of the network. The connections between each node is considered as an edge. Look at it this way, a normal 3-d cube has 3 connections per node. If you take the same 8 nodes and have 4 connections per node, it is equivalent to a hypercube, which is a 4 dimensional cube. You don't really need a higher evolved nervous system to view it, but any netowrk analysis you do on it would be equivalent to analyzing a path that follows the edges on a 4 dimensional structure. As wanna notes, keeping this degree of interconnections for very many nodes tends to multiply the number of required connections at a pretty fast rate. To fully interconnect two nodes takes one bidirectional connection per node, 3 nodes needs 2, 4 nodes needs 3, 8 nodes needs 4, etc. While log base 2 of n plus 1 isn't bad if you are doing analysis of algorithms, it sucks in the real world..."

Thanks, mentioning the hypercube label invokes memories of so many nova specials and such where they had goofy glass/tape models to try to explain the hypercube.

I guess I can see why they might use the term 4D when discussing the topology, guess I was just being closeminded or at least too quick to judge.
Volume:
Day Range:
Bid:
Ask:
Last Trade Time:
Total Trades:
  • 1D
  • 1M
  • 3M
  • 6M
  • 1Y
  • 5Y
Recent AMD News