Re: [MURG] meets [POLITICS]

From: Robert J. Bradbury (bradbury@aeiveos.com)
Date: Fri Apr 12 2002 - 13:33:11 MDT


On Fri, 12 Apr 2002, Eugen Leitl wrote:

> Current ballpark figures of residental WANs is 40 kBytes/s (and about 15
> ms latency) is sure not GBit rates, but 40 kBytes/s of a tight code
> communication is not something that can be shrugged off.

Actually Eugene, I can get 50-60 kBytes/sec on my DSL line and it
isn't configured for the highest speeds. I think the max rates
may be 2-4 times greater than that (though I'd have to pay more
for it).

I think the install base for DSL is currently 4-6 million homes
and that's certainly going to grow.

A 50+ kB continuous transfer in using the HTTP protocol in Netscape
(not the most efficient) consumes less than 10% of the available CPU
of a dual 200 MHz Pentium Pro machine. The communications bottleneck,
at least for now, is the limiting factor. I don't see that changing
until people get into a mode of video/TV on demand. There isn't
anything else that humans do that would seem to demand higher
bandwidth.

As you point out -- it comes down to a question of how clever
the neural communications protocol designers are as to whether
they can compress significant amounts of useful information
into that bandwidth for offsite delivery.

I'm too lazy to do the calculation right now, but if we
assume a brain is a petaflop machine and has 40 billion
neurons, we ought to be able to figure out what the number
of neurons/PC would be their associated bandwidth requirements
are.

We could stop the handwaving and work with some real data.

Robert



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:13:27 MST