On Mon, Nov 21, 2011 at 4:09 PM, Russell Coker <russell@coker.com.au> wrote:
On Mon, 21 Nov 2011, Julien Goodwin <luv-lists@studio442.com.au> wrote:
> I will say that some of this is new algorithms that are only practical
> with large amounts of RAM, some of the simulation stuff we use in Google
> NetOps is like that.

Sure, but then there's the fact that web browsing worked quite nicely on a
system with 16M of RAM in 1995 and a significant part of my web browsing
hasn't changed much since then.


Really?
I can't speak for how you browse the web today, or in 1995, but my habits are vastly changed.

There were no tabbed browsers back then, so I remember a very serialised process of loading a page, reading it, and then following a link.
Doing one thing at a time.
There was no webmail, so I didn't have gmail open, that was done via telnet and pine.
There were no twitter, facebook, youtube, google music, web forums, wikis, and the like.  Those use a lot of ram.

Now, I have, on this computer, 4 Chromium windows open, each one with between 5 and 15 tabs open.
I switch between tabs as I need, and open many tabs in the background then process them, rather than doing things in a serial way.

This has a cost in memory, clearly, but that's why I have machines with 8gb of ram in them.

I think if you're honest with yourself you'll admit that the way you browsed the web in 1995 has absolutely no similarities with how you do it now.

cheers,

     / Brett