BoatBanter.com - View Single Post

Mr. Luddite

On 4/1/2014 1:28 PM, wrote:
On Tue, 01 Apr 2014 12:43:16 -0400, Wayne.B
wrote:

On Tue, 01 Apr 2014 11:38:15 -0400, wrote:

You are still hitting the wall.
Regular chips are about tapped out.
We are rapidly approaching the point that we will be super cooling
processors to get quantum effects.
There is only so much you can do to shorten the data path.
They are just making them wider. (multiple processors, wider buses)

===

There's talk of stacking vertical substrates also. However the big
future opportunities are in developing better software that can take
advantage of massively parallel processors like IBM's Watson. Those
machines are very esoteric and expensive with today's hardware but
it's only a matter of time before they can stamp them out like jelly
beans. Present software systems have to be highly customized to take
advantage of that kind of power and more generic solutions are needed.
If quantum computers ever become a reality, and they probably will,
they will all will be massively parallel. The possibility of
simulating human thought at blindingly fast speeds is somewhere out
there on the horizon for better or worse, along with instant and
accurate language translation, monitoring millions of security cameras
simultaneously, accurate long range weather forecasting, and a whole
bunch of stuff that hasn't even been thought of yet. Computer
applications are already designing new computer hardware and have been
for some time. What we need now are applications that design and
produce new software.

That was happening to mainframes before I left IBM. We were replacing
giant water cooled systems that would barely fit on a tennis court
with 3 or 4 racks. One was the prossessing array, a rack full of
processors and fiber controllers. Fiber ran to another rack or 2 with
a **** load of RAIDed 3.5" drives (4g each in the mid 90s).
I assume they are 4T each now.

Hook all of that to the internet and you become Google.

Everything fails "soft" and you can hot swap in the new part.

It certainly became clear the hardware business was a dead end job but
I figured that out in the 80s.

Maybe but until we evolve to the point where we are born with USB ports
behind our ears there's going to be a need for a HMI.