Re: Another New Hardware Thought

[Charles Shannon Hendrix <shannon@news.widomaker.com>]

On 2007-05-04, Michael J. Mahon <mjmahon@aol.com> wrote:

> I agree--though it was also quite common to be a Computer Science grad
> 20 years ago and to know nothing about the inner workings of computers.

Not in my school... we called those guys "IT Majors".

I had to take computer architecture to graduate.

I suppose I could have found a way around it, but they made it
difficult, and I didn't want to anyway.

> In many schools hardware was considered a necessary evil, and software
> was a branch of mathematics!

Well, that's how it started out years ago. There was no computer science
degree program until years after WWII.

> But the real obstacles to complete understanding modern systems are
> 1) their inconceivable complexity (yes, even their designers don't
> *really* understand their complete workings), and 2) their vanishing
> physicality--there's nothing there but a sliver of silicon, WITH OVER
> 100 MILLION TRANSISTORS, SWITCHING OVER A BILLION TIMES PER SECOND!

Yes.  No test points, and it is hard for you to make modifications to a
system.

Even just creating an I/O card these days is quite hard, and expensive
too.  I suppose once you learn things like the PCI/PCIe glue needed 
it isn't too bad, but that stuff is expensive and has a high learning
curve.

Another issue is that modern busses are fast, but not that great in
other ways.  PCI downright sucks.

PCIe solves some problems, but has such high latency that a lot of
manufacturers are having a really hard time making a decent sound card
for it.

Even the big players are having trouble. Good luck with your hobby
project.

> They could experience its behavior as a "black box", they could "take
> off the top" and examine and probe the insides, add and remove cards,
> and even "construct" virtual cards to plug into it (for extra credit).

True, but there really is nothing like the real hardware.

Systems that can be expanded or modified with discrete logic give you a
physical understanding, and it is also a lot cheaper.

Yes, you can create a chip mask and send it off to be built, but you
have to be wealthy to afford that.

About all you can do at the hobby level is FPGAs, and that's pretty
expensive, and difficult.

One of my goals in life is to build my own computer.

I am trying to think of a way to do it in software, hopefully some that
can save hardware designs to let you build it for real when you are
done.


-- 
shannon "AT" widomaker.com -- ["There are nowadays professors of
philosophy, but not philosophers." ]