Foreword

"What is the best architecture for a CPU ?"

A question, which is often asked by beginners.

It depends on different factors.

It's a compromise between how much time, effort and money you are
willing to spend to gain enough functionality and speed for your project.

Too much functionality may reduce speed, but too less functionality
may limit the fun when trying writing software for your CPU later.

Also, please don't try to forget that you might feel the need to repair
or modify your design somedays.

Another question is availability of the parts...
and if you will be able to still build your design some years in the future.

"Why don't you use modern technology like FPGAs and such ?"

I'm already using FPGAs at work since a while.
(This article was written in 2008).

But if you would be working in a fast food place for years,
you would eat (nearly) anything but burgers at home.

Sure, ICs are getting faster and cheaper with every year, but quality
and the potential for practical use seems to be spiraling downwards.

If you want to pull a brewery waggon, you won't go far with racing horses.

As a hobbyist, I would be more happy with reliable and robust technology.
Reinventing the wheel every now and then for non_technical reasons ain't fun...

But I'm getting off topic.

Now back to "the making" of the 32 Bit TTL RISC CPU.

I really felt tempted to try a transistorised implementation for our cute little
"dinosaur". But the experiments with DECL based flipflops didn't go well, and
it became obvious that I would have to invest some more time, money and effort
into basic research...

While strolling through a city in Germany, I did decide to visit a small
electronics hobby shop. Did have an interesting talk with the owner
of said shop.

"GALs and EEPROMs ? We don't have them on stock, because nobody buys this
stuff. No, we don't have 74573/74574, and we won't order them. In fact, if one
of the 74LS drawers in this shelf goes empty, we don't refill it anymore."

There is a joke that half of the city is employed at the university, while
the rest is employed at the R&D facilities of a very big electronics company.
...go figure.

Anyhow, I did decide to investigate what TTL ICs the store still had on stock.
It looked like some of the ICs were older than 10, 20 years.
(It's been a while, that Motorola did manufacture 74LS parts.)
And I then dug out a treasure:
a handful of ancient 54LS181 from Texas Instruments.
The 54LS181 basically is a 74LS181 ALU with extended temperature range
in a cute caramic package.

So I simply could not resist and did decide to buy them for building
a TTL CPU, while pushing all my transistor plans aside.

Sorry to spoil your expectations...

[HOME] [UP]/ [BACK] [1] [2] [3] [4] [5] [6] [7] [8] [NEXT]