Project Kiwi - an 68k Homebrew Computer

[Linde]

So...an MC68008 is actually an even crappier version of the 68000.  Sign me up :rolleyes:

Yes, why didn't he put an Intel Core i7 on it? I mean, most things are crappier. :rolleyes:

[Linde]

If speed was really such a concern to any of us, why would we be using Amigas at all?

[Linde]

I don't think people are concerned the device will be slow, that's a given. The problem is that the 8bit bus will artificially limit the CPU for no really good reason.

How is that an artificial limitation? The bus is a very tangible and real limitation for any CPU, boo-hoo.

Tell me, if they used a 68000, what would suffice as a "good reason" for that choice?

[Linde]

There are no 68000 instructions that are only one byte long. They are all a multiple of 2 bytes.
So a byte wide bus means you need to use at least 2 extra cycles per instruction on a 68008 compared to a 68000. If it makes your machine easier to construct then by all means go for it, but it is like tieing one hand on your back.

I know that, but it still doesn't answer my question.

[Linde]

I said "artificial limit" because the 68000 works best with a 16bit bus.

What does that have to do with the 68008? The 68008 works best with an 8-bit bus. There is no artificial limit imposed.

Seriously, the thing runs a line number BASIC editor, has a VDP from the early 90s, and a couple of buggy sound chips from the 80s, and the first thing you can think of as "artificially limited" is the CPU? It's a dated and inefficient design, but that's obviously the whole point of the project. If he wanted to build a computer without "crippling performance issues" he wouldn't have used any CPU from the 68k series.

So I think you should all stop whining about a few extra cycles spent on memory reads and realize that in terms of performance, the 68000 and the 68008 are two drops in the bathtub you get when if buy a decent cell phone. Whatever argument you are making, it could as well escalate into "why not a 68020", "why not a Coldfire", "why not an ARM" or "why not Intel". But it doesn't, because you have no sense of scale.

[Linde]

I know it may make you feel very clever to compare one-off hobby projects or any non-cellphone platform that contains an ARM chip to a cell phone, but it just comes across as ignorant.

How so? If you're bent on computational performance, you have no business looking at 80s CPUs. The comparison is completely valid from a performance perspective.

A smartphone is not a general purpose computer, even if its quad core and running at 1.6Ghz+.

Not calling a smartphone a general purpose computer shows your complete ignorance of the meaning of the term. I might have agreed if you said that they're not "modular computers" or "open computers" but they are very much general purpose.

[Linde]

Its a spurious and silly comparison.  "OMG it can't outperform teh Droid phonez!" is not what the critique was about.  A standard 68000 or 68010 has 64 pins - that's hardly an unmanageable amount of pins.  If 8 bit is all that is wanted, fine - the 68008 is perfectly fine.

I'm not saying "OMG it can't outperform teh Droid phonez!". I'm saying that dismissing what is so obviously a hobbyist system not meant to compete in terms of performance in any way whatsoever on the basis of bad performance compared to a slightly faster stone-age CPU is totally ridiculous. Just as ridiculous as bringing ARMs to the table.

And of course he wanted an 8-bit bus. Would he blindly pick a CPU and build a working system around it not knowing its bus width?

This forum and amigans in general are fans of making retarded cellphone comparisons whenever performance or architecture things are brought up (rpi, odroid, ouya, etc etc), as if I can take my cellphone and easily install any operating system I want, utilize the bus to drive additional hardware, install keyboards and mice and larger screens easily, install whatever software I want easily, compile on it, etc.  They are not general purpose, they are radio receivers that also allow you some compute functionality.

General purpose computers existed long before computer keyboards, mice, large screens and compilers. Those things are totally irrelevant to the concept of general purpose computers, but sure, I get what you are trying to say. You just aren't familiar with the terminology.

[Linde]

Sure, they are called brains.  Hell, throw in the abacus, the slide rule and  Babbage's Difference engine.  You are just being pedantic and argumentative.  So, try not throwing out cellphone analogies (they don't fit), blindly jumping from single core CISC chips lacking mmu/fpu/pipelining/superscalar architectures (at least in the 68000 area) to quad core RISC architectures to prove some lame point about "power", and making non-relevant asides about fictitious "concepts" of "general purpose" computers... unless you have a link to an academic paper backing your snobbery up, that is - in which case I would be more than happy to read it.

Haha, do you doubt the fact that general purpose computing is a concept? Maybe you could link to an academic paper supporting your idea that cellphone analogies don't fit. Maybe that idea just based on a totally arbitrary threshold that only exists in your head.

But hey let's talk single core CISC. Pentium III?

[Linde]

I'd love to try that Hitachi CPU!