Alternate Processors

[samoz]

Apologies if this is in the wrong forum, please move if it's not.

I'm in an intro to digital system design course at school right now and we're getting to a point where we're developing our own simple "computer." It is very simple, only having around 8 instructions and a 5 bit address space (at least for now). This got me thinking though, it should be simple enough to use a bigger/more powerful processor to implement more instructions and a bigger address space.

I know most people are using x86 type chips for there operating system, but is anyone using any other sorts of chips? Or have you had any experience in the past? It could be an interesting experiment to buy the hardware and setup a computer (and possibly an OS!) on a smaller microprocessor...

Any thoughts? Stories?

[Love4Boobies]

Wrote some small assembly demos on a CPU that I designed some years ago (just for fun), but never had the patience to do an OS for it. It had little protection and only provided the flat memory model (naturally). I also wrote an emulator for this once, it was available on my web site before it went to hell.

[samoz]

And on a side, unrelated note, I just got 1 star

[Love4Boobies]

Don't they only do that in kidnergarten?

[samoz]

No not me... I was the bad child

What type of CPU did you design? What sort of components did you use? How many instructions did you implemenet?

You said you're website went to hell, but is there anything left about it on there? (I tried to find your website in your profile, but was unable to. If it is there in your profile, my apologies, I'm not too familiar with the new OSdev layout yet.)

[Love4Boobies]

What type of CPU did you design? What sort of components did you use? How many instructions did you implemenet?

Although I like x86(_64) best, it was more of a MIPS wanna-be design. My reasons were:

- I wanted a RISC processor and although nowadays Intel CPUs *are* RISC (with some compatibility layer on top, if you like), it's a lot of work.
- I really liked the idea of software controlled TLBs and virtual memory schemes. I never actually got as far as to implement all that, but I was planning to.

The project was never complete; for instance there was (sort of) no cache, no TLBs, no SMT, nor any other fancy features (except for one - precise interrupts ).

You said you're website went to hell, but is there anything left about it on there? (I tried to find your website in your profile, but was unable to. If it is there in your profile, my apologies, I'm not too familiar with the new OSdev layout yet.)

I'm planning to put my website online again sometime soon. In the meantime, there's no online copy of the emulator but I can send you one in a couple of weeks after I take a trip home (@ collage now).

[Mark139]

I did a simple OS for ARM processors. Didn't get it running on real h/w just emulators (skyeye and qemu), but there are a couple of interesting bits of h/w to look at for ARM coding. Gumstix (http://www.gumstix.com/index.html) and beagle board (http://beagleboard.org/) - I should have bought one when the pound was strong against the US $.

ARM coding is so much simpler than x86 - not too many instructuctions to remember

If work wasn't so hectic I'd get back into it.

I've also looked at designing a CPU using VHDL. Again only ever emulated as I didn't buy a FPGA kit.

Cheers

[JamesM]

Hi,

My OS supports x86, x86_64, PowerPC, MIPS and ARM. The last three are the most interesting but least used out of the five - my personal favourite being PPC

James

[jal]

but is anyone using any other sorts of chips?

In the embedded market, there's AVRs and such that get used. See e.g. here (PDF).


JAL

[Mark139]

Which emulator do you use for PPC work?
A few years back I did a brief bit fo work with the PPC and quite liked it. Maybe I should have a play with that.

[JamesM]

Which emulator do you use for PPC work?
A few years back I did a brief bit fo work with the PPC and quite liked it. Maybe I should have a play with that.

Mainly PearPC. I haven't got qemu-system-ppc to work yet.