Running CP/M on the Raspberry Pi Pico | Hacker News Reader

Running CP/M on the Raspberry Pi Pico

kevinboone.me

58 points_Microft23 comments

> CP/M programs often behave in ways which would be damaging to modern hardware -- endless input-checking loops at full CPU speed, for example. To be honest, I do not know if this kind of thing would be damaging to a Pico, but I don't want to risk it. CPICOM has a number of pause points built in, to give the microcontroller time to idle. These do slow the system down even further, and I'm not sure whether they're necessary.

This shouldn't be necessary on an MCU. Any kind of sleep call is probably just spinning the CPU anyway. There's often support for deeper power saving states, but MCUs are generally designed to be run full speed all the time.

Couple of things to note

1) The Pico doesn't have an OS, by default. It's designed to run bits of code people have written in the SDK. There are no guarantees that someone won't write code that endlessly busy-waits checking for a condition, and no warnings that doing so can be harmful

2) It costs about $4. Is there really much to lose if you "risk it" with this software?

its my understanding that you cannot damage the physical CPU by software instructions, period. However, heat from multi-core operations is not the same. Hardware gets damaged by heat all the time. So "loops waiting for input" may never damage the hardware, but "causing more cores to operate at the same time" can cause heat damage on some setups..

I don't think busy loops are damaging to the hardware. They just pointlessly waste CPU time.

There are CPUs out there that can be put into a "sleep" type mode. On the x86, there's the HLT instruction, which stops CPU execution until an interrupt happens. The 68k series has the STOP instruction, which does the same thing.

Halt and Catch Fire:

https://en.m.wikipedia.org/wiki/Halt_and_Catch_Fire_(computi...

> if the ZX80 could generate a video signal from its CPU while still running BASIC programs back in 1980,

It couldn't. The ZX81 could.

https://www.sinclairzxworld.com/viewtopic.php?p=20645&sid=86...

The chain certs expired last September.

https://www.ssllabs.com/ssltest/analyze.html?d=kevinboone.me

Preservation is always good, and some of the software from the CP/M world can be quite useful.

People can use constant reminders that software can be good without taking gigabytes of memory for the most basic of functions. We can literally emulate an entire computer and run its OS and software faster and more efficiently than we can render a page in a web browser.

Small and efficient is good :)

Indeed. You can go a very, very long way in productivity with Wordstar, Visicalc and Turbo Pascal on CP/M, each of which fit on a floppy. Many of us did.

Back when everything was new for the first time.

The only thing new about those tools was they were available on an affordable, personal platform. Wang, among others, had Wordstar-like computer-based word processing for years at that point. Pascal was old hat. Visicalc was new and revolutionary, but the idea of using computers to simplify computation was primordial.

I get that there's a certain mentality that thinks nothing happened in computer productivity until 2010 or so (date marching forward as new grads enter the workforce), but when you don't bury your word processor or spreadsheet under a metric ton of useless UI candy and 'features' that less than 1% of your users actually use, you get some pretty lean, useful tools.

Oh absolutely, well said. I miss those days when time spent at the screen was intent time. For a long while now for me, it's basically just been a TV replacement.

Your main point is valid but 2010 is maybe 25 years too late for the example programs listed.

You missed the point.

> CP/M predates the open source movement by thirty years.

CP/M was first developed in 1974, the FSF was founded in 1985, so I would say more like eleven years.

Beyond the FSF, Berkeley released sources to BSD as open source throughout the '70s.

IBM SHARE predates both BSD and FSF. Open Source used to common, especially before portable binaries.

If you mean bytecode as executable format, the Burroughs B5000 from 1961 was one of the first computers to use such concept.

I mean more the issue where the computers were differing wildly enough that binary distribution of software was very limiting, so it made sense to offer source that could be massaged to run on specific setup.

Just out of interest... there was a pre-existing CP/M emulator for the Pico.

https://github.com/guidol70/RunCPM_RPi_Pico

It's a port of RUNCPM:

https://github.com/MockbaTheBorg/RunCPM

And for the avoidance of doubt, the modern RunCPM emulator is nothing to do with the old MS-DOS tool of the same name.

I used it, but I can't remember whence that came. I also used 22NICE from Sydex. Info on such things:

https://www.cpm80.com/

https://www.sydex.com/index.html

http://www.primrosebank.net/computers/cpm/cpm_emul.htm

So this could run Borland Turbo Pascal up to version 3 (https://en.wikipedia.org/wiki/Turbo_Pascal)! Very cool: developing software on a microcontroller itself.

> Catalog vs > dir or > ls

Reply on news.ycombinator.com