2025-11-04 01:00:41 <tpnix       > +1
2025-11-04 01:03:28 <tpnix       > for instance on small Cortex-M0+ STM32 mcu's such as the STM32G030 which has 64KB flash, a RTOS won't fit, so it's not used. C is used, assembly is used, plus Forth or Basic to a vastly lesser degree
2025-11-04 01:05:04 <tpnix       > Mecrisp-Stellaris is designed for such targets as it only uses 20KB for the image, leaving a healthy 44KB for the Dictionary
2025-11-04 06:49:16 <M-Sci       > AFAICS lisbeths is a carbon-based LLM.
2025-11-04 09:20:29 <veltas      > RTOS's in embedded often are brought in to cover up other fundamental issues and don't solve an awful lot on their own
2025-11-04 09:20:59 <veltas      > FreeRTOS can get pretty tinsy, way under 64KB of flash I think?
2025-11-04 09:22:00 <veltas      > I just see it as a library for scheduling, that's all it really is, but frankly most embedded problems are best served with a simple loop, interrupts, maybe some queues if you want better throughput; don't need an RTOS
2025-11-04 09:22:18 <veltas      > I'd use an RTOS if I was writing some kind of larger safety critical thing
2025-11-04 09:31:42 <tpnix       > Ive never used a RTOS myself as the mcu's I use are far too small to run one, I have used a simple scheduler in a recent Forth powered device and that seems reliable
2025-11-04 09:32:08 <tpnix       > lol, even a simple scheduler takes up a lot of flash
2025-11-04 09:33:18 <tpnix       > in that case I used the multitasker Word with Mecrisp-Stellaris and it worked pretty well, I feel I nailed the main requirements with that one
2025-11-04 10:15:51 <veltas      > Supposedly you can run FreeRTOS with 32K flash
2025-11-04 10:16:12 <veltas      > I'm assuming that's disabling a lot of stuff, and with very small application code
2025-11-04 10:16:26 <veltas      > And optimised code output
2025-11-04 10:21:55 <tpnix       > veltas, a friend tried it on 32kb flash and said there was nothing left for the application
2025-11-04 10:23:18 <tpnix       > freertos looks to be very complex to me, but Ive never run it
2025-11-04 10:23:44 <veltas      > I don't doubt the forum post I saw claiming they had, I also don't doubt it would require a lot of tweaking and be pointless yes
2025-11-04 10:23:59 <veltas      > And no or little room for app
2025-11-04 10:24:34 <veltas      > FreeRTOS is actually quite well designed and lightweight, and not too complicated, but it's just not necessary at all for most embedded apps
2025-11-04 10:24:41 <veltas      > People just throw it in because they want to play with it
2025-11-04 10:24:51 <tpnix       > I mean I wouldnt even bother with Mecrisp-Stellaris on 32kb, while only 20kb is needed, I feel there arent enough resources left
2025-11-04 10:24:58 <veltas      > And in embedded everything you don't need is bloat
2025-11-04 10:25:29 <tpnix       > true
2025-11-04 10:27:09 <tpnix       > I dont get why some people feel compelled to try big systems on small devices ? I mean RTOS is fine on a suitably larger system, be fun and reliable to use etc
2025-11-04 10:27:59 <tpnix       > Id use it on a STM32H750 as it has 2MB flash etc, 400Mhz clock, would be impressive if one likes C and RTOS
2025-11-04 10:28:38 <tpnix       > myself, I love the simplicity of Mecrisp-Stellaris Forth on cortec-m0 with 64kb flash
2025-11-04 10:29:18 <tpnix       > but I only do simple devices, thats where my interest lays
2025-11-04 10:39:49 <veltas      > I don't understand it at all, I came into embedded because I was interested in how things work, last thing I want to do is throw an OS abstraction on top of stuff
2025-11-04 10:40:22 <veltas      > I wanted to know how software actually *did* stuff, not 'how to use an OS' which I'd already done enough of
2025-11-04 10:42:32 <veltas      > I think a lot of people think an OS is really fundamental to computers, but the reality is that lots of computers operate without an OS just fine (albeit more practical if you don't need loads of drivers and a network stack)
2025-11-04 10:43:18 <veltas      > If you're working on an MCU then often there's very little bare metal code required to do most projects, especially just involving GPIO or simpler peripherals
2025-11-04 10:45:34 <veltas      > One of my favourite things I did at work was rewriting bare metal boot firmware in PPC assembly for a system-on-chip, although certain init things were quite thankless like initialising DDR
2025-11-04 10:53:28 <tpnix       > we think somewhat alike then
2025-11-04 10:54:15 <tpnix       > the first commercial device I ever made was a hospital liquids (normal saline) bottle filling machine controller, back in 1984
2025-11-04 10:55:10 <tpnix       > it used a Intel 8085 cpu, and I programmed everything in machine code, with only a hex keypad and 7 seg led displays
2025-11-04 10:55:31 <tpnix       > no os, no nothing, just machine code
2025-11-04 10:57:23 <tpnix       > I learned embedded design building a prototype radioative sensor iron ore flow guage with a National PACE 16 bit cpu in 1976
2025-11-04 10:57:47 <tpnix       > that used the National Semi assembler, teletype and paper tape etc
2025-11-04 10:58:27 <tpnix       > but I was only the tech building it under the design of the lead engineer, I had no clue how much of it worked
2025-11-04 10:59:32 <tpnix       > I built all of it by hand, boards I hand made, pressed in solder turret pins etc, all the other S100 boards were wire wrapped
2025-11-04 11:00:35 <tpnix       > and finally I installed it at the top of a huge iron ore crusher building in Port Hedland in 1977
2025-11-04 11:01:52 <tpnix       > it worked bu using two radioactive sources to measure the thickness of a falling curtain of iron ore and compute the flow in tons per minute
2025-11-04 11:02:59 <tpnix       > it was I guess a legondary introduction to embedded. I was about 20 years old at the timer
2025-11-04 11:37:57 <veltas      > All the older embedded engineers I know have stories like this, it's fascinating
2025-11-04 11:46:00 <tpnix       > they were fascinating days for me
2025-11-04 11:47:43 <tpnix       > Ive been lucky to have such experience being australian
2025-11-04 13:16:39 <KipIngram   > Oh, I think the latter part of the twentieth century was the most fascinating time to be in the profession.  I have my own share of stories and interesting products I've had the privilege of working on over the years.
2025-11-04 13:17:01 <KipIngram   > It feels to me like it's all become rather "sterile" by comparison these days.
2025-11-04 13:17:30 <KipIngram   > Maybe it's just that the steep parts of learning curves are always the most fun.
2025-11-04 13:19:07 <KipIngram   > Take those first years of the microprocessor era.  I missed those - I was a little too young.  But those had to be heady days.