4 ms·

The problem space a full Linux SoC solves is very different from the problem an MCU solves. (Imagine a lot of real time io, interacting with a lot of peripheral

• Joel_Mckay 10mo ago
  The Arduino 1kHz busy-loop and ISR designs is not real-time or necessarily guaranteed latency. RTLinux or FreeRTOS are guaranteed-latency OS, that can reduce timing jitter. "Real-time" is another term hijacked by people that think it means the same thing. It was a disservice to students, but thankfully FPGA or Raspberry Pico io DMA with mcu parallelism can teach people to understand clock domain crossing. Part of the Arduino appeal, was convincing people the mcu and application SoC problem domains were the same. They are not, and never were... Hence the new Arduino board dual chip layout is twice as bad in some ways. Cheers =3 https://www.youtube.com/watch?v=G2y8Sx4B2Sk https://www.youtube.com/watch?v=G2y8Sx4B2Sk risc v -- a new hope, but 20 years too late stm32 -- a solid competitor to Microchip/AVR, but 14 years too late esp32 -- FreeRTOS with hardware flash paging can pose an issue, but at least it already doesn't need Arduino software lol
  • snvzz 10mo ago
    >RTLinux or FreeRTOS are guaranteed-latency OS, that can reduce timing jitter. Unlike seL4[0], neither OS has a guarantee of worst case latency. RTLinux cannot realistically have one, as formal verification is not feasible due to its sheer size. 0. https://sel4.systems/ https://sel4.systems/
    • Joel_Mckay 10mo ago
      RTLinux had many variants, but in general nobody wants to introduce chaotic processes while running previously profiled runtime code. It was odd xenomai was not as common as preemt_rt in projects. =3