3 ms·

Not an ATmega328P (and not ARM), but the Radxa X4 has an Intel N100 with a RP2040 microcontroller built-in.

• teamonkey 1y ago
  I remember someone testing that board (Jeff Geerling, maybe?) and there was significant lag on the GPIO because the MCU is essentially an external device with communication over serial (IIRC), it just happens to be mounted on the motherboard.
  • ThatPlayer 1y ago
    The Raspberry Pi 5 has also done something similar: the CPU no longer controls the GPIO directly, but through a RP1 southbridge. Maybe that's what you're thinking of? Don't see any mention of latency on Jeff's review on it, and I doubt the Radxa X4 is popular enough for any big in-depth review like that. It looks like the RP2040 on the X4 is connected via USB. The RP2040's bootloader mode that enables flashing emulates a mass storage device over USB to transfer firmware, so it would be confusing to not connect it via USB.
    • teamonkey 1y ago
      To clarify, the issue with the X4 is that it essentially is no different than having a Pi Pico attached via USB. This means that your x86 program doesn’t directly control the GPIOs, it must communicate with the RP2040 over USB serial. The RP2040 must be flashed with a firmware that knows how to handle these serial commands. It works, but it’s an extra step and timing may be an issue. On the other hand, a Raspberry Pi Linux program can toggle GPIO pins natively simply by using the official libraries. I mentioned the RP1 in another comment.