µctrl Raspberryπ pico & arduino

[Liam Hays]

EdS2 Wrote:Just a small point: it's not accurate to think of the Zero, or any other of the original Pi boads, as necessarily running Linux.

I remember once seeing a chess game that ran entirely in ARM assembly on a Pi, and like you say, it could boot in a fraction of a second. For a calculator, that's fine (look at the 48---every time I turn mine on, the screen flashes for about that long). That makes me wonder how power consumption is, between Linux and a dedicated program, on a Pi Zero.

EdS2 Wrote:A paired Zero and Pico might well be a common solution.

I don't know how this could be done, but it seems possible: a Zero for the screen, USB mass storage, and execution of programs, and a Pico for real-time hardware and keypad I/O and control. The speed of the Zero would give the machine modern calculator speeds, and the Pico would make infrared serial and other peripheral control easy.

I also thought of something else: the M0-based RP2040, as well as the ESP32, don't have hardware floating point support. I don't know about a Pi, but they're fast enough that I suspect software floating point, if needed, would not create tremendous overhead.

Maximilian Hohmann Wrote:but provining the Arduino programming environment (C / C++) via the "normal" Ardunio IDE.

I would like to see that, and I think it's likely, since there's already Arduino support for the raw M0. It would be really neat if they could incorporate the PIO assembly thing into the simpler Arduino environment, with all the same capability. Plus, if it means the board is compatible with the many Arduino libraries and peripherals...it could make the Pico promising for a lot of interesting projects.

That's all from me, for now. (Look! I figured out how to quote properly! )

[ijabbott]

(01-27-2021 09:56 PM)Liam Hays Wrote:  I remember once seeing a chess game that ran entirely in ARM assembly on a Pi, and like you say, it could boot in a fraction of a second. For a calculator, that's fine (look at the 48---every time I turn mine on, the screen flashes for about that long). That makes me wonder how power consumption is, between Linux and a dedicated program, on a Pi Zero.

For comparison, a TI Nspire CX calculator takes about a minute to boot up if it hasn't been turned on for a while!

Liam Hays Wrote:I also thought of something else: the M0-based RP2040, as well as the ESP32, don't have hardware floating point support. I don't know about a Pi, but they're fast enough that I suspect software floating point, if needed, would not create tremendous overhead.

I think Casio had custom versions of the Hitachi SuperH (now Renesas SuperH) CPUs produced without hardware floating point for their graphing calculators, because the hardware floating point unit worked in binary, but the calculator worked in decimal (like most calculators).

[John Keith]

(01-28-2021 12:37 AM)ijabbott Wrote:  I think Casio had custom versions of the Hitachi SuperH (now Renesas SuperH) CPUs produced without hardware floating point for their graphing calculators, because the hardware floating point unit worked in binary, but the calculator worked in decimal (like most calculators).

Indeed, this is why Free42 binary runs faster than Free42 decimal. For a calculator* it is probably better to have a CPU architecture/instruction set that makes BCD operations as efficient as possible.


* Or computer math software generally.

[David Hayden]

I suspect that the RP nano still draws too much power for a calculator app. The Datasheet says it draws about 1mA, even in sleep mode. Running code it seems to draw around 10mA. Isn't that to high for a calculator?

[grsbanks]

(02-02-2021 10:35 PM)David Hayden Wrote:  I suspect that the RP nano still draws too much power for a calculator app. The Datasheet says it draws about 1mA, even in sleep mode. Running code it seems to draw around 10mA. Isn't that to high for a calculator?

The 10mA figure isn't too bad. That's not much more than the DM41X/DM42 running at full speed. The 1mA sleep current is a lot, though. The SwissMicros machines suck about 6µA when sleeping (off).

[EdS2]

The RP Pico might be a bit better than that: the SLEEP mode is not quite as deep as DORMANT mode, and even with the hello_dormant example, it looks to me as if the RAMs are not powered down, and neither are the PLLs. So, I think there might be scope for further reduced power - but the Pico board has more on it than the RP2040 chip, so even with the chip at lowest dissipation, it might not constitute an ultra-low-power solution.

[compsystems]

Hopefully, future flowCode can integrate this new boards into your IDE to code not only in written form, but also with flowcharts, blocks, and other programming modes.

https://www.flowcode.co.uk/