Re: [Simulavr-devel] Settings for stackpointer

[Petr Hluzín]

On 23 February 2016 at 01:32, Markus Hitter <address@hidden> wrote:
> Am 23.02.2016 um 01:16 schrieb Petr Hluzín:
>> IMO the reason of algorithm testing and examples for learning are fantasy.
>
> D'oh. So we're doing fantasy:
>
> http://reprap.org/wiki/Teacup_Firmware#Teacup_in_SimulAVR
>
> It's also used to run semi-automated regression tests for the firmware:

Automated testing is a common use case for simulavr (or any a MCU
simulator), maybe the most common.

But the Teacup firmware is not tested on tiny11/12/15, it needs ~16x
more of flash. Neither Teacup's Bresenham algorithm or other algorithm
is tested on the Tiny part. It would be possible to compile the
Bresenham algorithm alone and test it, but that is not practical.
Experiments with algorithms are done on the beefiest MCU/CPU available
because it is more convenient and author is not concerned about unit
cost at the stage yet.

It is about those tiny parts we are talking about. Big AVRs are no problem.

-- 
Petr Hluzin