Hello,
I did a quick test on AVR28DA128 with an app that runs 4 tasks which
spin on shuffling bytes around with some I/O mixed in (first task
receives data from serial port, pipes them through second and third to
fourth which echoes it back to
serial port.) Preemptive multitasking with 50ms period is used.
Prevalent activity is the spinning, I/O only happens every 250ms.
The patch adds 800 bytes to resulting binary (the chip has 128kB total.)
As for speed, I attempted to count how many times one of the tasks goes
through the busy loop over a period of time. I think the results are not
completely reliable but on average, I got some 2% decrease (average from
7 runs for each of with/without patch.)
On 2025-12-11 10:43, Tomek CEDRO wrote:
Hello world :-)
There are two interesting PRs that touch scheduler and may need come
extra attention. Please take a look, test, and review in a free moment
:-)
https://github.com/apache/nuttx/pull/17468
https://github.com/apache/nuttx-apps/pull/3246
Thanks :-)
Tomek
--
CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
