kmackay
An avr assemble fix for Issue #154
Yesterday, I am trying to compile the micro-ecc in MCU atmega328p (avr mcu) when uECC_OPTIMIZATION_LEVEL=4.
But I got an error about too long assembler code.
So I edit the code to split them in some process.
I know it will reduce the speed, but it's better than nothing.
This is an example when I am using uECC_OPTIMIZATION_LEVEL=4 (Running examples/ecc_test/ecc_test.ino):
20:25:52.404 -> Testing ecc
20:25:53.572 -> Made key 1 in 1171
20:25:54.655 -> Made key 2 in 1066
20:25:55.726 -> Shared secret 1 in 1063
20:25:56.754 -> Shared secret 2 in 1066
20:25:56.754 -> Shared secrets are identical
The test finals in 4.35s.
And this is an example when I am using uECC_OPTIMIZATION_LEVEL=2:
20:35:04.517 -> Testing ecc
20:35:05.638 -> Made key 1 in 1127
20:35:06.807 -> Made key 2 in 1183
20:35:07.875 -> Shared secret 1 in 1074
20:35:08.949 -> Shared secret 2 in 1068
20:35:08.949 -> Shared secrets are identical
This process finals in 4.432s.
I think the optimize -Os reduces the optimiztion.
So I use -Ofast, and got:
21:01:31.967 -> Testing ecc
21:01:32.756 -> Made key 1 in 791
21:01:33.362 -> Made key 2 in 632
21:01:33.968 -> Shared secret 1 in 583
21:01:34.529 -> Shared secret 2 in 580
21:01:34.529 -> Shared secrets are identical
only 2.562s!
@aeruder can you merge this or check it? this is a good PR for avr platform.