Depending on the optimization selection, the code could be very different. For a 8-bit core, speed and size are often compatible. With these 32 bit cores, there are many choices for the code generation, and optimization choice is important.
Yes, but CircleOS is supposed to be the same for everybody. Another solution would consist in modifying the divisors for the handlers, or the systick frequency, etc... Where do you need to make CircleOS faster ?
