There are "Arduinos" available now with much greater speed and memory capacity, while still being small and and inexpensive. Maybe with those you could avoid writing assembler and stick to C++. Then, if you are successful, you have software that is more easily transportable to other mcus in the future. Examples of these are the Teensy 3.x family, the Maple Mini, the AdaFruit Feather M0, ItsyBitsy M0 and Trinket M0.