Ladvien over at LetsMakeRobots writes, “Excellent article on clock-cycle costs of math functions for uCs and processors–as they relate to programming robots.”
This first point is the single most important one to remember. Integer math is almost always faster than floating point, even on processors that have floating point math units. A whole lot faster normally. But most of the processors we use don’t even have floating point math units. An ATMEGA chip has an integer math unit that can add, subract, and multiply…
Via the contact form.