Depends : It's harder to do static analysis than in 1980, because there are far more complex interactions going on internally with a SoC design. It is usually one, but your execution speed is going to depend on how the flash prefetch works (because the flash is slow, probably ~30-35ns access time, and the buffering mechanism assumes linear code flow) whereas the ram is single cycle. Your load/store instructions are going to depend on the bus speed of the source/destination, be it multiple wait states on the flash, or slower peripheral buses. Or if others are using the bus. Want to benchmark code, uses the DWT_CYCCNT cycle counter in the trace/debug unit. Editing apparently alters the post ordering.
Tips, Buy me a coffee, or three.. PayPal Venmo (See Profile) Up vote any posts that you find helpful, it shows what's working..