Yeah, I’ve been there, squashing the same loop until the clock starts jittering. The trick is to keep the state machine tight, but when you start having the code refer to itself it turns into a runaway recursion of pointers that even a 16‑bit core can’t handle. It’s a classic “optimize until you hit the wall” scenario. If you want speed, cut the self‑reference and let the hardware do the looping, or offload the heavy lifting to a co‑processor. The paradox? You spend more cycles tracing the code than doing useful work. So keep the logic simple and let the 16‑bit bus do what it was built for.