I solved the halting problem, it doesn't exist.

Just check every n + 1 clock cycles whether the previous savestate (a term from speedrunning) matches the current savestate exactly, byte for byte. You can discard the previous previous savestate.

n can be a arbitrary precision integer.

You add 1 to n every time you check.

Now where's my million dollars for proving P=NP...