Branchless, yes, but the tradeoff is that in practice the program will now execute both branches. No branching, but more code has to run. Being branchless is good for pipelining (no branch misses), but more code is bad for caching (more code, more cache misses). Overall I'd say leave it to the compiler to decide.