In case anybody was wondering, the x86 assembly is doing the multiply trick. The rather confusing "lea" instruction -- Load Effective Address -- is used as a space-efficient way of saying "take one register, multiply by 1, 2, 4, or 8, add another register, and put it in a third register". The thing in the register that gets multiplied by that constant is either a 1 or a 0.

If the constants had been less convenient numbers, like 42, the compiler could have instead used a conditional move instruction -- more general-purpose, but I believe slightly larger and often ever-so-slightly slower.