What are the SetX instructions and how do they work?

SetX instructions write a single byte to 0 or 1 based on the condition codes — turning a comparison's flags into an actual 0/1 value, as for a C boolean.

* A comparison becomes a boolean: setX turns the flags into one 0/1 byte, then movzbl cleans the rest of the register. *

After a cmp, the answer lives in the flags; setX materializes that answer as a byte you can store or return. There's one for each condition, mirroring the conditional jumps.

Because setX writes only a single byte, the usual idiom zero-extends it to a full register:

cmpq   %rsi, %rdi   # compare x and y
setg   %al          # %al = (x > y) ? 1 : 0
movzbl %al, %eax    # zero-extend to a clean 32/64-bit result

Tip: the suffixes (e, ne, g, l, a, b, …) are the same across setX, jX, and cmovX, so learning one set teaches all three.

Go deeper:

• Felix Cloutier — SETcc — the SETcc family materialising flags into a 0/1 byte.