In C, why is “signed int” faster than “unsigned int”?

Question

In C, why is signed int faster than unsigned int? True, I know that this has been asked and answered multiple times on this website (links below).

Answer 1

From Instruction specification on AMD/Intel we have (for K7):

Instruction Ops Latency Throughput
DIV r32/m32 32  24      23
IDIV r32    81  41      41
IDIV m32    89  41      41 

For i7, latency and throughput are the same for IDIVL and DIVL, a slight difference exists for the µops.

This may explain the difference as -O3 assembly codes only differ by signedness (DIVL vs IDIVL) on my machine.