On: Popcount CPU instruction

[kardos]

From the article,

GCC: https://godbolt.org/z/JUzmD8

Clang: https://godbolt.org/z/AVqMGl

[account42 → kardos]

Any idea why GCC zeroes out eax before overwriting it?

It never fails to amaze me how smart compilers can be while also being so stupid sometimes.

[vnorilo → account42]

It could be a bug; I don't see an immediate reason to zero eax in this fragment. Sometimes it's easy to miss the reasons why compilers zero apparently unused things though: it can be about breaking data dependencies or making sure flags are in a known state.

[gpderetta → vnorilo]

It is a bug, but in the CPU. Until recently, on intel cpu, the popcnt instruction had an implicit dependency on its output register, which can be a performance issue. Explicitly clearing the register (which is very fast and almost free), is an easy way to workaround it.

[account42 → gpderetta]

Oh, nice info. It's gone when using e.g. -march=znver1.

[vnorilo → kardos]

Indeed, thanks! Skimmed a bit too quickly.