"Masking" article

[betsch85]

https://github.com/algorithmica-org/algorithmica/blob/master/content/english/hpc/simd/masking.md claims that

// no branch:
for (int i = 0; i < N; i++)
    s += (a[i] < 50) * a[i];

// also no branch:
for (int i = 0; i < N; i++)
    s += (a[i] < 50 ? a[i] : 0);

are both branchless, while I'm not sure about the first one, the 2nd one is definitely not branchless. (cond?true:false) is just syntactic sugar for if-then-else

[sslotin]

This is true, but compilers treat ternary operators differently from if-else constructions and are more likely to replace them with cmov-based predication. The article links to https://en.algorithmica.org/hpc/pipelining/branchless/, which elaborates on why this is happening.