Available on x86 and target feature
fma only.Expand description
Multiplies packed single-precision (32-bit) floating-point elements in a
and b, and alternatively subtract and add packed elements in c from/to
the intermediate result.