Available on x86 and target feature
avx only.Expand description
Conditionally multiplies the packed single-precision (32-bit) floating-point
elements in a and b using the high 4 bits in imm8,
sum the four products, and conditionally return the sum
using the low 4 bits of imm8.