This is supported on
target_family="wasm"
and target feature simd128
only.Expand description
Subtracts two 128-bit vectors as if they were two packed eight 16-bit
signed integers, saturating on overflow to i16::MIN
.