devela::_dep::safe_arch

Function dot_product_m128

pub fn dot_product_m128<const IMM: i32>(a: m128, b: m128) -> m128
Available on crate feature dep_safe_arch only.
Expand description

Performs a dot product of two m128 registers.

The output details are determined by a control mask:

  • For each lane, you can multiply that lane from a and b or you can take a default of 0.0

  • Bits 4 through 7 determine if we should mul lanes 0 through 3.

  • This forms four temporary f32 values which are summed to a single f32.

  • For each output lane, you can have the sum in that lane or 0.0.

  • Bits 0 through 3 determines if the sum is in lanes 0 through 3.

  • Intrinsic: _mm_dp_ps

  • Assembly: dpps xmm, xmm, imm8