_mm_dp_ps

Microsoft Specific

Emits the Streaming SIMD Extensions 4 (SSE4) instruction dpps. This instruction computes the dot product of single precision floating point values.

__m128 _mm_dp_ps( 
   __m128 a,
   __m128 b,
   const int mask 
);

Parameters

• [in] a
  A 128-bit parameter that contains four 32-bit floating point values.

• [in] b
  A 128-bit parameter that contains four 32-bit floating point values.

• [in] mask
  A constant mask that determines which components will be multiplied and where to place the results.

Result value

A 128 bit parameter that contains the 32-bit results of the dot products.

The result can be expressed with the following equations:

tmp0 := (mask4 == 1) ? (a0 * b0) : +0.0
tmp1 := (mask5 == 1) ? (a1 * b1) : +0.0
tmp2 := (mask6 == 1) ? (a2 * b2) : +0.0
tmp3 := (mask7 == 1) ? (a3 * b3) : +0.0

tmp4 := tmp0 + tmp1 + tmp2 + tmp3

r0 := (mask0 == 1) ? tmp4 : +0.0
r1 := (mask1 == 1) ? tmp4 : +0.0
r2 := (mask2 == 1) ? tmp4 : +0.0
r3 := (mask3 == 1) ? tmp4 : +0.0

Requirements

Header file <smmintrin.h>

Remarks

The immediate bits 4-7 of mask determine which of the corresponding source operand pairs are to be multiplied. Bits 0-3 determine whether the dot product result will be written. If a mask bit is 0, the corresponding product result or written value is +0.0.

r0-r3, a0-a3, and b0-b3are the sequentially ordered 32-bit components of return value r and parameters a and b, respectively. r0, a0, and b0 are the least significant 32 bits.

maski is bit i of parameter mask, where bit 0 is the least significant bit.

Before you use this intrinsic, software must ensure that the underlying processor supports the instruction.

Example

#include <stdio.h>
#include <smmintrin.h>

int main ()
{
    __m128 a, b;
    const int mask = 0x55;

    a.m128_f32[0] = 1.5;
    a.m128_f32[1] = 10.25;
    a.m128_f32[2] = -11.0625;
    a.m128_f32[3] = 81.0;
    b.m128_f32[0] = -1.5;
    b.m128_f32[1] = 3.125;
    b.m128_f32[2] = -50.5;
    b.m128_f32[3] = 100.0;

    __m128 res = _mm_dp_ps(a, b, mask);

    printf_s("Original a: %f\t%f\t%f\t%f\nOriginal b: %f\t%f\t%f\t%f\n",
                a.m128_f32[0], a.m128_f32[1], a.m128_f32[2], a.m128_f32[3],
                b.m128_f32[0], b.m128_f32[1], b.m128_f32[2], b.m128_f32[3]);
    printf_s("Result res: %f\t%f\t%f\t%f\n",
                res.m128_f32[0], res.m128_f32[1], res.m128_f32[2], res.m128_f32[3]);

    return 0;
}

Original a: 1.500000    10.250000       -11.062500      81.000000
Original b: -1.500000   3.125000        -50.500000      100.000000
Result res: 556.406250  0.000000        556.406250      0.000000

See Also

Reference

Compiler Intrinsics