Filter
Top Products
Optimizing for SIMD Floating-point Applications 5
5-21
Use of cvttps2pi/cvttss2si Instructions
The cvttps2pi and cvttss2si instructions encode the truncate/chop
rounding mode implicitly in the instruction, thereby taking precedence
over the rounding mode specified in the
MXCSR register. This behavior
can eliminate the need to change the rounding mode from
round-nearest, to truncate/chop, and then back to round-nearest to
resume computation. Frequent changes to the
MXCSR register should be
Example 5-10 Horizontal Add Using Intrinsics with movhlps/movlhps
void horiz_add_intrin(Vertex_soa *in, float *out)
{
__m128 v1, v2, v3, v4;
__m128 tmm0,tmm1,tmm2,tmm3,tmm4,tmm5,tmm6;
// Temporary variables
tmm0 = _mm_load_ps(in->x); // tmm0 = A1 A2 A3 A4
tmm1 = _mm_load_ps(in->y); // tmm1 = B1 B2 B3 B4
tmm2 = _mm_load_ps(in->z); // tmm2 = C1 C2 C3 C4
tmm3 = _mm_load_ps(in->w); // tmm3 = D1 D2 D3 D4
tmm5 = tmm0; // tmm0 = A1 A2 A3 A4
tmm5 = _mm_movelh_ps(tmm5, tmm1); // tmm5 = A1 A2 B1 B2
tmm1 = _mm_movehl_ps(tmm1, tmm0); // tmm1 = A3 A4 B3 B4
tmm5 = _mm_add_ps(tmm5, tmm1); // tmm5 = A1+A3 A2+A4 B1+B3 B2+B4
tmm4 = tmm2;
tmm2 = _mm_movelh_ps(tmm2, tmm3); // tmm2 = C1 C2 D1 D2
tmm3 = _mm_movehl_ps(tmm3, tmm4); // tmm3 = C3 C4 D3 D4
tmm3 = _mm_add_ps(tmm3, tmm2); // tmm3 = C1+C3 C2+C4 D1+D3 D2+D4
tmm6 = tmm3; // tmm6 = C1+C3 C2+C4 D1+D3 D2+D4
tmm6 = _mm_shuffle_ps(tmm3, tmm5, 0xDD);
// tmm6 = A1+A3 B1+B3 C1+C3 D1+D3
tmm5 = _mm_shuffle_ps(tmm5, tmm6, 0x88);
// tmm5 = A2+A4 B2+B4 C2+C4 D2+D4
tmm6 = _mm_add_ps(tmm6, tmm5);
// tmm6 = A1+A2+A3+A4 B1+B2+B3+B4
// C1+C2+C3+C4 D1+D2+D3+D4
_mm_store_ps(out, tmm6);
}