_mm_hsubs_epi16
Microsoft Specific
Emits the Supplemental Streaming SIMD Extensions 3 (SSSE3) instruction phsubsw. This instruction computes the difference between the elements of two 128-bit parameters.
__m128i _mm_hsubs_epi16( __m128i a, __m128i b );
A 128-bit value that contains eight 16-bit signed integers. Each integer is the difference between adjacent pairs of elements in the input parameters.
The result can be expressed with the following equations:
r0 := SATURATE_16(a0 - a1) r1 := SATURATE_16(a2 - a3) r2 := SATURATE_16(a4 - a5) r3 := SATURATE_16(a6 - a7) r4 := SATURATE_16(b0 - b1) r5 := SATURATE_16(b2 - b3) r6 := SATURATE_16(b4 - b5) r7 := SATURATE_16(b6 - b7)
r0-r7, a0-a7, and b0-b7 are the sequentially ordered 16-bit components of return value r and parameters a and b. r0, a0, and b0 are the least significant 16 bits.
SATURATE_16(x) is defined as ((x > 32767) ? 32767 : ((x < -32768) ? -32768 : x))
Before you use this intrinsic, software must ensure that the processor supports the instruction.
#include <stdio.h>
#include <tmmintrin.h>
int main ()
{
__m128i a, b;
a.m128i_i16[0] = 32;
a.m128i_i16[1] = 32;
a.m128i_i16[2] = 4096;
a.m128i_i16[3] = -4096;
a.m128i_i16[4] = -128;
a.m128i_i16[5] = 128;
a.m128i_i16[6] = 100;
a.m128i_i16[7] = 32767;
b.m128i_i16[0] = 32700;
b.m128i_i16[1] = -1000;
b.m128i_i16[2] = -8192;
b.m128i_i16[3] = 30000;
b.m128i_i16[4] = 512;
b.m128i_i16[5] = 0;
b.m128i_i16[6] = 0;
b.m128i_i16[7] = 2;
__m128i res = _mm_hsubs_epi16(a, b);
printf_s("Original a:\t%6d\t%6d\t%6d\t%6d\n\t\t%6d\t%6d\t%6d\t%6d\n",
a.m128i_i16[0], a.m128i_i16[1], a.m128i_i16[2], a.m128i_i16[3],
a.m128i_i16[4], a.m128i_i16[5], a.m128i_i16[6], a.m128i_i16[7]);
printf_s("Original b:\t%6d\t%6d\t%6d\t%6d\n\t\t%6d\t%6d\t%6d\t%6d\n",
b.m128i_i16[0], b.m128i_i16[1], b.m128i_i16[2], b.m128i_i16[3],
b.m128i_i16[4], b.m128i_i16[5], b.m128i_i16[6], b.m128i_i16[7]);
printf_s("Result res:\t%6d\t%6d\t%6d\t%6d\n\t\t%6d\t%6d\t%6d\t%6d\n",
res.m128i_i16[0], res.m128i_i16[1], res.m128i_i16[2], res.m128i_i16[3],
res.m128i_i16[4], res.m128i_i16[5], res.m128i_i16[6], res.m128i_i16[7]);
return 0;
}
Original a: 32 32 4096 -4096
-128 128 100 32767
Original b: 32700 -1000 -8192 30000
512 0 0 2
Result res: 0 8192 -256 -32667
32767 -32768 512 -2