__m64_pavg1_nraz
Microsoft Specific
Emits the 1-byte form of the IPF Parallel Average (pavg1) instruction, which computes the average of the two unsigned 8-bit integers in each byte of the input arguments and puts that average in the corresponding byte of the result.
__m64 __m64_pavg1_nraz(
__m64 a,
__m64 b
);
Parameters
- [in] a
An __m64 union containing an array of eight unsigned 8-bit integers.
- [in] b
An __m64 union containing an array of eight unsigned 8-bit integers.
Return Value
An __m64 union containing the array of eight unsigned 8-bit integers, each position of which is the average of the corresponding integers in the source arrays.
Requirements
Intrinsic | Architecture |
---|---|
__m64_pavg1_nraz |
IPF |
Header file <intrin.h>
Remarks
If fractional, the result in each byte is rounded off to the nearest odd number. The suffix nraz means "not rounding away from zero."
Example
// pavg1_nraz.cpp
// processor: IPF
#include <stdio.h>
#include <intrin.h>
#pragma intrinsic(__m64_pavg1_nraz)
void print8u(unsigned __int8* ia)
{
printf_s("{ %4u, %4u, %4u, %4u, %4u, %4u, %4u, %4u }\n",
ia[0], ia[1], ia[2], ia[3], ia[4], ia[5], ia[6], ia[7] );
}
int main()
{
__m64 m, n, result;
unsigned __int8 ra[8], i;
unsigned __int8 ma[8] = { 10, 10, 10, 10, 10, 10, 10, 10 };
unsigned __int8 na[8] = { 1, 3, 5, 7, 9, 11, 15, 1 };
for (i = 0; i < 8; i++)
{
m.m64_u8[i] = ma[i];
n.m64_u8[i] = na[i];
}
printf_s("a: \n");
print8u(ma);
printf_s("b: \n");
print8u(na);
printf_s("\n");
result = __m64_pavg1_nraz(m, n);
printf_s("Parallel Average (a[]+b[])/2 : \n");
print8u(result.m64_u8);
printf_s("\n");
// This operation is commutative.
result = __m64_pavg1_nraz(n, m);
printf_s("Parallel Average (b[]+a[])/2 : \n");
print8u(result.m64_u8);
printf_s("\n");
}
Output
a: { 10, 10, 10, 10, 10, 10, 10, 10 } b: { 1, 3, 5, 7, 9, 11, 15, 1 } Parallel Average (a[]+b[])/2 : { 5, 7, 7, 9, 9, 11, 13, 5 } Parallel Average (b[]+a[])/2 : { 5, 7, 7, 9, 9, 11, 13, 5 }