Developer Guide and Reference

Contents

_mm256_sub_epi8/16/32/64

Subtracts signed/unsigned packed 8/16/32/64-bit integer data elements of two vectors. The corresponding Intel® AVX2 instruction is
VPSUBB
,
VPSUBW
,
VPADDD
, or
VPSUBQ
.

Syntax

extern __m256i _mm256_sub_epi8(__m256i s1, __m256i s2);
extern __m256i _mm256_sub_epi16(__m256i s1, __m256i s2);
extern __m256i _mm256_sub_epi32(__m256i s1, __m256i s2);
extern __m256i _mm256_sub_epi64(__m256i s1, __m256i s2);
Arguments
s1
integer source vector used for the operation
s2
integer source vector used for the operation
Description
Subtracts packed signed/unsigned 8-, 16-, 32-, or 64-bit integers of the second source vector
s2
from corresponding bits of the first source vector
s1
and stores the packed integer result in the destination vector. When an individual result is too large to be represented in 8/16/32/64 bits (overflow), the result is wrapped around and the low 8/16/32/64 bits are written to the destination vector (that is, the carry is ignored).
You must control the range of values operated upon to prevent undetected overflow conditions.
Returns
Result of the subtraction operation.