Count Matches

Count how many elements in an array are equal to a given target value, using AVX2 intrinsics.

Function Signature

#include <immintrin.h>
#include <cstdint>

int count_matches(const int32_t* arr, int n, int32_t target);

Parameters:

arr — pointer to an array of n signed 32-bit integers, guaranteed 32-byte aligned
n — number of elements, guaranteed to be a multiple of 8 and at least 8
target — the value to count

Returns: the number of elements equal to target

Input:  arr = [1, 2, 3, 2, 5, 2, 7, 2], target = 2
Output: 4

There are two common approaches to this problem:

Movemask + popcount: Compare, extract an 8-bit mask with movemask, and count bits with __builtin_popcount.
Mask-as-minus-one trick: cmpeq produces all-ones per matching lane — which is -1 as a signed integer. Accumulate with add_epi32, then negate the final sum.

The second approach avoids the movemask bottleneck entirely and is a useful pattern to know.

Intrinsic	Description
`_mm256_load_si256(ptr)`	Load 256 bits from aligned memory
`_mm256_set1_epi32(x)`	Broadcast a 32-bit integer to all lanes
`_mm256_cmpeq_epi32(a, b)`	Compare packed 32-bit integers for equality
`_mm256_movemask_ps(v)`	Extract the sign bit of each 32-bit float (8-bit mask)
`_mm256_add_epi32(a, b)`	Add packed 32-bit integers
`__builtin_popcount(x)`	Count set bits in an integer