I have two functions counting the occurrences of a target char in the given input buffer. The functions vary only in how they communicate the result back to the caller; one returns the result and the other writes to a variable passed by reference.

#include 

#define BUF_LEN 0x1000

size_t check_count1(const char* buf, char target) {
    size_t count = 0;
    for (size_t i = 0; i < BUF_LEN; i++) {
        if (buf[i] == target) {
            count++;
        }
    }
    return count;
}

void check_count2(const char* buf, char target, size_t& count) {
    for (size_t i = 0; i < BUF_LEN; i++) {
        if (buf[i] == target) {
            count++;
        }
    }
}

I am puzzled by how Clang and GCC generate code for these two functions. The loop in check_count1 is vectorized, but for check_count2 it's not. Initially I thought this was due to pointer aliasing in the second case, but specifying __restrict has no effect. Here's the link to compiler explorer.

An older ICC compiler did just fine with both loops. What changed?