Debugging / neon performance threats in neon hand code

Question

Debugging / neon performance threats in neon hand code

Initially, the problem arose when I tried to optimize the algorithm for the neon hand, and some part of it was 80% according to the profiler. I tried to check what can be done to improve it, and for this I created an array of function pointers for different versions of my optimized function, and then ran them in a loop to see in the profiler which works better:

typedef unsigned(*CalcMaxFunc)(const uint16_t a[8][4], const uint16_t b[4][4]);
CalcMaxFunc CalcMaxFuncs[] =
{
    CalcMaxFunc_NEON_0,
    CalcMaxFunc_NEON_1,
    CalcMaxFunc_NEON_2,
    CalcMaxFunc_NEON_3,
    CalcMaxFunc_C_0
};


int N = sizeof(CalcMaxFunc) / sizeof(CalcMaxFunc[0]);
for (int i = 0; i < 10 * N; ++i)
{
    auto f = CalcMaxFunc[i % N];
    unsigned retI = f(a, b);

    // just random code to ensure that cpu waits for the results
    // and compiler doesn't optimize it away
    if (retI > 1000000)
        break;
    ret |= retI;
}