Compute SIMD lanes individually

The Dependencies analysis shows there is a reduction pattern dependency in the loop. Enable vectorization using the directive #pragma omp simd reduction(operator:list).

Example

#pragma omp simd reduction(+:sumx)
...
#pragma omp simd reduction(+:sumx)
for (k = 0;k < size2; k++)
{
    sumx += x[k]*b[k];
}

Read More