Effective SIMD Vectorization for Intel Xeon Phi Coprocessors

Algorithm 1

float x, y31;

for (k=0; k<31; k++) {

  x = x + fsqrt(yk);