Dot Product on TigerSHARC

loop:

XR1:0 = Q[j10+=2]; YR1:0 = Q[k0+=2];; // load 8 samples

XR3:2 = Q[j0+=2]; YR3:2 = Q[k0+=2];; // load 8 coeffs

// one cycle stall happens here

if NLC0E, jump loop; MR3:0 += R1:0*R3:2;; // loop, 8 MACs