I'm just wondering if there is a special reason to use xmm0 twice to calculate DCOffset of a signal block?
I thought that DCOffset of a signal block would be like (sum(xmm0)+sum(xmm1)+...+sum(xmm6))/28 (sum(xmm#) means summation of four complex signals)
But in SORA, to my understanding, it is (sum(xmm0)*2+sum(xmm1)+...+sum(xmm6))/32
I think it's an approximation. The purpose is to save computational cost. Divided by 32 is much cheaper than divided by 28 in terms of computational cost.