In this pipeline, one can see the data selection and shuffling units, PMXL, PMXR, and PMC. The pre-add (PRA) is just before the multiply block and then two lane reduction blocks (PSA, PSB) allows to perform up to 128 multiplies and get an output on 16 lanes down to two lanes. The accumulator block is fed either by its own output (AM) or by the upshift output. The feedback on the ACC block is only one clock cycle.