I was going to say as well. It seems like if caches are the bane of sequential processing(CPU) then routing has to be the counterpart on the parallel(FPGA/ASIC)

I was going to say as well. It seems like if caches are the bane of sequential processing(CPU) then routing has to be the counterpart on the parallel(FPGA/ASIC) side of the equation.