In all schematics illustrating the Streaming Multiprocessor’s content we always see just the eight Streaming Processors, but then we read that there are also two Special Function Units for more sophisticated mathematical functions (square root, sine, etc) and also a double-precision unit. So my question is, are those SFUs and the DP unit executing separately from the SPs, or are they called upon by the SPs? In other words - can an SM perform more than 8 ops simultaneously? I’m 90% sure the DP unit is independent, but I have no clue how it is with the two SFUs. I think I’ve read somewhere that each of them is connected to four SPs, which would imply some sort of dependence.