When completed, this worksheet will consist of four sections. The first makes use of the pipe-like pipe with floating point execution pipelines as presented in Appendix A of H & P with a few minor modifications. Only part one is relevant to the take-home quiz.
The second uses Tomasulo's algorithm described in Chapter 3 of H & P, with some additional design assumptions required to support execution of the MIPS ISA. Finally, we return to the pipe-like pipe, but modify it based on more realistic assumptions regarding memory access times.
In each case, you will be required to calculate the latencies between pairs of instructions that share an operand. The reason we are doing this for the Tomasulo algorithm is that you HAVE to know the latencies to write down an execution profile without too much extrae work. Furthermore, when we move to speculative execution (where we execute stuff based on what we think the branch will do) and super scalar (where we can issue more than one instruction per clock cycle), you will be too busy tracking 18 other assumptions to have time to deal with the latency.