13
for class on Thursday Oct. 26, 2000
Please read Section 6.1 of the Patterson & Hennessy text, and be prepared to discuss the following issue:
Rather than jump right into a 5-stage pipelined design, let's look first at a simpler structure. Suppose you want to pipeline Chapter 5's single-cycle design (Figure 5.29, say) by dividing it into just two pipeline stages. How would you do it? You can disregard the control for now, and concentrate on the datapath. What would be the effect on CPI, cycle time, throughput? What should ``CPI'' mean for a pipelined design, anyway?