for class on Thursday Oct. 10, 1996
Please read Sections 5.4 and 5.5 of the Patterson & Hennessy text, and be prepared to discuss the following issues:
1. We'll look closely at the design of the multi-cycle datapath (the corrected Figure 5.35 in the handout). Compared with the single-cycle design of last two classes, this design has some new registers. Please consider the logical rationale for such registers: are they needed for correct functionality, for better performance, for improved pedagogy, or what? The first edition of this book had just two new registers: the instruction register and a register for holding branch targets.
2.There are two methods for controlling this design, finite-state machine and microprogramming. These methods seem very similar. Are they really the same, or are there subtle differences? (There seem to be some unintentional differences between the two methods, probably due to the addition of all those registers--perhaps each author revised one section!)