are there choices in the data path?

First, let me say that the trip to Phoenix and Tucson was successful,
but it is great to be back in Austin.  I am told that one panel I was on
was videotaped, and will be available for your amusement at some point
soon.  The topic was "Single-threaded vs. Multi-threaded," and Professor
Mark Hill from Wisconsin and I took opposite sides of the issue.

Anyhow, back to 360N.  

A student writes:

	Dr. Patt,

	I have a question about the data path for State 12 of the JMP 
	instruction.  Maybe I'm not seeing something, but it seems that 
	there can be two different paths that the data can travel to 
	execute PC<-BaseReg:

I removed the two paths to keep from spoiling the fun for the 
rest of you. :-)

	Will it work either way, and if so, does it matter which we 
	implement in the Control Store?

Yes, it will work either way, and no, it does not matter which you 
implement in the Control Store.

Usually, one designs the data path so only one way makes sense.  But,
it is often the case, such as here, that two ways end up being possible
simply because the microarchitect put in one way to implement one function,
another way to implement a second function, and then later finds out that a
third function can be implemented both ways.  As long as the propagation
delays through both paths complete before the end of the clock cycle, either
way is fine. is the case here.


	<<name withheld to protect the student who is suprised he has a choice>>

Good luck finishing it.
Yale Patt