CS 354, version A
Spring 2006 		Name:___________________
ID:___________________
Login:_____________________
Exam 3

No electronic devices may be used while taking this exam. Examples of devices not allowed are calculators, pagers, cell phones, wrist calculators/computers, laptop computers, pocket computers. Each student is allowed one 8.5 by 11 inch sheet of paper with handwritten notes. The notes may be on both sides of the paper.

Show all work, and do any/all calculations on the exam. Extra scratch paper may not be used.

Exam Score
Q1 = _____ /   6
Q2 = _____ /   8
Q3 = _____ /   5
Q4 = _____ /   6
Q5 = _____ /   8
Q6 = _____ /   5
Q7 = _____ /   8
Q8 = _____ /   8
Q9 = _____ / 12
Total = _____ / 58

Question 1 (6 points total)
Give a TAL assembly language synthesis for each of the following MIPS pseudoinstructions. Assume that label X has been assigned address 0x00ff3300. 


  sw   $8, X




  rem  $10, $11, $12

Question 2 (8 points total)
This MIPS code fragment is to be used inside of an extremely simple kernel (that uses only spin wait loops; no interrupts) to implement the puts syscall. 


_k_put_s:  lbu  $k0, ($a0)

           beqz $k0, _k_exit_put_s

           sw   $k0, Display_Data

           add  $a0, $a0, 1

           b    _k_put_s

_k_exit_put_s:
           # "return" from syscall code here
Part A (3 points)
(Briefly) Explain why this code fragment does not correctly implement the puts syscall.





Part B (5 points)
Show how to modify the code fragment such that it does correctly implement the puts syscall.

Question 3 (5 points) The standard formula for calculating the time it takes to read or write a sector on a disk is

time = seek time + rotational latency + read/write time

Part A (2 points) Which of these 3 components (most often) takes the longest?


Part B (3 points) (Briefly) Why?



Question 4 (6 points) Within our kernel (the one used for Assignment 8), there is code to extract the cause (ExcCode), and use a switch statement to effect a multiway branch. This code is 


    mfc0  $k0, 13
    andi  $k0, $k0, 0x3c
    lw    $k0, _k_JumpTable($k0)
    jr    $k0
Write a TAL assembly language synthesis of the load (lw) instruction. Assume that the label _k_JumpTable is assigned address 0x810000c0.





Question 5 (8 points) The MIPS assembly language instruction 


    beq $8, $9, loop25
is assigned by the assembler to reside at address 0x00400a60. The machine code for this instruction is 0x1109fffa.

Part A (3 points) What is the (hexadecimal) address of the next instruction to be fetched and executed if the branch is NOT taken?




Part B (3 points) What is the (hexadecimal) address of the next instruction to be fetched and executed if the branch is taken?








Part C (2 points) When the branch is taken, does this go forward within the code or backward within the code?



Question 6 (5 points)
Classify each of the following events as generating either a trap or an interrupt.



Question 7 (8 points total)
Part A (4 points)
Was the kernel used for assignment 8 (the logical AND test program) reentrant or nonreentrant?

Part B (4 points)
Suppose that a MIPS computer system has a simple, nonreentrant kernel. An application has been executing, and the application executes the syscall instruction. As the first instruction within the kernel is fetched, diagram the values in the four (4) least significant bits of the Status register.







Question 8 (8 points total)
Part A (4 points) Name the 2 categories of dependencies.

1.

2.

Part B (4 points)
(Briefly) Why does the engineer designing a processor need to know about dependencies?









Question 9 (12 points total)
The Reallygood Company is putting an L1 cache that holds only instructions (not data) into their computer system that has a MIPS processor.


Part A (2 points)
How many MIPS instructions can this L1 instruction cache hold?



Part B (2 points)
How many Valid bits does this cache have?



Part C (8 points)
Consider the following MIPS code fragment that implements a loop. If only this code fragment is executed, show your work to calculate the cache's hit ratio. Since we are not using calculators, leave your answer as a ratio. 

                          # familiar MAL equivalents in comments
    addi $t0, $0, 1000    # li $t0, 1000
    addi $t1, $0, -12     # li $t1, -12
    addi $t2, $0, 500     # li $t2, 500
loop:
    beq  $t0, $0, loop_done
    mult $t2, $t1
    mflo $t3
    addi $t3, $t3, 13
    div  $t3, $t1
    mflo $t4
    srl  $t2, $t1, 4
    addi $t0, $t0, -1     # sub $t0, $t0, 1
    beq  $0 $0, loop      # b   loop
loop_done: