CIT 593: Introduction to Computer Systems Midterm: FALL 2017C/C++

Java Python CIT 593: Introduction to Computer Systems

Midterm: FALL 2017

Part I – Short Answer Questions {20 points} (Provide brief answers to the following questions)

1)   Convert the following decimal #’s into 8-bit 2’s complement binary numbers & their HEX equivalent:

a)   #17       in binary:  0001 0001                          in hex: x11                                            

b)   #-17     in binary:  1110 1111                          in hex: xEF                                            

c)   #128    in binary:  cannot fit in 8-bits           in hex: N/A, but not x80                   

d)   #-128   in binary:  1000 0000                          in hex: x80                                            

2)   The following are 2’s complement binary numbers.  Perform. the following operations & indicate if any of the operations generate overflow.

a)   1100 + 11                                     = 1011                                                    overflow?  NO

b)   100000000 + 110000                = 011110000                                       overflow?  YES

c)   01111 + 011                              = 10010                                                overflow?  YES

d)   10001 - xA                                   = 10111                                                overflow?  NO

3)   The format for a 32-bit Floating Point # is shown below.  For

S (1-bit)	Exponent (8-bits)	Fraction (23-bits)
1	00000111	10100000000000000000000

a)   Is the number positive or negative? _negative                                         

b)   Accounting for the bias (127), what is the actual value of the exponent?

7-127 = -120                                

c)   What is the fractional component in decimal: _101 = 0.5 + .125 = .625                     

d)   E.c. (2 points) Convert the entire # to decimal form

-1.625 x 2-120                                                                                                                                                                               

4)   What is the address space, addressability, and total capacity of the LC4s memory?

Address space = _216                                                                

Addressability = _16-bit                                      

Total capacity = _216  x 16-bit = 65,536 drawers * 16 bits per drawer = 1,048,576 bits =

131,072 bytes = 131kB total capacity                                  

5)   What is the difference between combinational logic and sequential logic?

The output of combinational logic depends only on a combination of its inputs, while the output of sequential logic depends on a combination of its inputs and in some cases its previous output.

6)   What is the difference between a D-latch and a D-flip/flop?

A D-latch’s output follows the input, whenever its WE is high.  A DFF’s only accepts new input at the transition of the clk (from 0 to 1 for a positive edge DFF).

7)   Briefly explain the purpose of feedback in logic circuits.

Feedback connects the output back to the input of a logic circuit, giving it the ability to use its last output as apossible input!  This enables the device to behave as a sequential lo CIT 593: Introduction to Computer Systems Midterm: FALL 2017C/C++ gic   device, and potentially store data from one input combination to the next.

8)   From the perspective of the growth of an ISA, why doesn’t the program memory contain the exact control signals for a CPU?

If the program memory actually stored the full set of control signals, if a future version of the ISA required more hardware to implement new instructions, more control signals would likely be required, expanding the width of program memory and breaking backwards compatibility with the previous processor’s ISA.

9)   If a CPU has a CLK signal with a frequency of 20 Hz, and it takes 30 ms (recall the unit milli = 10-3) for its register file to fetch or store new data, how fast must the ALU be to complete computations within 1 cycle of the clock?

CLK’s period = 1/20Hz = 50ms

Recall that the output of the register file connects to the input of the ALU, and the output of

the ALU back to the register file.  So data must be fetched from the register and the

calculation required for whatever instruction the CPU is working on, must be completed in under 50ms.

Say for a read-only instruction, data takes 30ms to read, the ALU must complete its work in 50-30ms, or just 20ms.

Part II - Long Form. Questions {80 points}:

1) {20 points} For parts (a)-(g) below, refer to the following sequential logic circuit:

a) {2 points} Write out the Boolean logic function that describes the above logic circuit: 0020

C =  (A OR C) AND (B’)

b)   {1 points} If an inverter has a delay of 1ns, an AND gate has a delay of 2ns, and an OR gate has a delay of 3ns, what is the propagation delay of the above logic circuit?

Propagation Delay (worst case): OR + AND = 3ns + 2ns = 5ns (OR and INVrun in parallel, so we only have to consider the longer operation (OR) and the AND gate)

[Best case: INV + AND = 1ns+2ns = 3ns (if C’s last output was 1, then there is no delay with the OR gate and its output (1) is available as the input of the AND gate)]

c)   {5 points} Create atruth table for the above logic circuit; it should have 5 rows and “arrows” indicating any sequence necessary to move from one row to the next.

d)   {1 pt} For each row of your truth table, go back and identify the “actions” of the logic circuit.

e)   {5 points} Generate a timing diagram for this logic circuit; it should encompass all actions from your truth table.

(many possible solutions, left as an exercise for the students)

 f)    {1 pt} Is this logic circuit a latch or a flip/flop?  Why?

A latch because new inputs aren’t accepted as the transition of the change of the inputs, they are accepted anytime the inputs are correct.

 g)   {5 points} Draw out the CMOS implementation for the above logic circuit

(must show the actual feedback wire to get full credit – labeling is not sufficient)

2)   {20 points} The table below shows the contents of a region of User Program Memory in

PennSim. First convert the machine instructions you see here to an equivalent sequence of assembly instructions so you can read them.   Write your answers directly in the tables below.  After you have done this, show what would happen when the program is executed by filling in the second table which shows the state of all of the registers at the start of each instruction cycle. For the register values R0-R7 you only need to fill in the value of the register that has changed from the previous cycle, if any. Keep the PSR in HEX.  For R0-R7 you may use either decimal or HEX (but indicate which is which with a leading # or x).  Hint:   start by looking at the four bit opcode - be careful how you breakup the 16 bit fields, one bit can make a big difference.

NOTE: this column is state ofthe machine when your program begins.

Refer to this column when you execute your first assembly instruction.

3)   {20 points} The 2-bit wide control signal: RegInputMux.CTL is one of few control signals that can be determined solely from the value of the OpCode of any instruction in the LC4-ISA.  Recall that the

Decoder is responsible for examining an instruction’s OpCode and determining the control signals necessary to setup the datapath for the instruction.  In the truth table below, all of the possible LC4 OpCode combinations are listed by category.

a)   {15 points} Use the ISA to determine the value of the 2-bits of RegInputMux.CTL for each instruction category listed and fill in those missing values in the truth table shown below. For any instructions that result in a don’tcare (X), simply fill in with a 0. Some OpCodes are not used in the LC4, you’ll see those combinations blacked out.

The outputs that are highlighted yellow, are the only ones that will be used in the PLA

b)   {5 points} Using the completed truth table, create a PLA that can be used in the decoder to generate the 2-bits of RegInputMux.CTL.  Your PLA should have 4 inputs and 2 outputs