Description
Reading Assignment:
• Chapter 6, of the 4th Edition: pp. 570-572, 575-595 (input-output).
Note: This is from the Fourth Edition of the Patterson & Hennessy book. This chapter is available in
the ‘‘Supplementary Readings’’ part of the class web site.
• Chapter 7, in the textbook by Hamacher, et al.: pp. 228-229.
Note: This is available in the ‘‘Supplementary Readings’’ on the class website, under ‘‘I/O Bus
Structure and PCI Express.’’
• Chapter 5: pp. 381-383 (‘‘Disk Memory’’).
• Chapter 4: pp. 325-332 (‘‘implementing exceptions in the pipelined MIPS’’).
• Appendix A: pp. A-33 − A-37 (Exceptions and Interrupts on MIPS).
• We will use some material from chapter 8 of the 2nd edition of the Patterson and Hennessy book. We
will refer to this as Ed2-Chap8. This material is available on the class web page in the ‘‘Supplementary
Readings’’ section.
− Ed2-Chap8: pp. 646-647 (‘‘Magnetic Disks’’)
− Ed2-Chap8: 656-659 (ignore figures 8.7 and 8.8)
− The examples in Ed2-Chap8: pp. 676-682
− Ed2-Chap8: pp. 662-673
Problems:
(1) A disk has an average seek time of 4ms and rotates at 5,400 RPM. There are 1100 sectors per track.
How long (in seconds) does it take to read the data from three consecutive sectors on the same track?
Explain your answer.
(2) Consider a system where all I/O is done using programmed I/O. The I/O device, CPU, and memory
are connected to the same data bus. The data bus is 64 bits wide. It takes 12 nanoseconds for one
transfer across the bus. What is the maximum possible bandwidth (in bytes per seconds) at which
data can be transferred from the I/O device to memory? Explain your answer!
(3) A program repeatedly performs a three-step process: it reads in a 4 KB block of data from disk, does
some processing on that data, and then writes out the result as another 4 KB block elsewhere on the
disk. Each block is contiguous and randomly located on a single track on the disk. The disk drive
rotates at 12,000 RPM, has an average seek time of 4.5 ms, and has a transfer rate of 80 MB/sec. The
controller overhead is 0.3 ms. No other program is using the disk or processor, and there is no
overlapping of disk operation with processing. The processing step takes 20 million clock cycles,
and the clock rate is 5 GHz. What is the overall speed of the system in blocks processed per second?
(4) An important advantage of interrupts over polling is the ability of the processor to perform other
tasks while waiting for communication from an I/O device. Suppose that a 1.8 GHz processor needs
to read 1000 bytes of data from a particular I/O device. The I/O device supplies 1 byte of data every
0.02 ms. The code to process the data and store it in a buffer takes 1000 cycles.
A) If the processor detects that a byte of data is ready through polling, and a polling iteration takes
60 cycles, how many cycles does the entire operation take?
B) If instead, the processor is interrupted when a byte is ready, and the processor spends the time
between interrupts on another task, how many cycles of this other task can the processor
complete while the I/O communication is taking place? The overhead for handling an interrupt is
200 cycles.
Clearly state your assumptions.
YT Spring 2018
-2-
(5) A pipelined MIPS implementation with support for overflow exceptions is described in the book, pp.
327-331. Consider the following claim: The description in the book is incomplete since it fails to
modify the operation of the Forwarding unit when an exception occurs.
A) Is the claim above correct ? Your answer must be Yes or No.
B) Explain your answer to Part A.
(6) Consider the pipelined MIPS implementation shown in slides 9.30-9.32 and explained in the book,
pp. 327-331. Your task is to enhance this implementation to add support for the illegal opcode
exception. As described on page 327, the Cause register must be modified based on whether the
exception is an illegal opcode (value 10) or an overflow (value 12). Your implementation must work
correctly even for the case where one instruction causes an overflow while the very next instruction
contains an illegal opcode.
A) Provide a list of your modifications, where each one of the modifications is described in 1-2 clear
sentences.
B) The existing datapath implementation is shown on slide 9.30 (available in the Useful Figures
section of the course website). Are any modification to this figure required? If so, you can show
them on a copy of the figure from slide 9.30. If there isn’t enough room on the figure to show the
modifications, just indicate the location of the modifications on the figure and show the details
separately.
C) Are any new control signals required? If so, list them with an explanation and identify them on
the datapath diagram.
D) Changes are required to the main Control circuit. Show those changes using a table similar to the
one shown on slide 9.32. The contents of the table you provide must reflect correct operation
when there are no exceptions, when there is an overflow exception, when there is an illegal
opcode exception, and when overflow and illegal opcode exceptions are detected simultaneously.
Practice problems: You do not need to hand in a solution to the problems below.
(7) Consider the factors that affect the performance of a disk. All other parameters (including total disk capacity)
being equal, would you expect better performance from a disk with fewer platters or more platters. Explain
your answer.
(8) The rotational speed of a disk is doubled. How does that affect the access time of the disk? Your answer must
be as detailed and quantitative as possible.
(9) Consider a disk with the following parameters: 512-byte sectors, rotation speed of 10,000RPM, average seek
time of 6 ms, transfer rate of 50 MB/s, and controller overhead of 0.2 ms. The system described in Ed2-Chap8,
page 665 is used to handle disk accesses. If the I/O is allowed to consume 100% of the bus and memory
bandwidth, what is the maximum number of simultaneous disk transfers that can be sustained for the two
block sizes?
(10) Consider the MIPS implementation shown in Figure 4.66 (page 329) in the book. Due to a fault in the main
Control circuit, whenever the input to the circuit is the opcode of sw (i.e., 101011), regardless of any other
input, the EX.Flush control output is asserted.
The Control circuit and the rest of the processor operate normally in every other respect.
Explain in full detail what will be the consequences of this fault when the processor executes programs —
how will it change the behavior of the processor as observed byauser/programmer who does not know
and does not care how the processor is implemented internally? Be sure to clearly identify each and every
consequence of this fault.
YT Spring 2018