CSCI 2321 (Computer Design), Spring 2018: 
Homework 1
- Credit:
- 40 points.
 
Be sure you have read, or at least skimmed, the assigned readings
from Chapter 1.
Please include with each part of the assignment the Honor Code pledge or
just the word ``pledged'', plus one or more of the following about
collaboration and help (as many as apply).1Text in italics is explanatory or something for you to
fill in.
For written assignments, it should go right after your name and
the assignment number; for programming assignments, it should go
in comments at the start of your program(s).
- This assignment is entirely my own work.
	(Here, ``entirely my own work'' means that it's
	your own work except for anything you got from the
	assignment itself -- some programming assignments
	include ``starter code'', for example -- or 
	from the course Web site.
	In particular, for programming assignments you can
	copy freely from anything on the ``sample programs page''.)
- I worked with names of other students on this
	assignment.
- I got help with this assignment from 
	source of help -- ACM
	tutoring, another student in the course, the instructor, etc.
	(Here, ``help'' means significant help,
	beyond a little assistance with tools or compiler errors.)
- I got help from outside source -- 
	a book other than the textbook (give title and author),
	a Web site (give its URL), etc..
	(Here too, you only need to mention significant help --
	you don't need to tell me that you
	looked up an error message on the Web, but if you found 
	an algorithm or a code sketch, tell me about that.)
- I provided help to names of students on this
	assignment.
	(And here too, you only need to tell me about
	significant help.)
Answer the following questions.  You may write out your answers by
hand or using a word processor or other program, but please submit
hard copy, either in class or in one of my mailboxes (outside my
office or in the ASO).
 
Tips:
- If the assignment asks you to do one or more problems from
	the textbook,
	be sure you get them from the edition specified in
	the syllabus;
	these sets of problems change from edition to edition.
- If a question requires you to do calculations,
	please show enough work to help me understand how you got
	the answer you did, so if you make a mistake I can give partial	
	credit for anything you did get right.
	Also,
	it's a good idea to apply to any result a ``reasonableness
	test'' -- for example, if a processor's clock rate is
	in the GHz range and it's executing a program with billions
	of instructions, execution time is likely to be in the
	range of a few seconds, and if it's different by a couple
	of orders of magnitude you probably made a mistake!
- (10 points)
	Below are two lists, one of terms from Chapter 1
	and one of short definitions/descriptions.
	For each definition say which term matches it best.
 Terms:
 
	 
| • desktop computer |  | • server |  | • low-end server |  | • supercomputer |  | • embedded computer |  | • multicore processor |  | • RAM |  | • CPU |  | • operating system |  | • compiler |  | • bit |  | • instruction |  | • assembly language |  | • machine language |  | • assembler |  | • high-level language |  | • system software |  | • application software |  
 
 
Definitions/descriptions:
 
 
- Computer used to run large problems and
		usually accessed via a network.
		
- Computer composed of hundreds or even thousands	
		of processor and terabytes of memory and having
		the highest performance and cost.
		
- Random access memory.
		
- Central processing unit.
		
- Microprocessor containing several processors in
		the same chip.
		
- Desktop-class computer without a screen or keyboard
		and usually accessed via a network.
		
- Computer used for running one predetermined
		application or collection of applications,
		often found as part of another piece of hardware.
		
- Personal computer delivering good performance to
		a single user at low cost.
		
- Program that translates statements in high-level
		language to assembly language.
		
- Program that translates symbolic instructions
		to binary instructions
		
- Binary language that a processor can understand.
		
- Command that a processor understands.
		
- Symbolic representation of machine instructions.
		
- Interface between user's program and 
		hardware providing a variety of services
		and supervision functions.
		
- Software/programs developed by a user.
		
- Binary digit (value 0 or 1).
		
- Software layer between the application software
		and the hardware that includes
		the operating system and compilers.
		
- Portable language usually composed of words
		and algebraic expressions that must be translated
		into assembly language before being run on 
		a computer.
 
 
- (10 points)
	Suppose two processors implement a given instruction set
	architecture,
	in which the instructions can be grouped into four classes
	A, B, C, and D.
	Processor P1 has a clock rate of 2.5GHz,
	and the CPI (cycles per instruction) for the four classes
	are 1, 2, 3, and 3 respectively.
	Processor P2 has a clock rate of 3GHz and a CPI of 2 for
	all classes.
	For a program that at runtime executes 10E9 ( )
	instructions, of which 
	10% are class A,
	20% class B,
	50% class C,
	and 20% class D,
	answer the following: )
	instructions, of which 
	10% are class A,
	20% class B,
	50% class C,
	and 20% class D,
	answer the following:
 
- How many seconds does this program take on P1?  on P2?
 
 
 
- If ``global CPI'' is an average CPI
		over all four classes, calculate global CPI
		for this program on both P1 and P2.
		(Note here that this average is probably best
		computed as total cycles divided by total
		instructions, 
		to allow for the different frequencies
		of the different classes of instructions.)
 
 
 
 
 
- (10 points)
	
 Review the scenario described in problem 1.9
	in the textbook (a program being executed on 
	a particular multiprocessor system)
	and answer the following questions:
 
- Assuming there is what you might call an original
		sequential program that just executes all the
		instructions mentioned in the description
		as well as a parallel program for which the number
		of instructions per processor is as described,
		calculate execution time for the sequential
		program and for the parallel program running
		on 1, 2, 4, and 8 processors,
		and compute speedups 
		relative to the sequential program
		(i.e., sequential time divided by time on  processors). processors).
Suggestion:  
		The calculations here are kind of tedious.
		If you like programming you might consider
		writing a short throw-away program to help you,
		in whatever language appeals to you.
		If you do, include a copy of your source code;
		that will be the equivalent of ``showing your work''.
		(Note that while normally 
		I prefer to get source code by e-mail,
		since I don't plan to test anything you write for
		this problem it will be simpler just to turn in
		hardcopy.)
 
 
 
 
- What would happen if the clock rate were changed
		to 2.5GHz but arithmetic instructions now had
		a CPI of 2?  
		(Recalculate all the times in the	
		first part of the problem and also the speedups.)
 
 
 
- For up to 2 extra-credit points, 
		plot the calculated speedups and a line showing
		what linear speedup would look like.
		(If you don't already have a favorite program
		for making plots, ask me about gnuplot.)
 
 
 
 
- (10 points)
	Do problems 1.12.1, 1.12.2, and 1.12.3 from the textbook.
	(Note that you will need information from the paragraph
	numbered 1.12.)
 
 
	
Footnotes
- ... apply).1
- 
Credit where credit is due:  
I based the wording of this list on a posting to a SIGCSE mailing
list.  SIGCSE is the ACM's Special Interest Group on CS Education.
 
Berna Massingill 
2018-01-20