Introduction toComputer Architecture

Topics:Topics: Course policies and overview Theme Five great realities of computer systems

Powerpoint Lecture Notes for Computer Systems: A Programmer'sPerspective, R. Bryant and D. O'Hallaron, Prentice Hall, 2003

CISC 360-010

Michela TauferSeptember 4, 2008

– 2 – CISC 360-010, Fa08

Meet your Colleagues…Ask your classmate: His/Her name? What are his/her favorite topics in computer

science? List up to three topics. What is he/she planning to do after his/her

graduation?

– 3 – CISC 360-010, Fa08

Course Theme Abstraction is good, but donʼt forget reality!

Courses to date emphasize abstractionCourses to date emphasize abstraction Abstract data types Asymptotic analysis (big oh)

These abstractions have limitsThese abstractions have limits Especially in the presence of bugs Need to understand underlying implementations

Useful outcomesUseful outcomes Become more effective programmers

Able to find and eliminate bugs efficiently Able to tune program performance

Prepare for later “systems” classes in CS & ECE Compilers, Operating Systems, Networks, Embedded Systems

– 4 – CISC 360-010, Fa08

Teaching Staff Instructors

Prof. Michela Taufer (2:30PM to 3:30PM or by appointment)

TAʼs Yang Gang

– 5 – CISC 360-010, Fa08

Textbooks (Mandatory)Randal E. Bryant andRandal E. Bryant and

David R. David R. OO ʼ̓HallaronHallaron,, “Computer Systems: A

ProgrammerʼsPerspective”, PrenticeHall 2003.

csapp.cs.cmu.edu

– 6 – CISC 360-010, Fa08

Textbooks (Recommended)Brian Kernighan and DennisBrian Kernighan and Dennis

Ritchie,Ritchie, “The C Programming

Language, SecondEdition”, Prentice Hall,1988

– 7 – CISC 360-010, Fa08

Course ComponentsLecturesLectures

Higher level concepts

QuizzesQuizzes Applied concepts, important tools and skills for labs

ProjectsProjects Four lab assignments The heart of the course ~2 weeks Provide in-depth understanding of an aspect of systems Programming and measurement

– 8 – CISC 360-010, Fa08

Getting HelpCourse webpage at Course webpage at http:http://cis//cis..udeludel..edu/~taufer/coursesedu/~taufer/courses

Copies of lectures and project assignments Clarifications to assignments, deadlines Syllabus and class schedule

User:User: cisc360studentcisc360studentPassword: Password: Learn4Fun!Learn4Fun!

DiscussionsDiscussions through mailing listthrough mailing list Clarifications to assignments, general discussion Send e-mail to all with: cisc360010_fa08@gcl.cis.udel.edu

Personal helpPersonal help Professor: lecture, exams, quizzes, grades TA: lab assignments

– 9 – CISC 360-010, Fa08

Policies: Quizzes Quizzes test your understanding of the text

You are expected to do the reading assignmentYou are expected to do the reading assignmentBEFOREBEFORE the quizthe quiz

There are There are NONO make-up quizzes make-up quizzes

– 10 – CISC 360-010, Fa08

Policies: Lab AssignmentsSubmissionsSubmissions

Assignments due at 11:59PM on specified duedate

Electronic submissions ONLY - procedure isexplained in assignment

Makeup exams and assignmentsMakeup exams and assignments OK, but must make PRIOR arrangements with

Prof. Taufer.Appealing gradesAppealing grades

Within 7 days of due date or exam date

– 11 – CISC 360-010, Fa08

CheatingWhat is cheating?What is cheating?

Sharing code: either by copying, retyping, looking at, orsupplying a copy of a file.

What is NOT cheating?What is NOT cheating? Helping others use systems or tools. Helping others with high-level design issues. Helping others debug their code.

Penalty for cheating:Penalty for cheating: Removal from course with failing grade.

– 12 – CISC 360-010, Fa08

Policies: GradingExams (38%)Exams (38%)

Two in class exams (10% each) Final (28%) All exams are open book

Projects (40%)Projects (40%) 4 project (10% each)

Quizzes (12%)Quizzes (12%)

Grading CharacteristicsGrading Characteristics Lab scores tend to be high

Serious handicap if you donʼt hand a lab in Tests typically have a wider range of scores

– 13 – CISC 360-010, Fa08

FacilitiesAssignments will use Linux OS machinesAssignments will use Linux OS machines

McKinly Lab  Lab 051 - 22 Linux systems for CIS labs Remote SSH Alias: hplab.acad.cis.udel.edu

– 14 – CISC 360-010, Fa08

Schedule of the Next Two WeeksClass 01 - Introduction (Sep 4)

o Course Overviewo Course Organizationo Reading chap 1 assigned

Class 02 - Bits and Bytes (Sep 9)o Reading chap 2 assignedo Quiz chap 1

Class 03 - Integers (Sep 11)

– 15 – CISC 360-010, Fa08

Course PerspectiveMost Systems Courses are Builder-CentricMost Systems Courses are Builder-Centric

Computer Architecture Design pipelined processor in Verilog

Operating Systems Implement large portions of operating system

Compilers Write compiler for simple language

Networking Implement and simulate network protocols

– 16 – CISC 360-010, Fa08

Course Perspective (Cont.)This Course is Programmer-CentricThis Course is Programmer-Centric

Purpose is to show how by knowing more about theunderlying system, one can be more effective as aprogrammer

Enable you to Write programs that are more reliable and efficient Incorporate features that require hooks into OS

» E.g., concurrency, signal handlers Not just a course for dedicated hackers

We bring out the hidden hacker in everyone Cover material in this course that you wonʼt see elsewhere

– 17 – CISC 360-010, Fa08

Programs and Data (9)

TopicsTopics Bits operations, arithmetic, assembly language programs,

representation of C control and data structures Includes aspects of architecture and compilers

Lab AssignmentsLab Assignments L1: Manipulating bits L2: Defusing a binary bomb

– 18 – CISC 360-010, Fa08

Computer Architecture (6)TopicsTopics

Computer architecture and logic design,pipelinedarchitecture

Develop understanding of processor design and the closerelationship between software and hardware

Lab AssignmentsLab Assignments L3: Design and implementation of a pipelined Y86 processor

– 19 – CISC 360-010, Fa08

Performance (2)

TopicsTopics High level processor models, code optimization (control and

data) Includes aspects of architecture, compilers, and OS

Lab AssignmentsLab Assignments L4: Optimizing Code Performance

– 20 – CISC 360-010, Fa08

The Memory Hierarchy (2)

TopicsTopics Memory technology, memory hierarchy, caches, disks,

locality Includes aspects of architecture and OS.

AssignmentsAssignments L4: Optimizing Code Performance

– 21 – CISC 360-010, Fa08

Virtual memory (4)

TopicsTopics Virtual memory, address translation, dynamic storage

allocation Includes aspects of architecture and OS

– 22 – CISC 360-010, Fa08

Project RationaleEachEach project should have a well-defined goal such as solving aproject should have a well-defined goal such as solving a

puzzle or winning a contest.puzzle or winning a contest. Defusing a binary bomb.

Doing aDoing a project should result in new skills and conceptsproject should result in new skills and concepts Data Lab: computer arithmetic, digital logic. Bomb Lab: assembly language, using a debugger,

understanding the stack Arch Lab: learn about the design and implementation of a

pipelined Y86 processor, optimizing its performance on abenchmark Y86 array copy program called ncopy.ys

Perf Lab: profiling, measurement, performance debugging.

– 23 – CISC 360-010, Fa08

Great Reality #1IntIntʼ̓s s are not Integers, Floatare not Integers, Floatʼ̓s are not s are not RealsReals

ExamplesExamples Is x2 ≥ 0?

Floatʼs: Yes! Intʼs:

» 40000 * 40000 --> 1600000000» 200 * 300 * 400 * 500 --> ??

Is (x + y) + z = x + (y + z)? Unsigned & Signed Intʼs: Yes! Floatʼs:

» (1e20 + -1e20) + 3.14 --> 3.14» 1e20 + (-1e20 + 3.14) --> ??

– 24 – CISC 360-010, Fa08

Computer ArithmeticDoes not generate random valuesDoes not generate random values

Arithmetic operations have important mathematicalproperties

Cannot assume Cannot assume ““usualusual”” properties properties Due to finiteness of representations Integer operations satisfy “ring” properties

Commutativity, associativity, distributivity Floating point operations satisfy “ordering” properties

Monotonicity, values of signs

ObservationObservation Need to understand which abstractions apply in which

contexts Important issues for compiler writers and serious application

programmers

– 25 – CISC 360-010, Fa08

Great Reality #2YouYouʼ̓ve got to know assemblyve got to know assembly

Chances are, youChances are, youʼ̓ll never write program in assemblyll never write program in assembly Compilers are much better & more patient than you are

Understanding assembly key to machine-levelUnderstanding assembly key to machine-levelexecution modelexecution model Tuning program performance

Understanding sources of program inefficiency Behavior of programs in presence of bugs

High-level language model breaks down Implementing system software

Compiler has machine code as target Operating systems must manage process state

– 26 – CISC 360-010, Fa08

Assembly Code ExampleTime Stamp CounterTime Stamp Counter

Special 64-bit register in Intel-compatible machines Incremented every clock cycle Read with rdtsc instruction

ApplicationApplication Measure time required by procedure

In units of clock cycles

double t;start_counter();P();t = get_counter();printf("P required %f clock cycles\n", t);

– 27 – CISC 360-010, Fa08

Code to Read Counter Write small amount of assembly code using GCCʼs asm

facility Inserts assembly code into machine code generated by

compilerstatic unsigned cyc_hi = 0;static unsigned cyc_lo = 0;

/* Set *hi and *lo to the high and low order bits of the cycle counter.*/void access_counter(unsigned *hi, unsigned *lo){ asm("rdtsc; movl %%edx,%0; movl %%eax,%1"

: "=r" (*hi), "=r" (*lo):: "%edx", "%eax");

}

– 28 – CISC 360-010, Fa08

Code to Read Counter/* Record the current value of the cycle counter. */void start_counter(){ access_counter(&cyc_hi, &cyc_lo);}

/* Number of cycles since the last call to start_counter. */double get_counter(){ unsigned ncyc_hi, ncyc_lo; unsigned hi, lo, borrow; /* Get cycle counter */ access_counter(&ncyc_hi, &ncyc_lo); /* Do double precision subtraction */ lo = ncyc_lo - cyc_lo; borrow = lo > ncyc_lo; hi = ncyc_hi - cyc_hi - borrow; return (double) hi * (1 << 30) * 4 + lo;}

– 29 – CISC 360-010, Fa08

Measuring TimeTrickier than it Might LookTrickier than it Might Look

Many sources of variation

ExampleExample Sum integers from 1 to n

n Cycles Cycles/n100 961 9.61

1,000 8,407 8.411,000 8,426 8.43

10,000 82,861 8.2910,000 82,876 8.29

1,000,000 8,419,907 8.421,000,000 8,425,181 8.43

1,000,000,000 8,371,2305,591 8.37

– 30 – CISC 360-010, Fa08

Great Reality #3Memory MattersMemory Matters

Memory is not unboundedMemory is not unbounded It must be allocated and managed Many applications are memory dominated

Memory referencing bugs especially perniciousMemory referencing bugs especially pernicious Effects are distant in both time and space

Memory performance is not uniformMemory performance is not uniform Cache and virtual memory effects can greatly affect program

performance Adapting program to characteristics of memory system can

lead to major speed improvements

– 31 – CISC 360-010, Fa08

Memory Referencing Bug Example

main (){ long int a[2]; double d = 3.14; a[2] = 1073741824; /* Out of bounds reference */ printf("d = %.15g\n", d); exit(0);}

Alpha MIPS Linux-g 5.30498947741318e-315 3.1399998664856 3.14

-O 3.14 3.14 3.14

(Linux version gives correct result, butimplementing as separate function givessegmentation fault.)

– 32 – CISC 360-010, Fa08

Memory Referencing ErrorsC and C++ do not provide any memory protectionC and C++ do not provide any memory protection

Out of bounds array references Invalid pointer values Abuses of malloc/free

Can lead to nasty bugsCan lead to nasty bugs Whether or not bug has any effect depends on system and

compiler Action at a distance

Corrupted object logically unrelated to one being accessed Effect of bug may be first observed long after it is generated

How can I deal with this?How can I deal with this? Program in Java, Lisp, or ML Understand what possible interactions may occur Use or develop tools to detect referencing errors

– 33 – CISC 360-010, Fa08

Memory Performance ExampleImplementations of Matrix MultiplicationImplementations of Matrix Multiplication

Multiple ways to nest loops

/* ijk */for (i=0; i<n; i++) { for (j=0; j<n; j++) { sum = 0.0; for (k=0; k<n; k++) sum += a[i][k] * b[k][j]; c[i][j] = sum; }}

/* jik */for (j=0; j<n; j++) { for (i=0; i<n; i++) { sum = 0.0; for (k=0; k<n; k++) sum += a[i][k] * b[k][j]; c[i][j] = sum }}

– 34 – CISC 360-010, Fa08

0

20

40

60

80

100

120

140

160

matrix size (n)

ijkikjjikjkikijkji

Matmult Performance (Alpha 21164)Too big for L1 Cache Too big for L2 Cache

– 35 – CISC 360-010, Fa08

Blocked matmult perf (Alpha 21164)

0

20

40

60

80

100

120

140

160

50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500matrix size (n)

bijkbikjijkikj

– 36 – CISC 360-010, Fa08

Great Reality #4ThereThereʼ̓s more to performance than asymptotics more to performance than asymptotic

complexitycomplexityConstant factors matter too!Constant factors matter too!

Easily see 10:1 performance range depending on how codewritten

Must optimize at multiple levels: algorithm, datarepresentations, procedures, and loops

Must understand system to optimize performanceMust understand system to optimize performance How programs compiled and executed How to measure program performance and identify

bottlenecks How to improve performance without destroying code

modularity and generality

– 37 – CISC 360-010, Fa08

Great Reality #5Computers do more than execute programsComputers do more than execute programs

They need to get data in and outThey need to get data in and out I/O system critical to program reliability and performance

They communicate with each other over networksThey communicate with each other over networks Many system-level issues arise in presence of network

Concurrent operations by autonomous processes Coping with unreliable media Cross platform compatibility Complex performance issues

– 38 – CISC 360-010, Fa08

Good Luck!