CS 401 Computer Architectures

Spring 2016
Computer Science
Sabanci University

Announcements

  • February 1: Days, time, and room: Tuesday - 09:40-11:30 – FENS L065, Thursday 15:40-16:30 – FASS G022
  • February 1: See the Syllabus and Schedule
  • February 1: No lab this week.
  • February 1: From now on the announcements will be available through sucourse.

Textbook

John L. Hennessy & David A. Patterson. Computer Organization and Design: The Hardware/Software Interface, 5th Edition, Morgan Kaufmann Publishers, 2014. ISBN: 978-0-12-407726-3.

Recommended Resources

Time & Place

  • Lecture: Tuesday 09:40 – 11:30, FENS L065
  • Lecture: Thursday 15:40-16:30 – FASS G022
  • Lab/Recitation: Friday 12:40 – 13:30, FASS G056

Office Hours

Friday 09:40 – 11:30, FENS 1098 (Office Hours)  

Teaching Assistant

Atıl Utku Ay:  FENS 2014 (time and place to be announced)

Motivation

This is an introductory course on the architecture and organization of computer hardware. The most important objective of the course is to explore the interaction between the hardware organization of modern computers and software, and to reveal the impact of the hardware organization on the performance of the software. Thus, the emphasis in this course will be the basic concepts and techniques that are fundamental for modern computers such as datapath design, pipelining, memory hierarchy etc. MIPS architecture as a representative modern RISC architecture is chosen to explain some fundamental concepts. And also basics of assembly language programming for this architecture will be introduced. More specifically, we show how numbers are represented within a computer and how the circuits that perform arithmetic operations on those numerals are organized. With a high-level overview of digital logic design to support us, we look at how the datapath and control circuits of processors are designed, and in particular we gain some insight for the pipelined processor design, which is the key organizational principle at work in most present-day processors. We examine the use of memory hierarchy (cache memory and virtual memory) to provide the illusion of a large and fast memory from the reality of limited fast memory plus a larger but slower memory. We'll look at input/output devices and buses. Throughout the course there will be an emphasis on the quantitative performance characteristics of computer systems; we'll look at the influence of architecture and organization on performance, and take an introductory look at the empirical and analytical tools appropriate to the study of performance.

Topics

  • Introduction: Computer Abstractions, Technology, Terminology, and History.
  • The Role of Performance: Definition, Measurement and Metrics, Comparison
  • Instructions: Operations of the Computer Hardware, Operands, representation of the Instructions, Procedures, and Addressing.
  • Computer Arithmetic: Signed and Unsigned Numbers, Addition and Subtraction, Logical operations, ALU Construction, Multiplication, Division, Floating Point Arithmetic.
  • Datapath and Control: Building the Datapath, Single-cycle and Multicycle Implementations, Control Design and Microprogramming, Exception Handling.
  • Pipelining: Pipelined datapath, Pipelined Control, Data hazards and Forwarding, Pipeline Stalls, Branch Hazards, Exceptions, Superscalar and Dynamic Pipelining.
  • Memory Hierarchy: Memory Hierarchy, The Basics of Cache, Measuring and Improving Cache Performance, Virtual Memory.
  • I/O: I/O Performance Measures, Types and Characteristics of I/O Devices, Buses, and Interfacing I/O Devices to the Memory, Processor, and Operating System.
  • Multiprocessor Systems: Multicore processors, basics of multicore programming

Homework Assignments

Handwritten assignments are not acceptable

Course Materials

Course materials such as presentation slides and handouts will be available through SUCourse

 

Tentative grading

  • First Midterm: 30 %
  • Final: 40 %
  • HW Assignments: 15%
  • Project: 10 %
  • Participation & Attendance: 5%

Important Dates

·         Midterm: 05 April 2016 (Tuesday) @ 09:40-11:30  

·         Final: As scheduled by Student Office

Homework and Class Projects

Homework assignments and class projects are important parts of the course. The homework assignments involve Assembly language programming for MIPS architectures. SPIM simulator implements almost the entire MIPS assembler-extended instruction set for the R2000/R3000 (It omits some of the complex floating point comparisons and details of the memory system page tables.) The MIPS architecture has evolved considerably since then (in particular, from 32 to 64 bits), which means that SPIM will not run programs compiled for recent MIPS or SGI processors. The students are required to learn SPIM to develop assembly codes for these architectures. Other types of homework assignments such as cache simulation can also be given.

Class projects will involve writing programs that simulate the datapath of MIPS or a similar architecture which implements a selected subset of the instruction set. The instructor may ask you to implement some exotic instructionz that are not available in the original architecture.  

Resources and Pointers

Dr. Erkay Savas