Computer Abstractions and Technology: Diego Montero

Notes
1. Computer Abstractions and Technology
Diego Montero
Slide Sources: Patterson & Hennessy MK 3rd and 5th edition, adapted and supplemented
Facultad de Ingenierı́a
Semestre Septiembre2021 - Marzo2022
Introduction
Outline Notes
1 Introduction
Classes of Computing Applications and Their Characteristics
Welcome to the PostPC Era
Objectives of this course
Understanding Program Performance
2 Eight Great Ideas in Computer Architecture
3 Below Your Program

From a High-Level Language to the Language of Hardware
4 Under the Covers

Opening the Box
5 Performance
Defining Performance
Measuring Execution Time
Instruction Count and CPI
Montero, Diego (OAC-18586) Abstractions and Technology 1 / 49

Introduction
Relevance Notes
Computers have led to a third revolution for civilization
Applications that were economically infeasible suddenly become

practical
Computers in automobiles
Cell phones
Human genome project
World Wide Web
Search engines
Introduction
Relevance Notes
Advances in this technology affects almost every aspect of our society
Hardware advances → programmers create wonderfully useful software
Today’s science fiction suggests tomorrow’s killer applications

self-driving cars
cashless society: bitcoins
augmented reality
artificial intelligence

Introduction Classes of Computing Applications
Classes of Computing Applications Notes
Personal Computer (PCs)

single user: incorporates a display, keyboard and a mouse
good performance, low cost
merely 40 years old
Servers
multiple users, carry sizable workloads
accessed via a network
resources expandable, dependability
Supercomputers: high-end servers

highest performance and cost: tens of thousands of processors and
many terabytes of memory
high-end scientific and engineering calculations: weather forecasting, oil
exploration, protein structure
Introduction Classes of Computing Applications
Embedded computers
designed to run one or a set of related applications
integrated with the hardware
largest class of computers and span the widest range of applications

and performance
found in your car, the computers in a television set, and the networks
of processors that control a modern airplane or cargo ship

Introduction PostPC
Welcome to the PostPC Era Notes
Figure: The number manufactured per year of tablets and smart phones, which
reflect the post-PC era, versus PC and traditional cell phones.
Introduction PostPC
Personal mobile device (PMD)

battery operated
wireless connectivity
smart phones, tablets
Cloud Computing
Warehouse Scale Computers (WSC)
Software as a Service (SaaS)
Portion of software run on a PMD and a portion run in the Cloud
Amazon, Google

Introduction Objectives
Objectives Notes
How are programs written in a high-level language, such as C or Java,

translated into the language of the hardware, and how does the
hardware execute the resulting program?
What is the interface between the software and the hardware, and
how does software instruct the hardware to perform needed functions?
What determines the performance of a program, and how can a

programmer improve the performance?
What techniques can be used by hardware designers to improve

performance?
Introduction Objectives
Objectives Notes
The answers to these questions will enable us to:

Improve the performance of a program
Select a specific hardware to execute a specific application
The rest of the course content:

basic ideas and definitions
Hardware and Software components
Performance evaluation

Introduction Performance
Performance of a program Notes
Effectiveness of the algorithms used

Software systems used to create and translate the program into
machine instructions
Effectiveness of the computer in executing those instructions (I/O
operations may be included)
Performance of a program Notes
Hardware or software How this component affects

component performance
Determines both the number of
source-level statements and the
Algorithm
number of I/O operations
executed
Determines the number of computer
Programming language,
instructions for each source-level
compiler, and architecture
statement
Processor and memory Determines how fast instructions
system can be executed
I/O system (hardware and Determines how fast I/O operations
operating system) may be executed

Exercise Notes
Can you think of examples where each of the following is the right place to
look for a performance bottleneck?
The algorithm chosen
The programming language or compiler
The operating system
The processor
The I/O system and devices
Ideas
Outline Notes
1 Introduction

4 Under the Covers

Opening the Box
5 Performance

Ideas
Eight Great Ideas in Computer Architecture Notes
1 Design for Moore’s Law: integrated circuit resources double every

18–24 months
2 Use Abstraction to Simplify Design: design at different levels of
representation; lower-level details are hidden→simpler model at higher
levels higher levels
3 Make the Common Case Fast
4 Performance via Parallelism
5 Performance via Pipelining
6 Performance via Prediction
7 Hierarchy of Memories
8 Dependability via Redundancy
Below Your Program
Outline Notes
1 Introduction

4 Under the Covers

Opening the Box
5 Performance

Below Your Program
Below Your Program Notes
Typical application → millions of lines of code and rely on

sophisticated libraries
The hardware in a computer can only execute extremely simple
low-level instructions
From complex application to simple instructions
several layers of software interpret/translate high-level operations into
simple computer-specific instructions
abstractions
Below Your Program
Below Your Program Notes
Figure: A simplified view of hardware and software as hierarchical layers.

Below Your Program
Systems software Notes
Operating system
Interfaces between a user’s program and the hardware
Handles basic input and output operations
Allocates storage and memory
Provides protected sharing of the computer among multiple
applications using it simultaneously.
Linux, iOS, Android and Windows
Compilers
translation of a program written in a high-level language, such as C,
C++, Java, or Visual Basic into instructions that the hardware can
execute
Below Your Program High-Level to Hardware
From a High-Level Language to the Language of Hardware Notes
Computers understand “instructions”

collections of bits that the computer understands and obeys
1000110010100000 → ADD two numbers
First programmers communicated to computers in binary numbers
Assembler: translates a symbolic version of an instruction into the

binary version
add A, B → 1000110010100000
Machine language = binary language
Assembly language = symbolic language

write one line for every instruction

From a High-Level Language to the Language of Hardware Notes
Figure:
Montero,CDiego
program compiled intoAbstractions
(OAC-18586) assembly language and then assembled into
and Technology 20 / 49
binary machine language
High-Level programming languages Notes
Allow the programmer to think in a more natural language
Improve programmer productivity.
Allow programs to be independent of the computer on which they

were developed
Compilers and assemblers translate high-level language programs to

the binary instructions

Under the Covers
Outline Notes
1 Introduction

4 Under the Covers

Opening the Box
5 Performance
Under the Covers
Under the Covers Notes
The hardware in any

computer performs the same
basic functions
inputting data
outputting data
processing data
storing data
The five classic components

of a computer
Input
Output Figure: The organization of a
Memory computer, showing the five classic
Data and control path, or components.
processor

Under the Covers Opening the Box
Opening the Box Notes
Figure: Components of the Apple iPad 2 A1395.
Figure: Mainboard of the Apple iPad 2 A1395.
ARM processor 1 GHz

512 MB memory
32 GB flash drive

Mainboard
connects to I/O devices
Memory
The processor
CPU central processor unit

datapath: performs aritmetic operations
control path: tells the datapath, memory and I/O devices what to do
according to the instructions
Memory
built with DRAM chips
Figure: The processor integrated circuit inside the A5 package.

Cache Memory
small and fast memory
act as a buffer for the DRAM
SRAM technology: faster but less dence → more expensive
Storage
Main memory, volatile, DRAM
non-volatile storage: Flash drives, hard drives
Instruction Set Architecture Notes
ISA (Instruction Set Architecture) or Architecture
Interface hardware ↔ lowest-level software
Includes anything programmers need to know to make a binary

machine language program work correctly
OS encapsulate I/O, memory allocation, and other low-level system

functions
ISA + OS = Application binary interface (ABI)
ISA allows us to talk about functions independently from the hardware

Abstractions Notes
Both hardware and software consist of hierarchical layers using abstraction,

with each lower layer hiding details from the level above.
One key interface between the levels of abstraction is the instruction set
architecture—the interface between the hardware and low-level software.
This abstract interface enables many implementations of varying cost and
performance to run identical software.
Recap
Recap Notes
Code generation, translation and execution
Hardware components
Abstractions
ISA
Raspberry PI ISA: ARM Cortex-A72, ARMv8-A 64-bit instruction set.
Arduino UNO ISA: ATmega328P, AVR enhanced RISC architecture.

Advanced RISC Machine (ARM) is a processor architecture based on a
32-bit reduced instruction set (RISC) computer.

Performance
Outline Notes
1 Introduction

4 Under the Covers

Opening the Box
5 Performance
Performance Defining Performance
Performance Notes
One computer has better performance than another?

Which airplane has the best performance?
Passenger Cruising range Cruising speed Passenger throughput
Airplane
capacity (miles) (m.p.h) (passengers x m.p.h.)
Boeing 737 240 3000 564 135,360
BAC/Sud Concorde 132 4000 1350 178,200
Boeing 777-200LR 301 9395 554 166,761
Airbus A380-800 853 8477 587 500,711

Performance Notes
Comparing two different desktop PC

Response time or execution time: time for task completion (metric)
Datacenter
Throughput or bandwidth: total amount of work done in a given
time (metric), e.g., tasks/transactions/... per hour
Response Time and Throughput: Example Notes
How are response time and throughput affected by:
1 Replacing the processor in a computer with a faster version?
2 Adding more processors?
We will focus on response time . . .

Relative Performance Notes
For a computer X:
1
PerformanceX =
Execution timeX
For two computers X and Y, if the performance of X is greater than Y:
PerformanceX > PerformanceY
Execution timeY > Execution timeX
PerformanceX Execution timeY

= =n
PerformanceY Execution timeX
If X is n times as fast as Y, then the execution time on Y is n times as
long as it is on X.
Relative Performance: Example Notes
If computer A runs a program in 10 seconds and computer B runs the

same program in 15 seconds, how much faster is A than B?
PerformanceA Execution timeB
= =n
PerformanceB Execution timeA
15
= 1.5
10
A is therefore 1.5 times as fast as B.

Performance Measuring Execution Time
Measuring Execution Time Notes
Elapsed time
Total response time, including: processing, I/O, OS overhead, idle time
Determines system performance
CPU time
Time spent processing a given job — discounts I/O time, other jobs’
shares
Comprises user CPU time and system CPU time
Different programs are affected differently by CPU and system

performance
Focus on this one
CPU Clocking Notes
Operation of digital hardware governed by a a constant-rate clock
Clock period: duration of a clock cycle

e.g., 250ps = 0.25ns = 250Ö10–12s
Clock frequency (rate): cycles per second

e.g., 4.0GHz = 4000MHz = 4.0 times109 Hz

CPU Time Notes
CPU execution time CPU clock cycles

= × Clock cycle time
for a program for a program
CPU execution time CPU clock cycles for a program

=
for a program Clock rate
Performance improved by:

reducing the number of clock cycles required for a program
Increasing clock rate
Hardware designer must often trade off clock rate against cycle count
CPU Time: Example Notes
Computer A: 2GHz clock, 10s CPU time
Designing Computer B
Objective: 6s CPU time
Can do faster clock, but causes 1.2 × clock cycles of A

CPU Time: Example Notes
CPU clock cyclesA

CPU timeA =
Clock rateA
CPU clock cyclesA
10 seconds = cycles
2 × 109 second
cycles
CPU clock cyclesA = 10 seconds × 2 × 109 = 20 × 109 cycles
second
Clock rate for B?
1.2 × CPU clock cyclesA 1.2 × 20 × 109 cycles
CPU timeB = =
Clock rateB Clock rateB
1.2 × 20 × 109 cycles

Clock rateB = = 4 GHz
6 seconds
Performance Instruction Count and CPI
Instruction Count and CPI Notes
Average clock cycles

CPU clock cycles = Instructions Count ×
per instruction (CPI)
CPU time = Instruction count × CPI × Clock cycle time
Instruction count × CPI

CPU time =
Clock rate
Instruction Count for a program

Determined by program, ISA and compiler
Average Cycles per instruction (CPI)
Determined by CPU hardware
If different instructions have different CPI: average CPI affected by
instruction mix
CPI Example Notes
Computer A: Cycle Time = 250ps, CPI=2.0

Computer B: Cycle Time = 500ps, CPI=1.2
Same ISA → same instruction count I
Which is faster, and by how much?
CPUTimeA = I × CPIA × CycleTimeA

= I × 2.0 × 250ps = I × 500ps
CPUTimeB = I × CPIA × CycleTimeB
= I × 1.2 × 500ps = I × 600ps
CPUTimeA I × 600ps
= = 1.2
CPUTimeA I × 500ps
A is 1.2 times as faster as B.
CPI in more detail: comparing code segments Notes
Compiler designer is trying to decide between two code sequences for a

particular computer.
CPI for each instruction class
A B C
CPI 1 2 3
For a particular high-level language statement, the compiler writer is
considering two code sequences:
Instruction counts for
each instruction class
Code sequence A B C
1 2 1 2
2 4 1 1
Which code sequence executes the most instructions? Which will be faster?
What is the CPI for each sequence?
CPI in more detail: comparing code segments Notes
Recap: Clock Cycles = CPI × Instruction Count

if different instruction classes take different number of cycles
Pn
Clock Cycles = CPIi × Instruction Counti
i=1
Sequence 1:
InstructionCount = 2 + 1 + 2 = 5 10
CPI1 = 5 =2
ClockCycles = 2 × 1 + 1 × 2 + 2 × 3 = 10
Sequence 2:
InstructionCount = 4 + 1 + 1 = 6 9
CPI2 = 6 = 1.5
ClockCycles = 4 × 1 + 1 × 2 + 1 × 3 = 9
Performance Summary Notes
The BIG Picture

Seconds Instructions Clock cycles Seconds
Time = = × ×
Program program Instructions ClockCycle
Components of performance Units of measure

CPU execution time for a program Seconds for the program
Instructions executed for the
Instruction count
program
Average number of clock c
Clock cycles per instruction (CPI)
ycles per instruction
Clock cycle time Seconds per clock cycle

Performance Summary: understanding program Notes

performance
Hardware
or software Affects what?
component
Instruction count,
Algorithm
possibly CPI
Programming Instruction count,
language CPI
Instruction count,
Compiler
CPI
Instruction set Instruction count,
architecture clock rate, CPI
Concluding Remarks Notes
Cost/performance is improving
Due to underlying technology development
Hierarchical layers of abstraction

In both hardware and software
Instruction set architecture

The hardware/software interface
Execution time: the best performance measure

Computer Abstractions and Technology: Diego Montero

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Computer Abstractions and Technology: Diego Montero

Uploaded by

Copyright:

Available Formats

Notes

1. Computer Abstractions and Technology

Semestre Septiembre2021 - Marzo2022

2 Eight Great Ideas in Computer Architecture

3 Below Your Program

4 Under the Covers

Montero, Diego (OAC-18586) Abstractions and Technology 1 / 49

Computers have led to a third revolution for civilization

Applications that were economically infeasible suddenly become

Montero, Diego (OAC-18586) Abstractions and Technology 2 / 49

Advances in this technology affects almost every aspect of our society

Hardware advances → programmers create wonderfully useful software

Today’s science fiction suggests tomorrow’s killer applications

Montero, Diego (OAC-18586) Abstractions and Technology 3 / 49

Classes of Computing Applications Notes

Personal Computer (PCs)

Supercomputers: high-end servers

Montero, Diego (OAC-18586) Abstractions and Technology 4 / 49

Introduction Classes of Computing Applications

Classes of Computing Applications Notes

integrated with the hardware

largest class of computers and span the widest range of applications

Montero, Diego (OAC-18586) Abstractions and Technology 5 / 49

Welcome to the PostPC Era Notes

Montero, Diego (OAC-18586) Abstractions and Technology 6 / 49

Classes of Computing Applications Notes

Personal mobile device (PMD)

Montero, Diego (OAC-18586) Abstractions and Technology 7 / 49

How are programs written in a high-level language, such as C or Java,

What determines the performance of a program, and how can a

What techniques can be used by hardware designers to improve

Montero, Diego (OAC-18586) Abstractions and Technology 8 / 49

The answers to these questions will enable us to:

The rest of the course content:

Montero, Diego (OAC-18586) Abstractions and Technology 9 / 49

Performance of a program Notes

Effectiveness of the algorithms used

Montero, Diego (OAC-18586) Abstractions and Technology 10 / 49

Performance of a program Notes

Hardware or software How this component affects

Montero, Diego (OAC-18586) Abstractions and Technology 11 / 49

Montero, Diego (OAC-18586) Abstractions and Technology 12 / 49

2 Eight Great Ideas in Computer Architecture

3 Below Your Program

4 Under the Covers

Montero, Diego (OAC-18586) Abstractions and Technology 13 / 49

Eight Great Ideas in Computer Architecture Notes

1 Design for Moore’s Law: integrated circuit resources double every

Montero, Diego (OAC-18586) Abstractions and Technology 14 / 49

Below Your Program

2 Eight Great Ideas in Computer Architecture

3 Below Your Program

4 Under the Covers

Montero, Diego (OAC-18586) Abstractions and Technology 15 / 49

Below Your Program Notes

Typical application → millions of lines of code and rely on

Montero, Diego (OAC-18586) Abstractions and Technology 16 / 49

Below Your Program

Below Your Program Notes

Figure: A simplified view of hardware and software as hierarchical layers.

Montero, Diego (OAC-18586) Abstractions and Technology 17 / 49

Systems software Notes