Download ICS 199 (Spring 1997) Lect 1 Commodity Computing

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
COSMOS Summer 2008
Embedded Systems
Rajesh K. Gupta
Computer Science and Engineering
University of California, San Diego.
©2008 R. Gupta, UCSD
Roadmap
• Topic:
– What are embedded systems?
– What has that got to do with computers and ‘chips’?
• This lecture
– Concepts covered: computing, computers, computer architecture
– Computer peripherals
• Next lecture
– Computer interfaces and networks: I2C, SPI
• Reference
– None
©2008 R. Gupta, UCSD
Keywords:
Embedded Computer
Computing Machines
Computer Application
Computer Architecture
Computer Organization
CPU, Memory, Bus
ISA
Conventions (if I can follow them
• Important concept words are usually highlighted
– Or strongly highlighted.
• “Memory” means more than it says
• Challenges and Teasers
– Teaser questions are not meant to be
answered
©2008 R. Gupta, UCSD
©2008 R. Gupta, UCSD
Embedded Systems
• “Systems” that are part of another “Application”
• We are interested in “Embedded Computer Systems”
– Why? Hold that thought for now. (Hint: wait until ‘chips’)
• Computers are machines that are designed either for
general-purpose applications or for specific applications.
• Computing refers to problem solving using Computers
– machine-centric problem solving.
• In the past, computing are confined to scientific
applications, and machines were expensive and
necessarily shared.
©2008 R. Gupta, UCSD
Examples of Computational
Problem Solving
• Can you think of examples where you need
computers to solve problems?
– Day-to-day
– In-the-interest-science-and-humanity
• Actually by far the largest application of
computing was in building computers (hw, sw)
©2008 R. Gupta, UCSD
Problem Solving
• Problem:
– Find greatest common divisor of two numbers: 6, 15.
• Solutions:
1. Mathematical:
gcd(a, b) = gcd(a-b, b)
assume w.l.o.g. a >=b
gcd(a, 0) = a
2. Procedural (Sequential): Chinese Remainder Theorem.
©2008 R. Gupta, UCSD
if (x != 0 and y!= 0) {
repeat {
while (x>=y)
x = x - y;
swap x, y;
} until (y=0);
write gcd = x;
Traditional Problem Solving
using Computers
• Describe the method in a programming language
• Compile this program into a machine language
• Execute machine program
– supply input data
©2008 R. Gupta, UCSD
The Computing Experience
• The Computer Center
– Mainframe computing
• The Personal Computer
– Desktop, laptop, palm top computing
• The Ubiquitous Embedded Computer
– Mobile, purpose-built.
©2008 R. Gupta, UCSD
Computing Machines:
Touchpoints in History
• Generation 0: Mechanical machines
– Difference engine by Babbage (1834)
• Generation 1: Vacuum Tubes (1945-1955)
– ENIAC by Eckert and Mauchly
(1943)
• Generation 2: Transistors (1955-1965)
– PDP-1 by DEC (1960): first microcomputer
• Generation 3: Integrated Circuits (1965-1980)
– IBM 1401, 7094 and System/360 (1964)
• Generation 4: PCs and VLSI (1980-continuing)
©2008 R. Gupta, UCSD
The Computer
• Computers manipulate information by means of
programmed instructions
• A computer hardware that makes this possible is called a
stored program computer.
• A computer system designer must
– evaluate applications for which the computer is to be used
– build a system architecture from software to its hardware
organization
– (by architecture we often mean how the system appears to the
system programmer, though it is not strictly correct..)
• The lexicon: architecture, organization, usage.
©2008 R. Gupta, UCSD
Computer Architecture
• Architecture provides
– A model to organize, implement system parts
– A model to use the system
• Architectures can be diverse…
©2008 R. Gupta, UCSD
Architectural Design Paradigms
“Organically grown”
The Planned Community
The sacrificial altar
Specialized
Ambitious (and never finished) Engineered
Hardware Organization
• Computer Organization refers to the structure
of a computer:
CPU
Controller
Address
control
enables
selects
R/W
DataPath
Data
data
conditions
©2008 R. Gupta, UCSD
Memory
Software ‘Architecture’
Application
Compiler
Assembler
Micro-operations
Hardware
©2008 R. Gupta, UCSD
A Personal Computer (PC)
• Uses commodity components and standard
interfaces to build the machine
• Performance/cost variations due to choice of
components and (sometimes) their
interconnection.
©2008 R. Gupta, UCSD
PC Components
• Case:
– desktop, minitower, mid case, mid tower, full tower, large tower
– include a power supply (typically 200 watts)
• CPU and CPU fan
– sold as a chip
• Motherboard
– single-board computer
– contains place holders for CPU and Memories
– Memory modules
• Interfaces
– Video (card and monitor)
– Others: (I/O card) “Floppy, CD-ROM, Keyboard, Hard drive, Sound
card, Speakers, Modem, Tape/zip drive, Mouse”
©2008 R. Gupta, UCSD
168-pin SDRAM
The Stuff at Fry’s
184-pin DDR-SDRAM
©2008 R. Gupta, UCSD
Personal Computing
• “Low cost and general purpose”
INST
CPU
DATA
BUS
CPU
MEM
BUS
©2008 R. Gupta, UCSD
Computer Organization
3
1
CPU
MEM
2
4
©2008 R. Gupta, UCSD
BUS
Drives
Devices
Input devices
Devices
Devices
Output devices
Networking Interface
Memory
2-3 GHz
CPU
MEM
BUS
Devices
Devices
Devices
©2008 R. Gupta, UCSD
50 ns SIMM
PC2-5300 667 MHz DIMM
Lexicon: Non-intuitive & Not-obvious
• Random Access: RAMs
– static
– dynamic
– quasi-static
CPU
• Read Only: ROMs
– PROM
– EPROM
Cache
– EEPROM
– Flash EPROM
– “CMOS”
• “Cache”?
Memory
©2008 R. Gupta, UCSD
What are L1, L2, L3?
Buses
• Functions:
– transmit information
– coordinate activities
– interface standard
• Bandwidth, width/xfr * xfr/sec
• Memory addressing, addr/word, addr/block
• Latency tolerance: low, medium, high
• Length: microns, mm, inches, feet, km, 1000s
km
©2008 R. Gupta, UCSD
Buses
• CPU -- L1 Cache
– very high bandwidth, 50-500
MB/sec, addr/word
• L1 Cache -- L2 Cache
• Cache -- Main Memory
• Network -- Main Memory
– similar to disk-memory
• Memory/CPU -- Display
– high bandwidth, 1-20 MB/sec
• Computer -- Computer
– very high bandwidth, 1-200 MB/sec– medium bandwidth, kbps to Mbps,
addr/message.
• Main Memory -- Disk
– high bandwidth, 2 MB/sec,
addr/block
©2008 R. Gupta, UCSD
CPU – Memory Interaction
Clock
200 ns
CPU
Address
control
enables
selects
Controller
clock
R/W
DataPath
Memory
Data
data
conditions
• CPU controlled by a clock pulse.
• Memory is controlled by R/W control
signals.
• CPU must synchronize its memory
read, write operations with respect to
its internal clock.
• Example: 5 MHz clock, 500 ns
access time.
©2008 R. Gupta, UCSD
addr
select
VALID
500 ns
R/W’
data
VALID
WRITE CYCLE
CPU Operation
• A computing machine delivers its functionality
using instructions.
– instruction set architecture (ISA)
• A CPU executes each instruction in steps:
– Fetch instruction
– Decode instruction
– Execute instruction
• Fetch instruction
– location pointed by the Program Counter
©2008 R. Gupta, UCSD
CPU Simplified
Register
File
A
B
ALU
Accumulator
©2008 R. Gupta, UCSD
The Assembly Language
• Machine defined by an ISA
• A program is translated into an object program
• An assembly language consists of
– statement corresponding to machine instructions
– represented mnemonically: easier to remember than
numbers and bits
– typical mnemonics: ADD, SUB, MUL
– machine dependent
©2008 R. Gupta, UCSD
control
enables
selects
Data Path
Program
Controller
data
Level 3
data
conditions
A
CPU
prog.
MEM
B
Level 2
SR
SA
Lc
mux
ALU
Level 1
Register
File
C
A
Level 0
B
ALU
Accumulator
Hardware View
©2008 R. Gupta, UCSD
Software Vie
Let us see how a programmer
would build a CPU...
PC
Mem
Step 1: PC points to an
instruction in the memory.
©2008 R. Gupta, UCSD
Let us see how a programmer
would build a CPU...
PC
PC
Mem
Step 1: PC points to an
instruction in the memory.
©2008 R. Gupta, UCSD
Mem
IR
Step 2: Memory responds with “data” that
represent an instruction to be stored in IR.
Let us see how a
programmer would build a
PC
CPU...
PC
Mem
Step 1: PC points to an
instruction in the memory.
PC
Mem
IR
control
Step 3: an instruction is decoded into
“control” lines by the control block.
©2008 R. Gupta, UCSD
Mem
IR
Step 2: Memory responds with “data” that
represent an instruction to be storedin IR.
Let us see how a
programmer would build a
PC
CPU...
PC
Mem
Mem
Step 1: PC points to an
instruction in the memory.
IR
Step 2: Memory responds with “data”
that represent an instruction to be
storedin IR.
PC
PC
Mem
Mem
IR
control
Step 3: an instruction is decoded into
“control” lines by the control block.
©2008 R. Gupta, UCSD
ALU
IR
Acc
control
Step 4: Memory also provides “data”
that is used by the ALU to compute
variable values.
CPU Building (contd.)
load new address into PC,
jump operation.
PC
addr
data out
ALU
IR
Mem
data in
control
Acc
save data into memory.
Step 5: ALU output is stored back into memory; the ALU may
also compute address, in which case ACC output it input to PC.
ALU also computes data which is stored in memory.
©2008 R. Gupta, UCSD
The Changing Computer
Instrumented
wide-area spaces
Internet end-points
In-body, in-cell, in-vitro spaces
Personal area spaces
©2008 R. Gupta, UCSD
Driven By “Chips” Underneath
©2008 R. Gupta, UCSD
Cellphone Baseband
Graphics Controller
Embedded Systems: Sensor Node
©2008 R. Gupta, UCSD
©2008 R. Gupta, UCSD
Recap
• A computer is a ‘Computing Machine’
• We saw what a computer looks like
– Beyond what you see at Fry’s
– Architectural components & interfaces
• CPU, ALU, Memory, Bus
• Application = What can we do with computers
– Traditional numeric problem solving replaced by ‘special-purpose’
functions
• Sensing, control, media,…
• Next: what are peripherals and how we connect to these?
©2008 R. Gupta, UCSD