Documente Academic
Documente Profesional
Documente Cultură
1
An embedded system example -- a digital
Design challenge – optimizing design metrics
camera
Digital camera chip • Common metrics (continued)
CCD
A2D
CCD preprocessor Pixel coprocessor D2A – Time-to-prototype: the time needed to build a working version of the
system
lens
– Time-to-market: the time required to develop a system to the point that it
JPEG codec Microcontroller Multiplier/Accum
can be released and sold to customers
DMA controller Display ctrl – Maintainability: the ability to modify the system after its initial release
– Correctness, safety, many more
Memory controller ISA bus interface UART LCD ctrl
2
Losses due to delayed market entry NRE and unit cost metrics
to ta l c ost (x1000)
p e r p rod uc t c ost
Delayed – Triangle area equals revenue $120,000 $120
Losses due to delayed market entry (cont.) The performance design metric
delayed entry
• Percentage revenue loss = • Latency (response time)
On-time
– Time between task start and end
Market rise Market fall (D(3W-D)/2W2)*100% – e.g., Camera’s A and B process images in 0.25 seconds
Delayed • Try some examples • Throughput
– Lifetime 2W=52 wks, delay D=4 wks – Tasks per second, e.g. Camera A processes 4 images per second
D W 2W
– (4*(3*26 –4)/2*26^2) = 22% – Throughput can be more than latency seems to imply due to concurrency, e.g.
On-time Delayed Time – Lifetime 2W=52 wks, delay D=10 wks Camera B may process 8 images per second (by capturing a new image while
entry entry – (10*(3*26 –10)/2*26^2) = 50% previous image is being stored).
– Delays are costly! • Speedup of B over S = B’s performance / A’s performance
– Throughput speedup = 8/4 = 2
NRE and unit cost metrics Three key embedded system technologies
• Costs: • Technology
– Unit cost: the monetary cost of manufacturing each copy of the system,
excluding NRE cost – A manner of accomplishing a task, especially using technical
– NRE cost (Non-Recurring Engineering cost): The one-time monetary cost of processes, methods, or knowledge
designing the system
– total cost = NRE cost + unit cost * # of units • Three key technologies for embedded systems
– per-product cost = total cost / # of units – Processor technology
= (NRE cost / # of units) + unit cost
• Example – IC technology
– NRE=$2000, unit=$100 – Design technology
– For 10 units
– total cost = $2000 + 10*$100 = $3000
– per-product cost = $2000/10 + $100 = $300
Amortizing NRE cost over the units results in an
additional $200 per unit
3
Processor technology Single-purpose processors
• The architecture of the computation engine used to implement a • Digital circuit designed to execute exactly Controller Datapath
system’s desired functionality one program Control index
• Processor does not have to be programmable – a.k.a. coprocessor, accelerator or peripheral logic
total
– “Processor” not equal to general-purpose processor
Controller Datapath
• Features State
register +
Controller Datapath Controller Datapath
• Processors vary in their customization for the problem at hand • Programmable processor optimized for a Controller Datapath
total = 0
• Benefits for i =1 to …
General-purpose Application-specific Single-purpose
processor processor processor – Some flexibility, good performance, size and
power
Embedded Systems Design: A Unified 20 Embedded Systems Design: A Unified 23
Hardware/Software Introduction, (c) 2000 Vahid/Givargis Hardware/Software Introduction, (c) 2000 Vahid/Givargis
4
IC technology PLD (Programmable Logic Device)
• All layers are optimized for an embedded system’s • The most important trend in embedded systems
particular digital implementation – Predicted in 1965 by Intel co-founder Gordon Moore
– Placing transistors IC transistor capacity has doubled roughly every 18 months
– Sizing transistors for the past several decades
– Routing wires 10,000
1,000
5
Graphical illustration of Moore’s law The co-design ladder
1981 1984 1987 1990 1993 1996 1999 2002 Behavioral synthesis
– Hardware and software Compilers
(1990's)
(1960's,1970's)
design technologies were
10,000 150,000,000 Register transfers
transistors transistors very different Assembly instructions RT synthesis
– Recent maturation of Assemblers, linkers
(1980's, 1990's)
Leading edge Leading edge
chip in 1981 chip in 2002 synthesis enables a unified (1950's, 1960's) Logic equations / FSM's
Logic synthesis
view of hardware and (1970's, 1980's)
Machine instructions
software Logic gates
– A 2002 chip can hold about 15,000 1981 chips inside itself The choice of hardware versus software for a particular function is simply a tradeoff among various
design metrics, like performance, power, size, NRE cost, and especially flexibility; there is no
fundamental difference between what hardware or software can implement.
1,000
pace with chip capacity
Productivity
100
10,000 100,000
10
1,000 10,000
1 Logic transistors 100 1000
per chip 10 Gap 100 Productivity
0.1 IC capacity (K) Trans./Staff-Mo.
(in millions) 1 10
0.01 0.1 1
2003
2005
1983
1985
1987
1993
2001
1989
1991
1995
1997
1999
2007
2009
productivity
0.01 0.1
0.001 0.01
6
Design productivity gap
Summary