leiden embedded research center prof. dr. ed f. deprettere, dr. bart kienhuis, dr. todor stefanov...
TRANSCRIPT
Leiden Embedded Research Center
Prof. Dr. Ed F. Deprettere, Dr. Bart Kienhuis, Dr. Todor StefanovLeiden Embedded Research Center (LERC)
Leiden Institute of Advanced Computer ScienceLeiden Univerity
{edd, kienhuis, stefanov}@liacs.nlhttp://www.liacs.nl/lerc
LERC's challenges 08/02/09
Outline of Presentation
• Embedded is not merely 'small'.
A small, and a not so small example.
• Embedded means: time matters.
Not 'fast', but determinate and predictable
(Multi-threaded is neither of the three).
LERC's challenges 08/02/09
NASA's Mars Sojourner Rover (July 8, 1997)Microprocessor: 8-bit Intel 80C85(produced 1977-1990)
Rover senses the environment (input signals), decides on (computes) its actions (response) in real time.
A small Embedded System
iPhone. (Small?)
ARM11 coreVector Floating Point co-processor (3D-graphics)SIMD integer CPU (2,1 MIPS)DMA .45mW/MHz
LERC's challenges 08/02/09
A not so small Embedded System
station density high in core
central core contains supercomputer
350 km
text
Central core(2km)
Remotestationation
station contains• 100 LF antennas• 100 HF compound antennas
• LOFAR• SKA
Distributed hierarchicalradio telescopes
LERC's challenges 08/02/09
Embedded is What?An Embedded System is an information processing system that is:• application domain specific (not general purpose) • tightly coupled to its environmentExamples of application domains are: automotive, multimedia. Environment: type and properties of input/output information.Tightly coupled: environment dictates what the system’s response behavior must be. (“ES cannot synchronize with environment”)
Embedded Systems are reactive and real-time
LERC's challenges 08/02/09
Time Matters.
Computer Science has been concerned about abstraction.
An instruction set, a register file, and a program counter,is all we need to execute whatever application(s) we candream of.
True. But, no programming language (except assembly) considers timing requirements and constraints.
Modern languages and computers make it impossible to tellhow long it will take to execute a piece of code.
01EMBSYST2006 Ed F.Deprettere
8
System-Level Performance Analysis
InputStream
InputStream
I/O
P
Memory Requirements?
Processor Speeds?
Timing Properties?
Bus Utilization?
Bottleneck?
01EMBSYST2006 Ed F.Deprettere
9
Difficulties
InputStream
Complex Input:- Timing (jitter, bursts, ...)- Different Event Types
Task Communication
Task Scheduling
ab acc b
01EMBSYST2006 Ed F.Deprettere
10
DifficultiesProcessor
Task
BufferInput
Stream
Task Communication
Task Scheduling
ab acc b
Complex Input:- Timing (jitter, bursts, ...)- Different Event Types
Variable Resource AvailabilityVariable Execution Demand- Input (different event types)- Internal State (Program, Cache, ...)
01EMBSYST2006 Ed F.Deprettere
11
Why is it difficult?
It is difficult because there are too many parametersfor any optimization procedure to give an optimal matchbetween a (small?) set of (simultaneously?) runningapplications and a corresponding execution platform.
01EMBSYST2006 Ed F.Deprettere
12
Summary
What is specific to research in embedded systems and software?
The impact of the environment The performance analysis in terms of delays and buffer sizes
Where are the fundamental questions? Compositionality (modularity) composability (layering)
scalability, dependability and security Design space exploration
01EMBSYST2006 Ed F.Deprettere
13
Conclusion (2)
Move successful concepts (like abstraction and modularity) to embedded systems
Do meaningful experiments that are carefully planned (learn from physics …)