Minimizing System Modification in an
Incremental Design Approach
Paul Pop, Petru Eles, Traian Pop, Zebo Peng
Department of Computer and Information Science, Linköpings universitet, Sweden
Summary
I/O Interface
RAM
Future
applications
Do not exist yet
at Version N!
Current
applications
Map and schedule
so that the
future applications
will have a chance
to fit.
N
Version N+1
Incremental Design
N-1
Existing
applications
Modify (re-map)
so that the
current applications
will fit.
n Mapping and scheduling of distributed
embedded systems for hard-real time applications
ROM
CPU
ASIC
Comm. Controller
n Static cyclic scheduling of processes and messages,
n Bus access scheme: time-division multiple-access.
Node
n Incremental design process
n Already existing system,
n Implement new functionality,
n a) Existing system modified as little as possible,
b) new functionality can be easily added to the system.
S0
S1
S2
S0
S3
S1
S2
S3
Slot
n Mapping strategy
TDMA Round
Cycle of two rounds
n a) Subset selection to minimize modification time,
n b) Two design criteria, objective function.
n Start from an already existing system with applications
n Implement new functionality on this system
Mapping and Scheduling
Problem Formulation
n To reduce design and testing time:
As few as possible modifications of the existing applications
n Input
n After the new functionality has been implemented:
It should be easy to add functionality in the future
Characterizing existing applications:
150
n A set of existing applications.
n A current application to be mapped.
n The system architecture.
n Output
n Requirements
a)
b)
c)
50
Γ3
70
Γ4
50
Γ5
Γ6
Γ9
Γ10
20
Γ7
Characterizing future applications:
§ Typical WCETs
§ Typical message sizes
§ Smallest expected period Tmin
§ Expected necessary processor time tneed
§ Expected necessary bandwidth bneed
n a) constraints of the current application are satisfied
and minimal modifications are performed to the
existing applications.
P1
n Initial mapping and scheduling
n Requirement a)
Subset selection problem
Select that subset Ω of existing applications so that
the current application fits and the modification
cost R(Ω) is minimized:
R (Ω ) =
∑R
Γi ∈Ω
i
Three approaches to the subset selection problem
n Exhaustive Search (ES)
n Ad-Hoc Solution (AH)
n Second criterion: C2P, C2m
How well the slack is distributed in time
to
accommodate Tmin, tneed and bneed
Paul Pop, Petru Eles, Traian Pop, Zebo
Peng:
An approach to Incremental Design of
Distributed Embedded Systems,
Design Automation Conference, 2001
n Subset Selection Heuristic (SH)
n Requirement b)
Objective function minimization:
C = w1P (C1P ) + w1m (C1m ) + w2P max(0, tneed − C 2P ) + w2m max(0, bneed − C2m )
Average Modification Cost R(Ω)
360 ms
Periodic slack
min(40, 80, 0) = 0ms
n
min(40, 0, 80) = 0ms
min(80, 80, 40) = 40ms
S1 S2 S3 S1 S2 S3 S1 S2 S3 S1 S2 S3
C P2 = 40ms
a)
Round 0 Round 1 Round 2 Round 3
Period 0
Period 1
Period 2
min(40, 40, 40) = 40ms
min(40, 40, 40) = 40ms
min(80, 80, 40) = 40ms
S1 S2 S3 S1 S2 S3 S1 S2 S3 S1 S2 S3
CP
2 = 120ms
b)
Round 0 Round 1 Round 2 Round 3
T mi
Mapping Strategy
1200
AH
SH
ES
1000
800
600
400
200
0
320
400
480
560
640
Number of processes
140
Average Execution Time [min]
Slac
k
AH
SH
ES
120
100
80
60
40
20
0
320
Percentage of future applications [%]
Curren
tP2
n First criterion: C1P, C1m
How well the resulted slack sizes
accommodate a future
application
N1
N2
N3
Bus
Γ2
n b) new future applications can be mapped
on the resulted system.
Existin
g
Future
P1
:
N1
N2
N3
Bus
70
Γ8
n A mapping and scheduling of the current application,
so that the incremental design requirements are satisfied.
Design Criteria
Γ1
400
480
560
640
Number of processes
120
MS*
100
AD
80
60
40
20
0
40
1 of240 16
April 25, 2006
80
160
Number of processes in the current application