Professional Documents
Culture Documents
The Deadline-Based Scheduling of Divisible Real-Time Workloads On Multiprocessor Platforms
The Deadline-Based Scheduling of Divisible Real-Time Workloads On Multiprocessor Platforms
The Deadline-Based Scheduling of Divisible Real-Time Workloads On Multiprocessor Platforms
OF DIVISIBLE REAL-TIME
WORKLOADS ON MULTIPROCESSOR
PLATFORMS
Suriayati Chuprat
Supervisors:
Professor Dr Shaharuddin Salleh
Professor Dr Sanjoy K. Baruah
• Introduction
• Summary of Contributions
• Anomalous Observations & Theoretical
Explanations
• Optimal Algorithm --MINPROCS
• LP-Based Algorithm --MIN-ξ
• Conclusions
• Divisible workload,
J i ( Ai , i , Di )
– Arrival time, Ai ≥ 0
– Total data size of a job, i>0
– Relative deadline, Di>0, Ai+Di
• DLT allows for the simultaneous execution
of a job upon multiple processors
Real-time DLT
P1 P2 P3 … PN
Real-time DLT
J ( A, , D)
P0 11C
Cmm 22C
Cmm 3C
m3Cm nCm nCm
P1
1C
1C
p p
P2 2C
2C
pp
P3 3C
3C
p p
1
( , n) (C C C p )
1
Pn n C m n p n p
r1 r2 r3 rn
Data transmission and execution time=C
diagram
p/(Cp+Cm)
when processors have different ready times
Computing Min-N
( , n)
C pi ,
( , n) rn ri
5 i ( si ) ( (Cm C p ))
6 si 1 max(ri 1 , si Cm i )
7 alloc alloc i
8 i i 1
end while
9 if ( alloc 1 ) then success!!
10 nmin i
else cannot meet the deadline, regardless of the number of processors used
11 nmin
end if
Simulation Results
Cm=1, Cp=100, N=16, LoadSize=50, Arrival=100 Cm=1, Cp=100, N=32, LoadSize=50, Arrival=100
Our Algorithm Our Algorithm
50
– x 14
Number of Processors
Number of Processors
Cm=1, Cp=100, N=16, Deadline=2500, Arrival=100 Cm=1, Cp=100, N=32, Deadline=2500, Arrival=100
20
Number of Processors
Lin et al. Algorithm Lin et al. Algorithm
40
15
30
10
20
5
10
0 0
50 53 59 68 80 95 113 134 158 185 50 53 59 68 80 95 113 134 158 185
Load Size Load Size
Simulation Results
Cp=100, N=16, LoadSize=50, Deadline=2500, Arrival=100 Cp=100, N=32, LoadSize=50, Deadline=2500, Arrival=100
– x
Number of Processors
Number of Processors
12 Lin et al. Algorithm 12 Lin et al. Algorithm
10 10
8 8
6 6
4 4
2 2
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Cm Cm
Cm=1, N=16, LoadSize=50, Deadline=2500, Arrival=100 Cm=1, N=32, LoadSize=50, Deadline=2500, Arrival=100
Number of Processors
25 Lin et al. Algorithm 16 Lin et al. Algorithm
14
20
12
15 10
8
10 6
5 4
2
0 0
50 60 70 80 90 100 110 120 130 140 50 60 70 80 90 100 110 120 130 140
Cp Cp
Computing Completion Time
(1) 1 2 ... n 1
(2) 0 i 1 i n
(3) ri si , 1 i n
(4) si i Cm si 1 , 1 i n
(5) si i (Cm C p ) , 1 i n
____________________________________________________________________
Computing the completion time: LP formulation
Computing Completion Time
N=4, Cm=1, Cp=100 N=6, Cm=1, Cp=100 N=8, Cm=1, C=100
3500 3000 2400
LP Solution LP Solution LP Solution
Approx Solution Approx Solution 2200 Approx Solution
3000
2500
2000
2500 1800
Completion Time
Completion Time
Completion Time
2000
1600
2000
1400
1500
1500 1200
1000
1000
1000
800
Completion Time
Completion Time
1400 1700
1600
1200 1600
1400 1500
1000
1400
1200
800
1300