Escolar Documentos
Profissional Documentos
Cultura Documentos
Remaining
Now let Ta be a task with the smallest remaining Deadline Laxity
Execution Time
execution time among a set of tasks with the least laxities.
T1 3 6 3
This means Ta ’s deadline is the earliest among those
tasks. And let Tmin be a task which has the earliest T2 4 7 3
deadline among all tasks, but has larger laxity than Ta .
With no loss in schedulability (i.e. in order not to miss
the deadlines of other tasks), Ta may be executed with
no preemption for D min (t ) − L a (t ) , where Dmin (t ) is the T1
deadline of task Tmin , and La (t ) is the laxity of task
Ta . If D a (t ) ≤ D min (t ) is satisfied, Ta will be T2
completed with no preemption.
t
Therefore, MLLF scheduling algorithm allows laxity 0 1 2 3 4 5 6 7
Algorithm MLLF
begin
finds Ta that satisfies V1 = {Ti | Li (t ) ≤ L j (t ), Ti , T j ∈ T } and Ta = {Ti | Ei (t ) ≤ E j (t ), Ti , T j ∈ V1} ;
finds Tmin that satisfies Tmin = {Ti | Di (t ) ≤ D j (t ) and Li (t ), L j (t ) > La (t ), Ti , T j ∈ T } ;
executes Ta until satisfying any rescheduling point conditions ;
end
0.3
0.1
1.0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Processor Utilization
0.5
1.0
Figure 7. Global performance ratio with task =
0.9
0.8
10 and 20
0.7
4.0
N MLLF 0.6
N LLF
0.5
3.5
0.4
global performance ratio
0.3 3.0
10 tasks
0.2
20 tasks
0.1 2.5
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 2.0
Processor Utilization
1.5
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 5. Comparison of the number of context The number of periodic tasks
(processor utilization = 0.9)
switches with task = 10 and 20
Figure 8. Global performance ratio
In Figure 6, N MLLF N LLF is shown when the
processor utilization is 0.9. Although as the number of In Figure 8, the global performance ratio is shown
tasks increases the ratio also increases slightly, the when the processor utilization is 0.9. As the number of
number of context switches with MLLF is half on the tasks increases, the performance of MLLF goes down
average. due to the cost of the algorithm itself. Therefore, in order
1.0 to maximize the performance of MLLF algorithm, an
0.9
efficient implementation approach and data structure are
0.8
needed.
0.7
0.6
N MLLF
N LLF 0.5
4. Conclusion
0.4
References
[1] M.L. Dertouzos, “Control Robotics: the Procedural
Control of Physical Processes,” Information Processing 74,
North-Holland Publishing Company, 1974
[2] A.K. Mok, “Fundamental Design Problems of Distributed
Systems for the Hard-Real-Time Environment,” Ph.D.
Thesis, Department of Electrical Engineering and
Computer Science, Massachusetts Institute of Technology,
Cambridge, Massachusetts, May 1983
[3] Michael B. Jones, Joseph S. Barrera III, Alessandro Forin,
Paul J. Leach, Daniela Rosu and Marcel-Catalin Rosu,
“An Overview of the Rialto Real-Time Architecture,” In
Proceedings of the Seventh ACM SIGOPS European
Workshop, Connemara, Ireland, pages 249-256, September,
1996
[4] G.C. Buttazzo, “Hard Real-Time Computing Systems :
Predictable Scheduling Algorithms and Applications,”
Kluwer Academic Publishers, 1997