Eliminating Migration in Multi-processor Scheduling
โ Scribed by Bala Kalyanasundaram; Kirk R Pruhs
- Publisher
- Elsevier Science
- Year
- 2001
- Tongue
- English
- Weight
- 186 KB
- Volume
- 38
- Category
- Article
- ISSN
- 0196-6774
No coin nor oath required. For personal study only.
โฆ Synopsis
We investigate the power of migration in real-time multi-processor scheduling with preemption. We show that every collection of jobs that can be completed by some schedule S on m processors can also be completed by a nonmigratory schedule S on 6m -5 processors. We can conclude from this result that, for many scheduling problems, such as P r i pmtn w i 1 -U i and special cases thereof, the ability of the scheduler to migrate jobs is of only limited advantage. Our proof is constructive, and can be implemented to run in pseudo-polynomial time (or in polynomial time at the cost of raising the bound from 6m -5 to 12m -5). As an example of the usefulness of this result, we give a polynomial-time 6-approximation reduction from the problem P r i pmtn w i 1 -U i to the problem 1 r i pmtn w i 1 -U i . The proof of correctness of this reduction critically uses the fact that there is a nonmigratory schedule that closely approximates the optimal migratory schedule.
๐ SIMILAR VOLUMES
We consider an on-line list scheduling problem of multi-core processor tasks with virtualization to minimize makespan. The competitive ratio of an on-line algorithm is shown for every specific m, where m is the number of processors. Better on-line algorithms are presented for a small number of proce
This paper presents a procedure for eliminating redundant columns in generalized set-covering formulations (GSCFs) of continuous tour scheduling problems that are characterized by labor requirements of zero in some planning periods. We describe the procedure and discuss properties of certain schedul
Instruction scheduling and register allocation are two very important optimizations in modern compilers for advanced processors. These two optimizations must be performed simultaneously in order to maximize the instruction-level parallelism and to fully utilize the registers [1]. In this paper, we s
## Abstract The distributed nature of the grid results in the problem of scheduling parallel jobs produced by several independent organizations that have partial control over the system. We consider systems in which each organization owns a cluster of processors. Each organization wants its tasks t