Meeting Deadlines Cheaply

Legriel, Julien; Maler, Oded

doi:10.1109/ecrts.2011.25

Cited by 4 publications

(8 citation statements)

References 43 publications

(21 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent advances in SAT and SMT solvers and other constraint propagation techniques suggests an opportunity to formulate and solve the problem in a monolithic way, avoiding the sub-optimality of decomposed solutions. For example, [15] exploit SMT solvers to combine multiple deployment sub-problems: the task-to-processor assignment, the ordering of tasks on each processor and the assignment of scalable voltage per processor. For SDF graphs, [2] and [28] combine multiple phases using a constraint programming (CP) engine.…”

Section: Discussionmentioning

confidence: 99%

“…Expressing scheduling problems using constraints is fairly standard [1,2,28,15] and various formulations may differ in the assumptions they make about the application and the architecture and the aspects of the problem they choose to capture. We target shared-memory multi-core architectures such as [16,26,11].…”

Section: Constraint-based Feasible Cost-space Explorationmentioning

confidence: 99%

“…First we will employ more refined models of data communication where different mappings imply different data transfer costs. Secondly we will consider pipelined executions as was done in [15,2,28,27], using e.g., a finite unfolding. This will increase the number of tasks but will reduce the effect of precedence constraints.…”

Section: /15mentioning

confidence: 99%

“…We take advantage of recent progress in SMT (SAT modulo theory) solvers [17,7] to provide a good approximation of the Pareto front of the problem. We encode the precedence and resource constraints of the problem in the theory of linear arithmetic and, following [15,14], we submit queries to the solver concerning the existence of solutions whose costs reside in various parts of the multi-dimensional cost space. Based on the answers to these queries we obtain a good approximation of the optimal trade-off between these criteria.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Symmetry Breaking for Multi-criteria Mapping and Scheduling on Multicores

Tendulkar

Poplavko

Maler

2013

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Multiprocessor mapping and scheduling is a long-old difficult problem. In this work we propose a new methodology to perform mapping and scheduling along with buffer memory optimization using an SMT solver. We target split-join graphs, a formalism inspired by synchronous data-flow (SDF) which provides a compact symbolic representation of dataparallelism. Unlike the traditional design flow for SDF which involves splitting of a big problem into smaller heuristic sub-problems, we deal with this problem as a whole and try to compute exact Pareto-optimal solutions for it. We introduce symmetry breaking constraints in order to reduce the run-times of the solver. We have tested our work on a number of SDF graphs and demonstrated the practicality of our method. We validate our models by running an image decoding application on the Tilera multi-core platform.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Constraint-based Feasible Cost-space Explorationmentioning

confidence: 99%

Section: /15mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Symmetry Breaking for Multi-criteria Mapping and Scheduling on Multicores

Tendulkar

Poplavko

Maler

2013

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…In [6] we have used the SMT solver Yices [2] to solve the single-criterium problem of finding the cheapest (in terms of energy) configuration on which such a task graph can be scheduled while meeting a given deadline. We applied our algorithm to solve a multi-criteria version of the problem, namely to show trade offs between energy cost and execution time.…”

Section: Experimentationmentioning

confidence: 99%

Approximating the Pareto Front of Multi-criteria Optimization Problems

Legriel

Guernic

Cotton

et al. 2010

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. We propose a general methodology for approximating the Pareto front of multi-criteria optimization problems. Our search-based methodology consists of submitting queries to a constraint solver. Hence, in addition to a set of solutions, we can guarantee bounds on the distance to the actual Pareto front and use this distance to guide the search. Our implementation, which computes and updates the distance efficiently, has been tested on numerous examples.

show abstract

Performance-aware scheduler synthesis for control systems

Majumdar

Saha

2011

Proceedings of the Ninth ACM International Conference on Embedded Software

View full text Add to dashboard Cite

We consider the problem of designing a cyber-physical system where several control loops share the same architectural resources. Typically, the design of such systems proceeds in two steps. In the platform independent step, for each control loop in the system, the control designer calculates a control law and a sampling time that together ensure that the control loop has certain desired performance. Then, in the platform dependent step, these control tasks are scheduled on the platform, and a schedulability analysis determines if (and how) the control laws can be implemented and scheduled without missing the sampling deadlines.In this paper, we explore an alternative controllerscheduler co-design approach that aims to achieve optimal performance for all the individual control loops maintaining fairness. We first analyze the control systems to find out the rates at which control signals should be dropped to maintain schedulability and the optimal performance. We then use the rates to compute a schedule statically. We show a control theoretic approach to compute the effect of the rate of drops on the performance of the control systems and provide an SMT-based scheduling algorithm that takes as input control tasks, their periods, worst-case execution times, and their rate of drops, and outputs a static schedule that optimizes the performance of the control systems. We demonstrate our results through a case study on a family of inverted pendulums sharing the same processor for their control computations.

show abstract

Meeting Deadlines Cheaply

Cited by 4 publications

References 43 publications

Symmetry Breaking for Multi-criteria Mapping and Scheduling on Multicores

Symmetry Breaking for Multi-criteria Mapping and Scheduling on Multicores

Approximating the Pareto Front of Multi-criteria Optimization Problems

Performance-aware scheduler synthesis for control systems

Contact Info

Product

Resources

About