#### DMCA

## Buffer Minimization in Earliest-Deadline First Scheduling of Dataflow Graphs

Citations: | 3 - 2 self |

### Citations

598 | Static scheduling of synchronous dataflow programs for digital signal processing.
- Lee, Messerschmitt
- 1987
(Show Context)
Citation Context ...els. In this paper, we target the design of software applications where software tasks interact with each other via software buffers. Among dataflow models of computation, synchronous data-flow (SDF) =-=[15]-=- and cyclo-static dataflow (CSDF) [6] are popular in the embedded system community. An actor has constant production and consumption rates in SDF and periodic rates in CSDF. Figure 1(a) shows a cyclic... |

238 | Cyclo-Static Dataflow.
- Bilsen
- 1996
(Show Context)
Citation Context ...gn of software applications where software tasks interact with each other via software buffers. Among dataflow models of computation, synchronous data-flow (SDF) [15] and cyclo-static dataflow (CSDF) =-=[6]-=- are popular in the embedded system community. An actor has constant production and consumption rates in SDF and periodic rates in CSDF. Figure 1(a) shows a cyclic dataflow graph with ultimately perio... |

110 |
Real time scheduling theory: A historical perspective
- Sha, Abdelzaher, et al.
- 2004
(Show Context)
Citation Context ...heduling theory for both uniprocessor and multiprocessor systems has already provided a lot of algorithms to check whether a task system meets its timing requirements, even in the worst-case scenario =-=[11,18]-=-. However, most algorithms assume that timing parameters of tasks are known a priori. Our approach to the problem is rather to synthesize the timing parameters of actors in dataflow graphs so as to en... |

99 | A Survey of Hard Real-Time Scheduling for Multiprocessor Systems.”
- Davis, Burns
- 2011
(Show Context)
Citation Context ...heduling theory for both uniprocessor and multiprocessor systems has already provided a lot of algorithms to check whether a task system meets its timing requirements, even in the worst-case scenario =-=[11,18]-=-. However, most algorithms assume that timing parameters of tasks are known a priori. Our approach to the problem is rather to synthesize the timing parameters of actors in dataflow graphs so as to en... |

56 | Pt-scotch: A tool for efficient parallel graph ordering.
- Chevalier, Pellegrini
- 2008
(Show Context)
Citation Context ...e the implementation code of actors so that they produce tokens only at the end of their firings. However, firstly, a job may complete before its worst-case execution time. Secondly, if jobs p1[8], p1=-=[9]-=-, and p1[10] consume many tokens before the producer writes something to channel e3, this is likely to increase the minimum number of initial tokens on the channel and so the buffering requirements. O... |

56 |
SDF3: SDF For Free,” in
- Stuijk, Geilen, et al.
- 2006
(Show Context)
Citation Context ...paper. Figure 6 shows the ratios of the buffering requirements to the minimum (regardless of the throughput) achievable buffering requirements. The minimum buffer sizes are obtained by the SDF 3 tool =-=[21]-=- which computes static-periodic schedules of (C)SDF graphs (auto-concurrency disabled). Figure 7 shows the processor utilization factor obtained by each tool. RaFos to the minimum sizes 3 2,5 2 1,5 1 ... |

55 | Efficient Computation of Buffer Capacities for Cyclo-Static Dataflow Graphs.
- Wiggers
- 2007
(Show Context)
Citation Context ..., Section 9 ends the paper with conclusions.2. Related work Schedules of (C)SDF graphs that minimize buffer sizes under throughput constraints have been investigated in many studies, for instance in =-=[23]-=-. Those works aim at creating static cyclic schedules of actors for which hard real-time scheduling theories are not applicable. It is well known that the buffer sizes influence the maximum throughput... |

50 | Exploring Trade-Offs in Buffer Requirements and Throughput Constraints for Synchronous Dataflow Graphs.
- Stuijk
- 2006
(Show Context)
Citation Context ...at the buffer sizes influence the maximum throughput that can be achieved. Recently, a technique that computes the trade-offs between the throughput and buffer sizes for (C)SDF graphs was proposed in =-=[20,22]-=-. Our approach rather aims at creating EDF schedules that achieve a high throughput while trying to reduce the total amount of buffer storage capacities. It also considers a more general model than CS... |

45 |
Improvement in feasibility testing for real-time tasks.
- Ripoll, Crespo, et al.
- 1996
(Show Context)
Citation Context ...e. 6.1 Standard schedulability analysis The exact schedulability analysis of constrained periodic task systems is based on the processor-demand approach as given by the following lemma. Lemma 1 (from =-=[17]-=-). A synchronous periodic task set is schedulable if and only if U < 1 and ∀l ≤ L ∗ : h(l) ≤ l where L ∗ = ∑ (πi−di)Ui pi∈P 1−U and Ui = C i π i . L ∗ is called the feasibility bound. As proposed in [... |

36 | Throughput-buffering trade-off exploration for cyclo-static and synchronous dataflow graphs.
- Stuijk, Geilen, et al.
- 2008
(Show Context)
Citation Context ...at the buffer sizes influence the maximum throughput that can be achieved. Recently, a technique that computes the trade-offs between the throughput and buffer sizes for (C)SDF graphs was proposed in =-=[20,22]-=-. Our approach rather aims at creating EDF schedules that achieve a high throughput while trying to reduce the total amount of buffer storage capacities. It also considers a more general model than CS... |

15 |
Implementation of a speedup-optimal global EDF schedulability test.
- Baruah, Bonifaci, et al.
(Show Context)
Citation Context ... a global EDF symbolic schedulability analysis of dataflow graphs which is similar to the SQPA technique but based on (an improved version of) the speedup-optimal schedulability analysis presented in =-=[3, 5]-=-. • Once we have a symbolic schedulability analysis for constrained task systems, the deadlines adjustment technique, as illustrated in the previous example, can be used to reduce the buffering requir... |

14 |
2011. Multi-task implementation of multi-periodic synchronous programs. Discrete Event Dynamic Systems
- Pagetti, Forget, et al.
- 1998
(Show Context)
Citation Context ...c. EDF scheduling of data-dependent tasks was tackled in [12] by adjusting deadlines and release times and without using lockbased synchronizations. That approach, implemented in the Prelude compiler =-=[16]-=-, differs from the present one in three respects. 1) In Prelude, deadlines are adjusted so as to ensure data-dependencies w.r.t. multi-rate communication patterns; while precedences in the present mod... |

13 |
Tests for global EDF schedulability analysis
- Bertogna, Baruah
- 2011
(Show Context)
Citation Context ...e precedence between jobs p3[4] and p1[9] without using lock-based mechanisms. In a dual way, actor p3 is the consumer of channel e2; while actor p2 is its producer. As depicted in Figure 1(b), job p2=-=[5]-=- also preempts job p3[4]. If the deadline of job p3[4] is less or equal to that of p2[5], then this latter will not preempt job p3[4]. This technique of deadlines adjustment can be used to play on pri... |

12 | Symbolic computation of schedulability regions using parametric timed automata.
- Cimatti, Palopoli, et al.
- 2008
(Show Context)
Citation Context ..., we will investigate the impact of preemption on the computation of buffer sizes. Actor p3 is the producer of channel e3, while actor p1 is its consumer. As depicted in Figure 1(b), jobs p1[9] and p1=-=[10]-=- preempt job p3[4]. Since the technique proposed in [8] does not assume any knowledge about the implementation code of actors (i.e. it is unknown when exactly the actor reads and writes tokens), it co... |

9 | Hard-real-time scheduling of data-dependent tasks in embedded streaming applications
- Bamakhrama, Stefanov
(Show Context)
Citation Context .... Recently, techniques that schedule dataflow graphs as implicit-deadline periodic (IDP) task systems have been proposed with an aim of achieving a high processor utilization factor (i.e. throughput) =-=[1,8]-=-. An IDP task has a deadline equal to its period. Such scheduling approach gives the maximum achievable processor utilization for large set of dataflow(a) (b) Figure 1. (a) A cyclic dataflow graph wi... |

8 | Affine data-flow graphs for the synthesis of hard real-time applications
- Bouakaz, Talpin, et al.
- 2012
(Show Context)
Citation Context ...or a given sequence of inputs, the system will always produce the same sequence of outputs. This implies that neither overflow nor underflow exceptions over communication buffers may occur at runtime =-=[8]-=-. Unlike with the classical static-periodic scheduling of dataflow graphs, this paper considers a fixed-job-priority preemptive scheduling policy: the EDF policy. Real-time scheduling theory for both ... |

8 | Scheduling dependent periodic tasks without synchronization mechanisms
- Forget, Boniol, et al.
- 2010
(Show Context)
Citation Context ...of buffer storage capacities. It also considers a more general model than CSDF where production and consumption rates can be ultimately periodic. EDF scheduling of data-dependent tasks was tackled in =-=[12]-=- by adjusting deadlines and release times and without using lockbased synchronizations. That approach, implemented in the Prelude compiler [16], differs from the present one in three respects. 1) In P... |

8 | Validation of Mixed Signal-Alpha Real-Time Systems through Affine Calculus on
- SMARANDACHE, GAUTIER, et al.
- 1999
(Show Context)
Citation Context ...fine relations, defined below, between activation clocks are expressive enough to describe strictly periodic schedules of dataflow graphs. Affine transformations of abstract clocks were introduced in =-=[19]-=- and they enjoy a canonical form and other useful mathematical properties. Definition 1 (Affine relation). A (n, ϕ, d)-affine relation between two clocks ˆpi and ˆpk has three parameters n, d ∈ N ∗ an... |

7 |
The partitioned dynamic-priority scheduling of sporadic task systems. Real-Time Systems:
- Baruah, Fisher
- 2007
(Show Context)
Citation Context ...ocessor schedulability. For the multiprocessor case, we will consider the partitioned EDF policy on identical multiprocessors where each task is allocated to a processor and no migration is permitted =-=[4]-=-. The current state of the art favors the partitioned scheduling over the global one for reasons that concern the run-time cost of inter-processor migration and the optimality of schedulability tests.... |

7 | Robustness analysis for scheduling problems using the inverse method
- Fribourg, Lesens, et al.
- 2012
(Show Context)
Citation Context ...ic timed automata to symbolically compute the schedulability region (i.e. the region of the parameter space that corresponds to feasible designs) of a task system scheduled using fixed priorities. In =-=[13]-=-, the Inverse method for parametric timed automata is used to synthesize zones of the timing parameter space where the system is schedulable. Our solution is tailored to a specific scheduling policy a... |

5 | Managing latency in embedded streaming applications under hard-real-time scheduling
- Bamakhrama, Stefanov
- 2012
(Show Context)
Citation Context ...EDF symbolic schedulability of IDP task systems is quite simple with a polynomial time complexity [8]. However, the IDP model may increase the latency and the buffering requirements of the graphs. In =-=[2]-=-, authors have shown that the IDP model increases the latency for a class of graphs called unbalanced graphs. In this paper, we will investigate the impact of the IDP model on the buffering requiremen... |

5 |
Design of Streaming Applications on MPSoCs using Abstract Clocks
- GAMATIÉ
(Show Context)
Citation Context ...scheduling policy and to specific constraints; it has hence less complexity and implementation overhead. Regarding affine relations, a subclass of affine relations between abstract clocks was used in =-=[14]-=- to address time requirements of streaming applications on multiprocessor systems on chip. The whole class of affine relations is used in this paper. The affine model is also used in [7] to compute fi... |

4 | Design of Safety-Critical Java Level 1 Applications Using Affine Abstract Clocks,”
- Bouakaz, Talpin
- 2013
(Show Context)
Citation Context ... was used in [14] to address time requirements of streaming applications on multiprocessor systems on chip. The whole class of affine relations is used in this paper. The affine model is also used in =-=[7]-=- to compute fixed-priority schedules of dataflow graphs with constrained deadlines. 3. Background This section outlines the generic approach for strictly (as opposite to static) periodic scheduling of... |

4 | Improvement to Quick Processor-Demand Analysis for EDF-Scheduled Real-Time Systems
- Zhang, Burns
(Show Context)
Citation Context ...al EDF symbolic schedulability analysis of dataflow graphs with constrained deadlines called “Symbolic Quick-convergence Processor-demand Analysis” (SQPA). It is based on the well-known QPA algorithm =-=[24,25]-=-. Instead of testing schedulability for all possible timing parameters, our algorithm does that incrementally and so maximizes the processor utilization factor. For partitioned EDF scheduling of dataf... |