Utilization-Tensity Bound for Real-Time DAG Tasks under Global EDF Scheduling

Jiang, Xu; Sun, Jun; Tang, Yue; Guan, Nan

doi:10.1109/tc.2019.2936477

Cited by 21 publications

(5 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previously, works on CIM architectures have only supported the inference of the DNN. Some of the works have been conducted for the in situ training process on-chip, based on the SRAM [68,69] and RRAM [70][71][72]. The energy-efficient feed forward propagation and backpropagation can be realized in these architectures.…”

Section: D Nand-based Fefet For Trainingmentioning

confidence: 99%

Ferroelectric HfO₂-based synaptic devices: recent trends and prospects

Hur

Luo

et al. 2021

Semicond. Sci. Technol.

View full text Add to dashboard Cite

Neuro-inspired deep learning algorithms have shown promising futures in artificial intelligence. Despite the remarkable progress in software-based neural networks, the traditional von-Neumann hardware architecture has suffered from limited energy efficiency while facing unprecedented large amounts of data. To meet the performance requirements of neuro-inspired computing, large-scale vector-matrix multiplication is preferred to be performed in situ, namely compute-in-memory. Non-volatile memory devices with different materials have been proposed for weight storage as synaptic devices. Among them, HfO 2 -based ferroelectric devices have attracted great attention because of their low energy consumption, good complementarymetal-oxide-semiconductor (CMOS) compatibility and multi-bit per cell potential. In this review, recent trends and prospects of the ferroelectric synaptic devices are surveyed. First, we present the three-terminal synaptic devices based on the ferroelectric field effect transistor (FeFET), and discuss the switching physics of the intermediate states, the back-end-of-line integration and the 3D NAND architecture design. Then, we introduce a hybrid precision synapse concept that leverages the volatile charges on the gate capacitor of the FeFET and the non-volatile polarization on the gate dielectric of the FeFET. Lastly, we review two-terminal synaptic devices using the ferroelectric tunnel junction (FTJ) and ferroelectric capacitor (FeCAP). The design margins of the crossbar array with FTJ and FeCAP analyzed.

show abstract

Section: D Nand-based Fefet For Trainingmentioning

confidence: 99%

Ferroelectric HfO₂-based synaptic devices: recent trends and prospects

Hur

Luo

et al. 2021

Semicond. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…The drawback of capacity augmentation bound is that it gives the same threshold for both normalized utilization and tensity, i.e., a task set with small normalized utilization but maximum tensity slightly exceeding capacity augmentation bound cannot pass the schedulability test even it is easy to be scheduled. From this view, another metric named utilizationtensity bound which aims to have asymmetric thresholds for normalized utilization and tensity on unit-speed processors are derived in [12]. Instead of deriving schedulability conditions on speed-s processors and then transforming them to speed-1 processors, conditions to guarantee schedulability on unitspeed processors for G-EDF are directly derived by obtaining the "unit-speed version" of Lemma 3 and Lemma 4: Lemma 7 ( [12]).…”

Section: B Utilization-tensity Boundmentioning

confidence: 99%

“…From this view, another metric named utilizationtensity bound which aims to have asymmetric thresholds for normalized utilization and tensity on unit-speed processors are derived in [12]. Instead of deriving schedulability conditions on speed-s processors and then transforming them to speed-1 processors, conditions to guarantee schedulability on unitspeed processors for G-EDF are directly derived by obtaining the "unit-speed version" of Lemma 3 and Lemma 4: Lemma 7 ( [12]). For any task τ i and t > 0, it holds…”

Section: B Utilization-tensity Boundmentioning

confidence: 99%

“…In the following, we show that it is not necessary to enumerate each instance of t ≥ T i to check (12)…”

Section: A Utilization-tensity Boundmentioning

confidence: 99%

“…Lemma 16 can be seen as a byproduct of this paper. The schedulability of τ under G-RM can be decided by checking (12) for each task τ i ∈ τ for each discrete value of t ∈ [T i , 2T i ) separated by the minimum time unit. Clearly, such test has a polynomial-time complexity.…”

Section: A Utilization-tensity Boundmentioning

confidence: 99%

See 2 more Smart Citations

Schedulability Bounds for Parallel Real-Time Tasks under Global Rate-Monotonic Scheduling

Jiang,

Guan,

Yang

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Schedulability bounds not only serve as efficient tests to decide schedulability of real-time task systems, but also reveal insights about the worst-case performance of scheduling algorithms. Different from sequential real-time task systems for which utilization is a suitable metric to develop schedulability bounds, schedulability of parallel real-time tasks depends on not only utilization but also the workload graph structure of tasks, which can be well represented by the tensity metric. In this paper, we develop new analysis techniques for parallel real-time tasks systems under Global Rate-Monotonic (G-RM) scheduling and obtain new results on schedulability bounds based on these two metrics: utilization and tensity. First, we develop the first utilization-tensity bound for G-RM. Second, we improve the capacity augmentation bound of G-RM from the best known value 3.73 to 3.18. These schedulability bounds not only provide theoretical insights about real-time performance of G-RM, but also serve as highly efficient schedulability tests, which are particularly suitable to design scenarios in which detailed task graph structures are unknown or may change at run-time. Experiments with randomly generated task sets show that our new results consistently outperform the state-of-the-art with a significant margin under different parameter settings.

show abstract

The shape of a DAG: bounding the response time using long paths

et al. 2023

Real-Time Syst

View full text Add to dashboard Cite

Utilization-Tensity Bound for Real-Time DAG Tasks under Global EDF Scheduling

Cited by 21 publications

References 23 publications

Ferroelectric HfO₂-based synaptic devices: recent trends and prospects

Ferroelectric HfO₂-based synaptic devices: recent trends and prospects

Schedulability Bounds for Parallel Real-Time Tasks under Global Rate-Monotonic Scheduling

The shape of a DAG: bounding the response time using long paths

Contact Info

Product

Resources

About

Utilization-Tensity Bound for Real-Time DAG Tasks under Global EDF Scheduling

Cited by 21 publications

References 23 publications

Ferroelectric HfO2-based synaptic devices: recent trends and prospects

Ferroelectric HfO2-based synaptic devices: recent trends and prospects

Schedulability Bounds for Parallel Real-Time Tasks under Global Rate-Monotonic Scheduling

The shape of a DAG: bounding the response time using long paths

Contact Info

Product

Resources

About

Ferroelectric HfO₂-based synaptic devices: recent trends and prospects

Ferroelectric HfO₂-based synaptic devices: recent trends and prospects