PRUNE: Dynamic and Decidable Dataflow for Signal Processing on Heterogeneous Platforms

Boutellier, Jani; Wu, Jiahao; Huttunen, Heikki; Bhattacharyya, Shuvra S.

doi:10.1109/tsp.2017.2773424

Cited by 17 publications

(23 citation statements)

References 31 publications

(58 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…PRUNE [11] is an open source framework for design and execution of dynamic and decidable dataflow applications on heterogeneous platforms. The PRUNE framework defines its own dynamic MoC, which is developed specifically for the requirements of heterogeneous platforms.…”

Section: Related Workmentioning

confidence: 99%

“…9. Dataflow network of the dynamic predistortion filter application network has been designed for TTADF from a floating point version of DPD network presented in PRUNE [11]. In the experiments, the token size is set to 256 bytes, which equals to 64 32-bit integer numbers.…”

Section: Dynamic Predistortion Filtermentioning

confidence: 99%

“…Dataflow [6], a well-known and widely used programming paradigm for expressing the functionality of signal processing and streaming applications, has been proposed as a next-generation programming solution for heterogeneous MPSoCs [7] [8] [9] [10] [11]. Dataflow programming is an intuitive approach for describing the parallelism of application software, while at the same time increasing its modularity, flexibility and re-usability.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

TTADF: Power Efficient Dataflow-Based Multicore Co-Design Flow

Hautala

Boutellier

Silvén

2020

IEEE Trans. Comput.

Self Cite

View full text Add to dashboard Cite

The era of mobile communications and the Internet of Things (IoT) has introduced numerous challenges for mobile processing platforms that are responsible for increasingly complex signal processing tasks from different application domains. In recent years, the power efficiency of computing has been improved by adding more parallelism and workload-specific computing resources to such platforms. However, programming of parallel systems can be time-consuming and challenging if only low-level programming methods are used. This work presents a dataflow-based co-design framework TTADF that reduces the design effort of both software and hardware design for mobile processing platforms. The paper presents three application examples from the fields of video coding, machine vision, and wireless communications. The application examples are mapped and profiled both on a pipelined and a shared-memory multicore platform that is generated by TTADF. The results of the TTADF co-design-based solutions are compared against previous manually created designs and a recent dataflow-based design flow, showing that TTADF provides very high energy efficiency together with a high level of automation in software and hardware design.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Dynamic Predistortion Filtermentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

TTADF: Power Efficient Dataflow-Based Multicore Co-Design Flow

Hautala

Boutellier

Silvén

2020

IEEE Trans. Comput.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Moreover, effective scheduling and memory management decisions can be distributed among PEs and taken at run-time. PRUNE [10] is an example of hybrid approach that can be used to design dataflow applications that are analyzable and flexible. It uses an extension of the Synchronous Data-Flow (SDF) formalism where actors are allowed to dynamically vary their production/consumption rates.…”

Section: Rmentioning

confidence: 99%

Odyn: Deadlock Prevention and Hybrid Scheduling Algorithm for Real-Time Dataflow Applications

Dauphin

Pacalet

Enrici

et al. 2019

2019 22nd Euromicro Conference on Digital System Design (DSD)

View full text Add to dashboard Cite

In recent wireless communication standards (4G, 5G), the growing need for dynamic adjustments of transmission parameters (e.g., modulation, bandwidth, channel coding rate) makes traditional static scheduling approaches less and less efficient. The reason being that precomputed fixed mapping and scheduling prevent the system from dynamically adapting to changes of the operating conditions (e.g. wireless channel quality, available bandwidth). In this paper, we present Odyn, a hybrid approach for the scheduling and memory management of periodic dataflow applications on parallel, heterogeneous, Non-Uniform Memory Architecture (NUMA) platforms. In Odyn, the ordering of tasks and memory allocation are distributed and computed simultaneously at run-time for each Processing Element. Odyn also proposes a mechanism to prevent deadlocks caused by attempts to allocate buffers in size-limited memories. This technique, based on the static computation of exclusion relations among buffers in a target application, removes the need for backtracking that is typical of dynamic scheduling algorithms. We demonstrate the effectiveness of Odyn on a testbench that simulates the interactions of randomly generated concurrent applications. We also demonstrate its deadlock prevention technique on a selection of use cases.

show abstract

“…It is a data-centric model and can be represented as a graph, where nodes are the operations that are applied to the incoming data, and edges represent the flow of data. It is used in different domains like signal processing [2], [3], [4], [5] and image [6] or video processing [7]. New languages based on this approach have also been developed [8], [9], [10] and [11].…”

Section: Introductionmentioning

confidence: 99%

Tracing and Profiling Machine Learning Dataflow Applications on GPU

Zins

Dagenais

2019

Int J Parallel Prog

View full text Add to dashboard Cite

In this paper, we propose a profiling and tracing method for dataflow applications with GPU acceleration. Dataflow models can be represented by graphs and are widely used in many domains like signal processing or machine learning. Within the graph, the data flows along the edges, and the nodes correspond to the computing units that process the data. To accelerate the execution, some co-processing units, like GPUs, are often used for computing intensive nodes. The work in this paper aims at providing useful information about the execution of the dataflow graph on the available hardware, in order to understand and possibly improve the performance. The collected traces include low level information about the CPU, from the Linux Kernel, as well as high-level information about the dataflow model. This is followed by post-mortem analysis and visualization steps in order to enhance the trace and show useful information to the user. To demonstrate the effectiveness of the method, it was evaluated for TensorFlow, a well-known machine learning library that uses a dataflow computational graph to represent the algorithms. We present a few examples of machine learning applications that can be optimized with the help of the information provided by our proposed method.

show abstract

PRUNE: Dynamic and Decidable Dataflow for Signal Processing on Heterogeneous Platforms

Cited by 17 publications

References 31 publications

TTADF: Power Efficient Dataflow-Based Multicore Co-Design Flow

TTADF: Power Efficient Dataflow-Based Multicore Co-Design Flow

Odyn: Deadlock Prevention and Hybrid Scheduling Algorithm for Real-Time Dataflow Applications

Tracing and Profiling Machine Learning Dataflow Applications on GPU

Contact Info

Product

Resources

About