Hybrid Dataflow/von-Neumann Architectures

Yazdanpanah, Fahimeh; Martínez, Carlos Álvarez; Jiménez-González, Daniel; Etsion, Yoav

doi:10.1109/tpds.2013.125

Cited by 57 publications

(25 citation statements)

References 99 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The main difference lies in the second subfiring: the execution of the filter is replaced by the execution of the subgraph, called subiteration. For instance, in an H.264 video decoder, a subiteration may correspond to the decoding of one macroblock 4 . Subiterations are defined analogously to DKPN iterations: all actors in the subgraph have to be in end-of-iteration state for the subiteration to terminate.…”

Section: Hierarchical Compositionmentioning

confidence: 99%

“…To tackle this problem, this paper presents DKPN, a parameterized dataflow/Kahn process networks execution model that aims at: (1) bridging the gap between high-level models and real implementations; (2) supporting heterogeneous platforms, including hybrid Von Neumann/dataflow architectures [4]; (3) capturing the highest level of application dynamism. In detail, the contributions of the paper are: (1) a composition semantics for dataflow and Kahn process networks; (2) parameters to address variability and unpredictability at run time; (3) a scalable, dynamic, distributed scheduler to back the execution model; (4) effective scheduling strategies.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DKPN: A Composite Dataflow/Kahn Process Networks Execution Model

Arras

Fuin

Jeannot

et al. 2016

2016 24th Euromicro International Conference on Parallel, Distributed, and Network-Based Processing (PDP)

View full text Add to dashboard Cite

Section: Hierarchical Compositionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

DKPN: A Composite Dataflow/Kahn Process Networks Execution Model

Arras

Fuin

Jeannot

et al. 2016

2016 24th Euromicro International Conference on Parallel, Distributed, and Network-Based Processing (PDP)

View full text Add to dashboard Cite

“…The use of parallelism in the hybrid flow of input data. 7 6. The methods based on the use of parallelism in a homogeneous structure of processors.…”

Section: Resources and Related Workmentioning

confidence: 99%

Algorithmization and Optimization of Processing of Big Geographical Data

Bartoněk

Bureš

2016

j comput theor nanosci

View full text Add to dashboard Cite

This paper presents the optimization of evaluation of large volume of geographic data. The core of the method is hierarchical decomposition of the set of processes into elementary processes and the allocation of means to these processes. The means can be of three types: hardware, software or human factor, eventually combination of these types. Each elementary process can be processed at one of these means in certain time. Generally, the processes and the means can be interdependent or independent. The described problem can be represented using an oriented graph, where nodes correspond to the processes or the means and edges represent either the interdependence of processes and means, or the processing time of certain process on a given mean. The map of processes is formed on the basis of the graph. This map contains temporal continuity of solutions of sub-processes. Then, the duration of all processes is compiled from this map, which must be less than the time solving a task in the required quality of results. If not, the pairs of sub process-mean are replaced alternative pairs according to the map of processes with lower duration. The special algorithm was designed for this task. If the sum of the durations of all processes complies with solutions, the optimization ends and at this time the sub-processes and their allocated means are defined. The proposed method of data processing was realized in the project of data analysis of storage of gas facilities under certain types of terrain surface in the Czech Republic with the area of 64,350 km 2 .

show abstract

“…Data-flow execution models had been studied widely studied [16] as they provide a simple an elegant way to efficiently move data from one computational thread to another one [17], [18].…”

Section: Related Workmentioning

confidence: 99%

Scalable Embedded Systems: Towards the Convergence of High-Performance and Embedded Computing

Giorgi

2015

2015 IEEE 13th International Conference on Embedded and Ubiquitous Computing

View full text Add to dashboard Cite

Embedded System toolchains are highly customized for a specific System-on-Chip (SoC).\ud When the application needs more performance, the designer is typically forced\ud to adopt a new SoC and possibly another toolchain.\ud The rationale for not scaling performance by using, e.g., two SoCs, is\ud that maintining most of the operations on-chip may allow for higher energy efficiency.\ud We are exploring the feasibility and trade-offs of designing and manufacturing a new Single Board Computer (SBC) that could serve flexibly for a number of current and future applications, by allowing scalability through clusters of SBCs while keeping the same programming model for the SBC.\ud This board is based on FPGAs and embedded processors, and its key points are: i) a fast custom interconnect for board-to-board communication and ii) an easily programmable environment which would allow both the off-loading of code into accelerators (either soft-IP blocks or hard-IP blocks) and, at the same time, the distribution of computation across boards.\ud A key challenge to successfully deploying this paradigm is to properly distribute the threads across several boards without the explicit intervention of the programmer.\ud In this paper we describe how to dynamically and efficiently distribute the computational threads in symbiosis with an appropriate memory model to allow the system scalability, so that we can double the performance by simply connecting two boards without i) changing the basic hardware components (e.g., to a different System-On-Chip) and ii) changing the programming model to follow the vendor specific toolchain. Our approach is to reduce data movement across boards. Our initial experiments have confirmed the feasibility of our approach

show abstract

Hybrid Dataflow/von-Neumann Architectures

Cited by 57 publications

References 99 publications

DKPN: A Composite Dataflow/Kahn Process Networks Execution Model

DKPN: A Composite Dataflow/Kahn Process Networks Execution Model

Algorithmization and Optimization of Processing of Big Geographical Data

Scalable Embedded Systems: Towards the Convergence of High-Performance and Embedded Computing

Contact Info

Product

Resources

About