Customization methodology for implementation of streaming aggregation in embedded systems

Papadopoulos, Lazaros; Soudris, Dimitrios; Walulya, Ivan; Tsigas, Philippas

doi:10.1016/j.sysarc.2016.04.013

Cited by 5 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Efficient local memory utilization will reduce the number of DMA transfers which will benefit the energy consumption. Additionally, extensive experimentation with Myriad2 has shown that heavy DMA usage reduces the performance of the DMA engine, with negative impact on application's execution time [25]. Furthermore, the management of the available processing units, such as the efficient partitioning of the algorithm between them and the memory alignment issues should be carefully examined to effectively exploit the provided SIMD features.…”

Section: Edge Devices and Cnn Implementation Challengesmentioning

confidence: 99%

A Design Space Exploration Framework for Convolutional Neural Networks Implemented on Edge Devices

Tsimpourlas

Papadopoulos

Bartsokas

et al. 2018

IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst.

Self Cite

View full text Add to dashboard Cite

Deploying convolutional neural networks (CNNs) in embedded devices that operate at the edges of Internet of Things (IoT) networks provides various advantages in terms of performance, energy efficiency, and security in comparison with the alternative approach of transmitting large volumes of data for processing to the cloud. However, the implementation of CNNs on low power embedded devices is challenging due to the limited computational resources they provide and to the large resource requirements of state-of-the-art CNNs. In this paper, we propose a framework for the efficient deployment of CNNs in low power processor-based architectures used as edge devices in IoT networks. The framework leverages design space exploration (DSE) techniques to identify efficient implementations in terms of execution time and energy consumption. The exploration parameter is the utilization of hardware resources of the edge devices. The proposed framework is evaluated using a set of 6 state-of-the-art CNNs deployed in the Intel/Movidius Myriad2 low power embedded platform. The results show that using the maximum available amount of resources is not always the optimal solution in terms of performance and energy efficiency. Fine-tuned resource management based on DSE, reduces the execution time up to 3.6% and the energy consumption up to 7.7% in comparison with straightforward implementations.

show abstract

Section: Edge Devices and Cnn Implementation Challengesmentioning

confidence: 99%

A Design Space Exploration Framework for Convolutional Neural Networks Implemented on Edge Devices

Tsimpourlas

Papadopoulos

Bartsokas

et al. 2018

IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Furthermore, in ported memories, such as in Myr-iad2, stalls may appear under heavy data sharing. Finally, extensive experimentation in Myriad2 has shown that the DMA engine performance is reduced under heavy utilization [12]. Software techniques that optimize data management, improve the utilization of local memories and exploit parallelism can improve both performance and energy efficiency.…”

Section: Edge Devices and Cnn Implementation Challengesmentioning

confidence: 99%

“…However, this scheme imposes frequent DMA data transfers between the local and the global memory during the execution of a CNN inference, due to the limited local memory space available for data. Reducing the frequency of DMA transfers can benefit performance significantly: The effects of DMA data transfers to the performance and the energy consumption in Myriad have been extensively examined in [12]. Allocating the vector processing units' instruction code in the global memory and fetching instructions through the cache subsystem increases the local memory space available for CNN data.…”

Section: Data Transfer and Management Optimizationsmentioning

confidence: 99%

Efficient Winograd-based Convolution Kernel Implementation on Edge Devices

Xygkis

Soudris²,

Papadopoulos³

et al. 2018

2018 55th ACM/ESDA/IEEE Design Automation Conference (DAC)

Self Cite

View full text Add to dashboard Cite

The implementation of Convolutional Neural Networks on edge Internet of Things (IoT) devices is a significant programming challenge, due to the limited computational resources and the real-time requirements of modern applications. This work focuses on the efficient implementation of the Winograd convolution, based on a set of application-independent and Winograd-specific software techniques for improving the utilization of the edge devices computational resources. The proposed techniques were evaluated in Intel/Movidius Myriad2 platform, using 4 CNNs of various computational requirements. The results show significant performance improvements, up to 54%, over other convolution algorithms.

show abstract

“…Inefficient software can drive low-power hardware to waste the system's energy budget, regardless of the application performance [8]. Relevant works in the literature focus on sourceto-source optimizations at application level for increasing application performance and energy efficiency by improving the data flow and memory utilization [10,15].…”

Section: Introductionmentioning

confidence: 99%

A Flexible Tool for Estimating Applications Performance and Energy Consumption Through Static Analysis

et al. 2021

Self Cite

View full text Add to dashboard Cite

The design requirements of modern applications that target embedded systems, such as the need for high performance and low energy consumption, impose challenges on developers. Software tools capable of providing performance and energy consumption estimations are useful for addressing these challenges. Such tools aim to reduce development time and alleviate the time-to-market pressure. In this work, we propose a flexible tool that enables the estimation of performance and energy consumption of the application on embedded devices, providing a complete methodology based on which the user can add estimation models for various platforms. In contrast to existing tools that either rely on dynamic instrumentation or require detailed modeling of the hardware, the proposed tool leverages static analysis techniques applied at instruction level coupled with data-driven regression models. The proposed method is tested using a widely used benchmark suite for evaluation.

show abstract

Customization methodology for implementation of streaming aggregation in embedded systems

Cited by 5 publications

References 6 publications

A Design Space Exploration Framework for Convolutional Neural Networks Implemented on Edge Devices

A Design Space Exploration Framework for Convolutional Neural Networks Implemented on Edge Devices

Efficient Winograd-based Convolution Kernel Implementation on Edge Devices

A Flexible Tool for Estimating Applications Performance and Energy Consumption Through Static Analysis

Contact Info

Product

Resources

About