Benchmarking Deep Spiking Neural Networks on Neuromorphic Hardware

Ostrau, Christoph; Homburg, Jonas Dominik; Klarhorst, Christian; Thies, Michael; Rückert, Ulrich

doi:10.1007/978-3-030-61616-8_49

Cited by 5 publications

(6 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As the work of Ostrau et al ( 2020 ) shows some similarities to our work, it is noteworthy that they pre-learn several conventional ANNs which are transformed into SNNs. It is shown that the SpiNNaker system can be efficient if it is used to its full extend, whereas the GeNN implementations are most suitable if one focuses primarily on short simulation times.…”

Section: Introductionsupporting

confidence: 66%

“…However, as noted in Diamond et al ( 2016 ) this implies the risk that individual strengths of the simulators are not accounted for. Despite this issue, in this paper, all SNNs are modeled with PyNN, whereas Ostrau et al ( 2020 ) opted to use the Cypress library and Diamond et al ( 2016 ) decided to model their networks in the native modeling languages of the simulators instead of using their PyNN interface. However, using PyNN to model all networks allows to simulate exactly the same model on all backends.…”

Section: Methodsmentioning

confidence: 99%

“…Firstly, a consumer grade power meter that allows to store time stamped data which can be extracted via an SD card. In Ostrau et al ( 2020 ), the Ruideng UM25C power meter and PeakTech9035 power meter are used. Secondly, image recordings of the power meter with subsequential digital post-processing.…”

Section: Methodsmentioning

confidence: 99%

“…The authors also outline the possibility of using GPU-based SNN controllers in robots. Benchmark scenarios more focused on a task applicable outside of the neuroscientific community are carried out in Diamond et al ( 2016 ) and Ostrau et al ( 2020 ). Both apply image classification and evaluate their models by using the MNIST data set.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Benchmarking Highly Parallel Hardware for Spiking Neural Networks in Robotics

et al. 2021

View full text Add to dashboard Cite

Animal brains still outperform even the most performant machines with significantly lower speed. Nonetheless, impressive progress has been made in robotics in the areas of vision, motion- and path planning in the last decades. Brain-inspired Spiking Neural Networks (SNN) and the parallel hardware necessary to exploit their full potential have promising features for robotic application. Besides the most obvious platform for deploying SNN, brain-inspired neuromorphic hardware, Graphical Processing Units (GPU) are well capable of parallel computing as well. Libraries for generating CUDA-optimized code, like GeNN and affordable embedded systems make them an attractive alternative due to their low price and availability. While a few performance tests exist, there has been a lack of benchmarks targeting robotic applications. We compare the performance of a neural Wavefront algorithm as a representative of use cases in robotics on different hardware suitable for running SNN simulations. The SNN used for this benchmark is modeled in the simulator-independent declarative language PyNN, which allows using the same model for different simulator backends. Our emphasis is the comparison between Nest, running on serial CPU, SpiNNaker, as a representative of neuromorphic hardware, and an implementation in GeNN. Beyond that, we also investigate the differences of GeNN deployed to different hardware. A comparison between the different simulators and hardware is performed with regard to total simulation time, average energy consumption per run, and the length of the resulting path. We hope that the insights gained about performance details of parallel hardware solutions contribute to developing more efficient SNN implementations for robotics.

show abstract

Section: Introductionsupporting

confidence: 66%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Benchmarking Highly Parallel Hardware for Spiking Neural Networks in Robotics

et al. 2021

View full text Add to dashboard Cite

show abstract

“…A third benchmark is the conversion of deep neural networks to SNNs (Ostrau et al, 2020a ). For this conversion, DNNs are trained using ReLU activation functions without biases, and for simplicity, only densely connected layers.…”

Section: Methodsmentioning

confidence: 99%

Benchmarking Neuromorphic Hardware and Its Energy Expenditure

et al. 2022

Self Cite

View full text Add to dashboard Cite

We propose and discuss a platform overarching benchmark suite for neuromorphic hardware. This suite covers benchmarks from low-level characterization to high-level application evaluation using benchmark specific metrics. With this rather broad approach we are able to compare various hardware systems including mixed-signal and fully digital neuromorphic architectures. Selected benchmarks are discussed and results for several target platforms are presented revealing characteristic differences between the various systems. Furthermore, a proposed energy model allows to combine benchmark performance metrics with energy efficiency. This model enables the prediction of the energy expenditure of a network on a target system without actually having access to it. To quantify the efficiency gap between neuromorphics and the biological paragon of the human brain, the energy model is used to estimate the energy required for a full brain simulation. This reveals that current neuromorphic systems are at least four orders of magnitude less efficient. It is argued, that even with a modern fabrication process, two to three orders of magnitude are remaining. Finally, for selected benchmarks the performance and efficiency of the neuromorphic solution is compared to standard approaches.

show abstract