Benchmarking of Neuromorphic Hardware Systems

Ostrau, Christoph; Klarhorst, Christian; Thies, Michael; Rückert, Ulrich

doi:10.1145/3381755.3381772

Cited by 8 publications

(7 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Aubie's model is formulated in NEURON; hence, a targeted neuromorphic hardware needs to support the portability of NEURON code by an application programming interface. Benchmarking of neuromorphic hardware systems helps to define standardized criteria of code mapping, execution, and measuring performance (Ostrau et al, 2020). A few neuromorphic hardware resources are available (Thakur et al, 2018).…”

Section: Discussionmentioning

confidence: 99%

Periodicity Pitch Perception

Klefenz

Harczos

2020

Front. Neurosci.

View full text Add to dashboard Cite

This study presents a computational model to reproduce the biological dynamics of "listening to music." A biologically plausible model of periodicity pitch detection is proposed and simulated. Periodicity pitch is computed across a range of the auditory spectrum. Periodicity pitch is detected from subsets of activated auditory nerve fibers (ANFs). These activate connected model octopus cells, which trigger model neurons detecting onsets and offsets; thence model interval-tuned neurons are innervated at the right interval times; and finally, a set of common interval-detecting neurons indicate pitch. Octopus cells rhythmically spike with the pitch periodicity of the sound. Batteries of interval-tuned neurons stopwatch-like measure the inter-spike intervals of the octopus cells by coding interval durations as first spike latencies (FSLs). The FSL-triggered spikes synchronously coincide through a monolayer spiking neural network at the corresponding receiver pitch neurons.

show abstract

Section: Discussionmentioning

confidence: 99%

Periodicity Pitch Perception

Klefenz

Harczos

2020

Front. Neurosci.

View full text Add to dashboard Cite

show abstract

“…We see this as a compromise between real “black-box” benchmarking and individual implementations for every platform. This has been accounted for in the Spiking Neural Architecture Benchmark Suite (SNABSuite) and its architecture was already proposed in Ostrau et al ( 2020b ) combined with a coarse overview of the benchmark approach. Its modular structure factored out all backend specific configuration and the benchmark implementation.…”

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

“…For NEST, Spikey and SpiNNaker the framework makes use of their PyNN interfaces, however, for BrainScaleS and GeNN a lower-level C++ interface is used. Furthermore, the proposed networks studied below are part of the Spiking Neural Architecture Benchmark Suite 3 (SNABSuite) [17,18], which also covers benchmarks like low-level synthetic characterizations and application-inspired (sub-)tasks with an associated framework for automated evaluation.…”

Section: Target Systems and Softwarementioning

confidence: 99%

Benchmarking Deep Spiking Neural Networks on Neuromorphic Hardware

Ostrau,

Homburg,

Klarhorst

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

With more and more event-based neuromorphic hardware systems being developed at universities and in industry, there is a growing need for assessing their performance with domain specific measures. In this work, we use the methodology of converting pre-trained nonspiking to spiking neural networks to evaluate the performance loss and measure the energy-per-inference for three neuromorphic hardware systems (BrainScaleS, Spikey, SpiNNaker) and common simulation frameworks for CPU (NEST) and CPU/GPU (GeNN). For analog hardware we further apply a re-training technique known as hardware-in-the-loop training to cope with device mismatch. This analysis is performed for five different networks, including three networks that have been found by an automated optimization with a neural architecture search framework. We demonstrate that the conversion loss is usually below one percent for digital implementations, and moderately higher for analog systems with the benefit of much lower energy-per-inference costs.

show abstract

Benchmarking of Neuromorphic Hardware Systems

Cited by 8 publications

References 15 publications

Periodicity Pitch Perception

Periodicity Pitch Perception

Benchmarking Neuromorphic Hardware and Its Energy Expenditure

Benchmarking Deep Spiking Neural Networks on Neuromorphic Hardware

Contact Info

Product

Resources

About