# An Evaluation of Edge TPU Accelerators for Convolutional Neural Networks

arXiv: https://arxiv.org/abs/2102.10423 · HTML: https://mlsanity.com/paper/2102.10423/

## Key facts

- Cited by 39 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 52 references that are in the map.
- First posted on arXiv 20 Feb 2021.

## Builds on (most cited first, 17 of 17)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Semi-Supervised Classification with Graph Convolutional Networks](https://mlsanity.com/paper/1609.02907/) (2016)
- [Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift](https://mlsanity.com/paper/1502.03167/) (2015)
- [Inductive Representation Learning on Large Graphs](https://mlsanity.com/paper/1706.02216/) (2017)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Relational inductive biases, deep learning, and graph networks](https://mlsanity.com/paper/1806.01261/) (2018)
- [In-Datacenter Performance Analysis of a Tensor Processing Unit](https://mlsanity.com/paper/1704.04760/) (2017)
- [NAS-Bench-101: Towards Reproducible Neural Architecture Search](https://mlsanity.com/paper/1902.09635/) (2019)
- [BANANAS: Bayesian Optimization with Neural Architectures for Neural Architecture Search](https://mlsanity.com/paper/1910.11858/) (2019)
- [Neural Predictor for Neural Architecture Search](https://mlsanity.com/paper/1912.00848/) (2019)
- [Ithemal: Accurate, Portable and Fast Basic Block Throughput Estimation using Deep Neural Networks](https://mlsanity.com/paper/1808.07412/) (2018)
- [Meta Architecture Search](https://mlsanity.com/paper/1812.09584/) (2018)
- [Apollo: Transferable Architecture Exploration](https://mlsanity.com/paper/2102.01723/) (2021)
- [Efficient Forward Architecture Search](https://mlsanity.com/paper/1905.13360/) (2019)
- [Learning Execution through Neural Code Fusion](https://mlsanity.com/paper/1906.07181/) (2019)
- [NASGEM: Neural Architecture Search via Graph Embedding Method](https://mlsanity.com/paper/2007.04452/) (2020)
- [Automated Instruction Stream Throughput Prediction for Intel and AMD Microarchitectures](https://mlsanity.com/paper/1809.00912/) (2018)

## Built on by (most cited first, 30 of 39)

- [Learning Performance-Improving Code Edits](https://mlsanity.com/paper/2302.07867/) (2023)
- [FLAT: An Optimized Dataflow for Mitigating Attention Bottlenecks](https://mlsanity.com/paper/2107.06419/) (2021)
- [Data-Driven Offline Optimization For Architecting Hardware Accelerators](https://mlsanity.com/paper/2110.11346/) (2021)
- [A Machine Learning-oriented Survey on Tiny Machine Learning](https://mlsanity.com/paper/2309.11932/) (2023)
- [A Full-Stack Search Technique for Domain Optimized Deep Learning Accelerators](https://mlsanity.com/paper/2105.12842/) (2021)
- [Face Deblurring using Dual Camera Fusion on Mobile Phones](https://mlsanity.com/paper/2207.11617/) (2022)
- [Real-Time Multi-View 3D Human Pose Estimation using Semantic Feedback to Smart Edge Sensors](https://mlsanity.com/paper/2106.14729/) (2021)
- [Binarized Neural Machine Translation](https://mlsanity.com/paper/2302.04907/) (2023)
- [Sparse Attention Acceleration with Synergistic In-Memory Pruning and On-Chip Recomputation](https://mlsanity.com/paper/2209.00606/) (2022)
- [GRANITE: A Graph Neural Network Model for Basic Block Throughput Estimation](https://mlsanity.com/paper/2210.03894/) (2022)
- [Accelerating Attention through Gradient-Based Learned Runtime Pruning](https://mlsanity.com/paper/2204.03227/) (2022)
- [Fail-Safe Human Detection for Drones Using a Multi-Modal Curriculum Learning Approach](https://mlsanity.com/paper/2109.13666/) (2021)
- [Edge Video Analytics: A Survey on Applications, Systems and Enabling Techniques](https://mlsanity.com/paper/2211.15751/) (2022)
- [RESPECT: Reinforcement Learning based Edge Scheduling on Pipelined Coral Edge TPUs](https://mlsanity.com/paper/2304.04716/) (2023)
- [ArchGym: An Open-Source Gymnasium for Machine Learning Assisted Architecture Design](https://mlsanity.com/paper/2306.08888/) (2023)
- [Parallel Algorithms Align with Neural Execution](https://mlsanity.com/paper/2307.04049/) (2023)
- [Hardware Acceleration of Explainable Machine Learning using Tensor Processing Units](https://mlsanity.com/paper/2103.11927/) (2021)
- [Edge Inference with Fully Differentiable Quantized Mixed Precision Neural Networks](https://mlsanity.com/paper/2206.07741/) (2022)
- [Deep Learning on Edge TPUs](https://mlsanity.com/paper/2108.13732/) (2021)
- [Rubik's Optical Neural Networks: Multi-task Learning with Physics-aware Rotation Architecture](https://mlsanity.com/paper/2304.12985/) (2023)
- [Precision-aware Latency and Energy Balancing on Multi-Accelerator Platforms for DNN Inference](https://mlsanity.com/paper/2306.05060/) (2023)
- [Tao: Re-Thinking DL-based Microarchitecture Simulation](https://mlsanity.com/paper/2404.10921/) (2024)
- [Google Coral-based edge computing person reidentification using human parsing combined with analytical method](https://mlsanity.com/paper/2209.11024/) (2022)
- [Machine Learning Fleet Efficiency: Analyzing and Optimizing Large-Scale Google TPU Systems with ML Productivity Goodput](https://mlsanity.com/paper/2502.06982/) (2025)
- [Exploring Deep Neural Networks on Edge TPU](https://mlsanity.com/paper/2110.08826/) (2021)
- [Flex-TPU: A Flexible TPU with Runtime Reconfigurable Dataflow Architecture](https://mlsanity.com/paper/2407.08700/) (2024)
- [GenAI for Energy-Efficient and Interference-Aware Compressed Sensing of GNSS Signals on a Google Edge TPU](https://mlsanity.com/paper/2605.14839/) (2026)
- [Monocular Cyclist Detection with Convolutional Neural Networks](https://mlsanity.com/paper/2303.11223/) (2023)
- [Hardware Acceleration of Explainable Artificial Intelligence](https://mlsanity.com/paper/2305.04887/) (2023)
- [Performance Analysis of DNN Inference/Training with Convolution and non-Convolution Operations](https://mlsanity.com/paper/2306.16767/) (2023)

## Abstract (arXiv)

Edge TPUs are a domain of accelerators for low-power, edge devices and are widely used in various Google products such as Coral and Pixel devices. In this paper, we first discuss the major microarchitectural details of Edge TPUs. Then, we extensively evaluate three classes of Edge TPUs, covering different computing ecosystems, that are either currently deployed in Google products or are the product pipeline, across 423K unique convolutional neural networks. Building upon this extensive study, we discuss critical and interpretable microarchitectural insights about the studied classes of Edge TPUs. Mainly, we discuss how Edge TPU accelerators perform across convolutional neural networks with different structures. Finally, we present our ongoing efforts in developing high-accuracy learned machine learning models to estimate the major performance metrics of accelerators such as latency and energy consumption. These learned models enable significantly faster (in the order of milliseconds) evaluations of accelerators as an alternative to time-consuming cycle-accurate simulators and establish an exciting opportunity for rapid hard-ware/software co-design.