[DL] A Survey of FPGA-based Neural Network Inference Accelerators
The DOI matches its Crossref record; 71 of the 71 references it cites were found in a registry or an index.
Abstract
Recent research on neural networks has shown a significant advantage in machine learning over traditional algorithms based on handcrafted features and models. Neural networks are now widely adopted in regions like image, speech, and video recognition. But the high computation and storage complexity of neural network inference poses great difficulty on its application. It is difficult for CPU platforms to offer enough computation capacity. GPU platforms are the first choice for neural network processes because of its high computation capacity and easy-to-use development frameworks. However, FPGA-based neural network inference accelerator is becoming a research topic. With specifically designed hardware, FPGA is the next possible solution to surpass GPU in speed and energy efficiency. Various FPGA-based accelerator designs have been proposed with software and hardware optimization techniques to achieve high speed and energy efficiency. In this article, we give an overview of previous work on neural network inference accelerators based on FPGA and summarize the main techniques used. An investigation from software to hardware, from circuit level to system level is carried out to complete analysis of FPGA-based neural network inference accelerator design and serves as a guide to future work.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“The research from Facebook shows the performance gain by porting DNN applications to specific hardware accelerators may not be able to compensate the significant engineering effort spent due to the poor ecosystems and program-ability [20].”
From Bringing ai to edge: from deep learning’s perspective (2022) · cites this paper as [20]
In the citing paper
“However, static DNN models are unable to achieve such guarantee under dynamic environment [20].”
From Bringing ai to edge: from deep learning’s perspective (2022) · cites this paper as [20]
In the citing paper
“Such variance may influence the quality of service (QoS) of applications without rigorously temporal requirement, like voice recognition, face recognition, machine translation, etc [20] and on the other hand may lead to catastrophic consequence for those with rigorously temporal requirement, (i.e., real-time requirement), like autonomous driving, UAV, etc.”
From Bringing ai to edge: from deep learning’s perspective (2022) · cites this paper as [20]
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 5×
“Besides CONV and FC layers, NN layers also have pooling, ReLU [28], concat [58], elementwise [22], and other types of layers.”
BackgroundMethodInfluentialVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 6×
“For certain networks like VGG [56], only 3 × 3 convolution kernels are used.”
BackgroundMethodInfluentialImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 26×
“In Table 1, we list the number of operations, number of parameters (add or multiplication), and top-1 accuracy on the ImageNet dataset [50] of state-of-the-art CNN models.”
BackgroundMethodInfluentialGoing deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380 · cited 7×
“In Table 1, we list the number of operations, number of parameters (add or multiplication), and top-1 accuracy on the ImageNet dataset [50] of state-of-the-art CNN models.”
MethodInfluentialRich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.81 · Cited by 32,684
SSD: Single Shot MultiBox Detector
Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.-Y., & Berg, A. C. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46448-0_2 · Cited by 22,128 · cited 1×
“New features like depthwise convolution [24] and the complex branch in SSD [34] may brings more design opportunities.”
BackgroundMobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., & Adam, H. (2017) · arXiv · doi:10.48550/arxiv.1704.04861 · Cited by 19,015 · cited 4×
“AlexNet [28] VGG19 [56] ResNet152 [22] MobileNet [24] ShuffleNet [79] Year 2012 2014 2016 2017 2017 # Param 60M 144M 57M 4.”
BackgroundMethodInfluentialProceedings of IEEE Conference on Computer Vision and Pattern Recognition
(1993) · doi:10.1109/cvpr.1993 · Cited by 17,667
Caffe: Convolutional Architecture for Fast Feature Embedding
Jia, Y., Shelhamer, E., Donahue, J., Karayev, S., Long, J., Girshick, R., Guadarrama, S., & Darrell, T. (2014) · Proceedings of the 22nd ACM international conference on Multimedia · doi:10.1145/2647868.2654889 · Cited by 11,077 · cited 3×
“Development frameworks like Caffe [26] and Tensorflow [4] also offer easy-to-use interfaces, which makes GPU the first choice of neural network acceleration.”
Method
Research works printed but not linked: 14
- DeePhi Tech . 2017. DNNDK. Retrieved December 3, 2018 from http://www.deephi.com/technology/dnndk. DeePhi Tech. 2017. DNNDK. Retrieved December 3, 2018 from http://www.deephi.com/technology/dnndk.
- Martín Abadi , Ashish Agarwal , Paul Barham , Eugene Brevdo , Zhifeng Chen , Craig Citro , Greg S. Corrado , Andy Davis , Jeffrey Dean , Matthieu Devin , 2016 . Tensorflow: A system for large-scale machine learning . In Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI’16) . 265--283. Martín Abadi, Ashish Agarwal, Paul Barham, Eugene Brevdo, Zhifeng Chen, Craig Citro, Greg S. Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, et al. 2016. Tensorflow: A system for large-scale machine learning. In Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI’16). 265--283.
- 10.5555/3195638.3195664 · doi:10.5555/3195638.3195664
- Wenlin Chen , James Wilson , Stephen Tyree , Kilian Weinberger , and Yixin Chen . 2015 . Compressing neural networks with the hashing trick . In Proceedings of the International Conference on Machine Learning. 2285--2294 . Wenlin Chen, James Wilson, Stephen Tyree, Kilian Weinberger, and Yixin Chen. 2015. Compressing neural networks with the hashing trick. In Proceedings of the International Conference on Machine Learning. 2285--2294.
- P. K. Gupta . 2016 . Accelerating datacenter workloads . In Proceedings of the 26th International Conference on Field Programmable Logic and Applications (FPL’16) . P. K. Gupta. 2016. Accelerating datacenter workloads. In Proceedings of the 26th International Conference on Field Programmable Logic and Applications (FPL’16).
- Awni Hannun Carl Case Jared Casper Bryan Catanzaro Greg Diamos Erich Elsen Ryan Prenger Sanjeev Satheesh Shubho Sengupta Adam Coates etal 2014. Deep speech: Scaling up end-to-end speech recognition. arXiv preprint arXiv:1412.5567 (2014). Awni Hannun Carl Case Jared Casper Bryan Catanzaro Greg Diamos Erich Elsen Ryan Prenger Sanjeev Satheesh Shubho Sengupta Adam Coates et al. 2014. Deep speech: Scaling up end-to-end speech recognition. arXiv preprint arXiv:1412.5567 (2014). · arXiv
- Fengfu Li , Bo Zhang , and Bin Liu . 2016. Ternary weight networks. arXiv preprint arXiv:1605.04711 ( 2016 ). Fengfu Li, Bo Zhang, and Bin Liu. 2016. Ternary weight networks. arXiv preprint arXiv:1605.04711 (2016). · arXiv
- Huimin Li , Xitian Fan , Li Jiao , Wei Cao , Xuegong Zhou , and Lingli Wang . 2016 . A high performance FPGA-based accelerator for large-scale convolutional neural networks . In Proceedings of the 26th International Conference on Field Programmable Logic and Applications (FPL’16) . IEEE, 1--9. Huimin Li, Xitian Fan, Li Jiao, Wei Cao, Xuegong Zhou, and Lingli Wang. 2016. A high performance FPGA-based accelerator for large-scale convolutional neural networks. In Proceedings of the 26th International Conference on Field Programmable Logic and Applications (FPL’16). IEEE, 1--9.
- Huizi Mao , Song Han , Jeff Pool , Wenshuo Li , Xingyu Liu , Yu Wang , and William J. Dally . 2017. Exploring the granularity of sparsity in convolutional neural networks . In Proceedings of the Computer Vision and Pattern Recognition Workshops. 1927--1934 . Huizi Mao, Song Han, Jeff Pool, Wenshuo Li, Xingyu Liu, Yu Wang, and William J. Dally. 2017. Exploring the granularity of sparsity in convolutional neural networks. In Proceedings of the Computer Vision and Pattern Recognition Workshops. 1927--1934.
- Duncan J. M. Moss , Eriko Nurvitadhi , Jaewoong Sim , Asit Mishra , Debbie Marr , Suchit Subhaschandra , and Philip H. W. Leong . 2017. High performance binary neural networks on the Xeon+FPGA™ platform . In Proceedings of the 27th International Conference on Field Programmable Logic and Applications (FPL’17) . IEEE, 1--4. Duncan J. M. Moss, Eriko Nurvitadhi, Jaewoong Sim, Asit Mishra, Debbie Marr, Suchit Subhaschandra, and Philip H. W. Leong. 2017. High performance binary neural networks on the Xeon+FPGA™ platform. In Proceedings of the 27th International Conference on Field Programmable Logic and Applications (FPL’17). IEEE, 1--4.
- 10.5555/3130379.3130591 · doi:10.5555/3130379.3130591
- 10.5555/3195638.3195659 · doi:10.5555/3195638.3195659
- Stylianos I. Venieris and Christos-Savvas Bouganis. 2017. Latency-driven design for FPGA-based convolutional neural networks . In Proceedings of the 27th International Conference on Field Programmable Logic and Applications (FPL’17) . IEEE, 1--8. Stylianos I. Venieris and Christos-Savvas Bouganis. 2017. Latency-driven design for FPGA-based convolutional neural networks. In Proceedings of the 27th International Conference on Field Programmable Logic and Applications (FPL’17). IEEE, 1--8.
- Xin Wang , Fisher Yu , Zi-Yi Dou , and Joseph E . Gonzalez . 2017 . Skipnet : Learning dynamic routing in convolutional networks. arXiv preprint arXiv:1711.09485 (2017). Xin Wang, Fisher Yu, Zi-Yi Dou, and Joseph E. Gonzalez. 2017. Skipnet: Learning dynamic routing in convolutional networks. arXiv preprint arXiv:1711.09485 (2017). · arXiv
Cited by
Items 201–300 of 575
A System-Level Transprecision FPGA Accelerator for BLSTM Using On-chip Memory Reshaping
Diamantopoulos, D., & Hagleitner, C. (2018) · International Conference on Field-Programmable Technology (FPT) · doi:10.1109/fpt.2018.00068 · Cited by 11
High-Throughput Machine Learning Approaches for Network Attacks Detection on FPGA
Ngo, D., Tran-Thanh, B., Dang, T., Tran, M. T., Thinh, T. N., & Pham‐Quoc, C. (2019) · Lecture notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering · doi:10.1007/978-3-030-34365-1_5 · Cited by 11
“While GPU is a good choice for the training phase of an ANN, FPGA is a promising candidate for processing inference phase [5,12].”
MethodAutomated Offline Design-Space Exploration and Online Design Reconfiguration for CNNs
Mazouz, A., & Bridges, C. (2020) · IEEE Conference on Evolving and Adaptive Intelligent Systems (EAIS) · doi:10.1109/eais48028.2020.9122697 · Cited by 11
Feature Map Analysis-Based Dynamic CNN Pruning and the Acceleration on FPGAs
Li, Q., Li, H., & Meng, L. (2022) · Electronics · doi:10.3390/electronics11182887 · Cited by 11
Aggressive undervolting of FPGAs : power & reliability trade-offs
Salami, B. (2018) · UPCommons institutional repository (Universitat Politècnica de Catalunya) · doi:10.5821/dissertation-2117-125840 · Cited by 11
Real-Time Driver Monitoring: Implementing FPGA-Accelerated CNNs for Pose Detection
Kim, M., & So, J. (2025) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2025.3554880 · Cited by 10
An Automated Design Flow for Adaptive Neural Network Hardware Accelerators
Ratto, F., Mainez, Á. P., Sau, C., Meloni, P., Deriu, G., Delucchi, S., Massa, M., Raffo, L., & Palumbo, F. (2023) · Journal of Signal Processing Systems · doi:10.1007/s11265-023-01855-x · Cited by 10
Vision-based robotics using open FPGAs
Machado, F., Nieto, R., Fernández-Conde, J., Lobato, D., & Cañas, J. M. (2023) · Microprocessors and Microsystems · doi:10.1016/j.micpro.2023.104974 · Cited by 10
Graph-OPU: A Highly Flexible FPGA-Based Overlay Processor for Graph Neural Networks
Tang, E., Li, S., Chen, R., Zhou, H., Ma, Y., Zhang, H., Yu, J. W., & Wang, K. (2024) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/3691636 · Cited by 10
“This substantial time investment is required when users need to adjust configuration parameters or switch models [12].”
Self-citationONNX-To-Hardware Design Flow for Adaptive Neural-Network Inference on FPGAs
Manca, F., Ratto, F., & Palumbo, F. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-78380-7_7 · Cited by 10
The AV says: Your Hardware Definitions Were Updated!
Botacin, M. F., Galante, L., Ceschin, F., Santos, P. C., Carro, L., de Geus, P. L., Gregio, A. R. A., & Alves, M. A. Z. (2019) · International Symposium on Reconfigurable Communication-centric Systems-on-Chip (ReCoSoC) · doi:10.1109/recosoc48741.2019.9034972 · Cited by 10
“Moreover, recent research work has been suggesting that FPGAs can outperform GPGPUs when operating as co-processors [16].”
BackgroundAn FPGA-based Convolution IP Core for Deep Neural Networks Acceleration
Nguyen, X.-Q., & Pham‐Quoc, C. (2022) · REV Journal on Electronics and Communications · doi:10.21553/rev-jec.286 · Cited by 10
Profiling of CNNs using the MATLAB FPGA-based Deep Learning Processor
Spanò, S., Canese, L., & Cardarilli, G. C. (2022) · Conference on Ph.D. Research in Microelectronics and Electronics · doi:10.1109/prime55000.2022.9816841 · Cited by 10
NERONE: The Fast Way to Efficiently Execute Your Deep Learning Algorithm at the Edge
Berzoini, R., D’Arnese, E., Conficconi, D., & Santambrogio, M. D. (2023) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2023.3296142 · Cited by 10
“Conversely, as shown in literature [13], [18], when the working scenario requires a memory footprint reduction, better energy efficiency, and less voluminous device, models with 8-bit integer (INT8) weights drastically reduce the required memory and power consumption with a negligible accuracy loss.”
BackgroundJakšić, Z., Cadenelli, N., Prats, D. B., Polo, J., Berral Garcia, J. L., & Perez, D. C. (2020) · Future Generation Computer Systems · doi:10.1016/j.future.2019.10.025 · Cited by 10
Exploring Resource-Efficient Acceleration Algorithm for Transposed Convolution of GANs on FPGA
Di, X., Yang, H., Huang, Z., Mao, N., Jia, Y., & Zheng, Y. (2019) · International Conference on Field-Programmable Technology (FPT) · doi:10.1109/icfpt47387.2019.00011 · Cited by 10
Self-citationOn-Chip Hardware-Aware Quantization for Mixed Precision Neural NetworksPreprint
Huang, W., Qin, H., Liu, Y., Liang, J., Zhang, Y., Li, Y., & Liu, X. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2309.01945 · Cited by 10
Accelerating ODE-Based Neural Networks on Low-Cost FPGAs
Watanabe, H., & Matsutani, H. (2021) · IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum · doi:10.1109/ipdpsw52791.2021.00021 · Cited by 9
Best Practices for the Deployment of Edge Inference: The Conclusions to Start Designing
Flamis, G., Kalapothas, S., & Kitsos, P. (2021) · Electronics · doi:10.3390/electronics10161912 · Cited by 9
A Novel Training Strategy for Deep Learning Model Compression Applied to Viral Classifications
Fernandes, M. A. C., & Kung, H. T. (2021) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn52387.2021.9534430 · Cited by 9
Niu, Z., Zhou, S., & Crespi, N. (2021) · Wiley eBooks · doi:10.1002/9781119765554.ch4 · Cited by 9
Extending the NOEL-V Platform with a RISC-V Vector Processor for Space Applications
Mascio, S. D., Menicucci, A., Gill, E. K. A., & Monteleone, C. (2023) · Journal of Aerospace Information Systems · doi:10.2514/1.i011097 · Cited by 9
A Low-Cost Neural ODE with Depthwise Separable Convolution for Edge Domain Adaptation on FPGAs
Kawakami, H., WATANABE, H., Sugiura, K., & Matsutani, H. (2023) · IEICE Transactions on Information and Systems · doi:10.1587/transinf.2022edp7149 · Cited by 9
Hardware Design of Lightweight Binary Classification Algorithms for Small-Size Images on FPGA
Saglam, S., & Bayar, S. (2024) · IEEE Access · doi:10.1109/access.2024.3390564 · Cited by 9
Diamantopoulos, D., Giefers, H., & Hagleitner, C. (2018) · IEEE Symposium in Low-Power and High-Speed Chips (COOL CHIPS) · doi:10.1109/coolchips.2018.8373077 · Cited by 9
A study on a low power optimization algorithm for an edge-AI device
Kaneko, T., Orimo, K., Hida, I., Takamaeda-Yamazaki, S., Ikebe, M., Motomura, M., & Asai, T. (2019) · Nonlinear Theory and Its Applications IEICE · doi:10.1587/nolta.10.373 · Cited by 9
Dynamically Reconfigurable Deep Learning for Efficient Video Processing in Smart IoT Systems
Eldash, O., Frost, A., Khalil, K., Kumar, A. S., & Bayoumi, M. (2020) · IEEE World Forum on Internet of Things (WF-IoT) · doi:10.1109/wf-iot48130.2020.9221101 · Cited by 9
INCAME: Interruptible CNN Accelerator for Multirobot Exploration
Yu, J., Xu, Z., Zeng, S., Yu, C., Qiu, J., Shen, Z., Xu, Y., Dai, G., Wang, Y., & Yang, H. (2021) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2021.3072340 · Cited by 9
Self-citationYang, T., He, Z., Kou, T., Li, Q., Han, Q., Yu, H., Liu, F., Liang, Y. E., & Jiang, L. (2021) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/3467476 · Cited by 9
Post Training Weight Compression with Distribution-based Filter-wise Quantization Step
Sasaki, S., Maki, A., Miyashita, D., & Deguchi, J. (2019) · IEEE Symposium in Low-Power and High-Speed Chips (COOL CHIPS) · doi:10.1109/coolchips.2019.8721356 · Cited by 9
FTT-NAS: Discovering Fault-Tolerant Neural Architecture.Preprint
Ning, X., Ge, G., Li, W., Zhu, Z., Zheng, Y., Chen, X., Gao, Z., Wang, Y., & Yang, H. (2020) · arXiv (Cornell University) · Cited by 9
Self-citationA runtime-adaptive transformer neural network accelerator on FPGAs
Kabir, E., Bakos, J. D., Andrews, D. A., & Huang, M. (2025) · Microprocessors and Microsystems · doi:10.1016/j.micpro.2025.105223 · Cited by 8
Mosses, A., & Prathap, P. M. J. (2024) · Journal of Computational Electronics · doi:10.1007/s10825-023-02123-8 · Cited by 8
“Due to non-ideal elements including process fluctuations, a lack of control accuracy, and dynamic disturbances, PNNs will unavoidably experience degradation in performance and or even malfunction [48].”
Karatsuba Matrix Multiplication and Its Efficient Custom Hardware Implementations
Pogue, T. E., & Nicolici, N. (2025) · IEEE Transactions on Computers · doi:10.1109/tc.2025.3525606 · Cited by 8
“While there are different possible multiplier circuit implementations, the area of multiplier circuits used in practice commonly scale quadratically with the area of a full adder [16], [17], [18], [22].”
Deepika, S., Arunachalam, V., & Raj, A. N. J. (2025) · Microprocessors and Microsystems · doi:10.1016/j.micpro.2025.105146 · Cited by 8
Lee, K., Cha, M., Kim, S., Nguyen, X. T., & Lee, H. (2025) · International Conference on Consumer Electronics · doi:10.1109/icce63647.2025.10930191 · Cited by 8
Kang, J., & Kim, T. (2019) · Integration · doi:10.1016/j.vlsi.2019.11.003 · Cited by 8
Data-Driven Real-Time Magnetic Tracking Applied to Myokinetic Interfaces
Mendez, S. P., Gherardini, M., Santos, G. V. D. P., Munoz, D. M., Ayala, H. V. H., & Cipriani, C. (2022) · IEEE Transactions on Biomedical Circuits and Systems · doi:10.1109/tbcas.2022.3161133 · Cited by 8
On How to Push Efficient Medical Semantic Segmentation to the Edge: the SENECA approach
Berzoini, R., D’Arnese, E., & Conficconi, D. (2022) · IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum · doi:10.1109/ipdpsw55747.2022.00027 · Cited by 8
“[11], [27], [28], employing weights with 8-bit integer (INT8)”
BackgroundMethodHardware-aware Partitioning of Convolutional Neural Network Inference for Embedded AI Applications
Kreß, F., Hoefer, J., Hotfilter, T., Walter, I., Sidorenko, V., Harbaum, T., & Becker, J. (2022) · International Conference on Distributed Computing in Smart Systems and the Internet of Things (DCOSS-IoT) · doi:10.1109/dcoss54816.2022.00034 · Cited by 8
Photonic tensor core for machine learning: a review
Peserico, N., Ma, X., Shastri, B., & Sorger, V. J. (2022) · Emerging Topics in Artificial Intelligence (ETAI) · doi:10.1117/12.2633916 · Cited by 8
Automatic Generation of Resource and Accuracy Configurable Processing Elements
León-Vega, L. G., Salazar-Villalobos, E., Rodriguez-Figueroa, A., & Godinez, J. C. (2023) · ACM Transactions on Embedded Computing Systems · doi:10.1145/3594540 · Cited by 8
“As a irst hint, the error mitigation might require ine-tuning training to it the model into the hardware to absorb the error characterisation of the accelerator [12].”
MethodHeidaryan, M., & Karimi, G. (2023) · Microprocessors and Microsystems · doi:10.1016/j.micpro.2023.104849 · Cited by 8
DOSA: Organic Compilation for Neural Network Inference on Distributed FPGAs
Ringlein, B., Abel, F., Diamantopoulos, D., Weiss, B., Hagleitner, C., & Fey, D. (2023) · IEEE International Conference on Edge Computing and Communications (EDGE) · doi:10.1109/edge60047.2023.00019 · Cited by 8
“To not exceed the scope of this section, we refer the reader to [20] for more background on OSAs and to [28], [29], or [40] for a detailed review of existing DNN-to-FPGA frameworks.”
BackgroundMethodAccelerating Compact Convolutional Neural Networks with Multi-threaded Data Streaming
Chen, W., Wang, Z., Li, S., Yu, Z., & Li, H. (2019) · IEEE Computer Society Annual Symposium on VLSI · doi:10.1109/isvlsi.2019.00099 · Cited by 8
“Additionally, depthwise convolution has a low computation to communication (CTC) ratio [8] as reported in [9], which makes it difficult to be executed as efficient as standard convolution.”
MethodActiMon: Unified JOP and ROP Detection With Active Function Lists on an SoC FPGA
Oh, H., Yang, M., Cho, Y., & Paek, Y. (2019) · IEEE Access · doi:10.1109/access.2019.2961416 · Cited by 8
Muon–Electron Pulse Shape Discrimination for Water Cherenkov Detectors Based on FPGA/SoC
Ordóñez, L. G. G., Molina, R., Crespo, M. L., Carrato, S., Ramponi, G., Cicuttin, A., Morales, I., & Perez, H. (2021) · Electronics · doi:10.3390/electronics10030224 · Cited by 8
Resource-constrained FPGA/DNN co-design
Zhang, Z., & Kouzani, A. Z. (2021) · Neural Computing and Applications · doi:10.1007/s00521-021-06113-4 · Cited by 8
“Field-programmable gate arrays (FPGAs) can efficiently support DNN inference due to their flexible and programmable architectures, power efficiency, and parallel architecture [35].”
BackgroundEdge Intelligence for Smart Metro Systems: Architecture and Enabling Technologies
Liu, X., Zhang, M., Zou, C., Yang, J., & Yan, X. (2021) · IEEE Network · doi:10.1109/mnet.211.2100302 · Cited by 8
Open-Source Educational Platform for FPGA Accelerated AI in Robotics
Malle, N. H., & Ebeid, E. (2022) · International Conference on Mechatronics and Robotics Engineering (ICMRE) · doi:10.1109/icmre54455.2022.9734102 · Cited by 8
“With their great flexibility and efficiency, FPGAs are well positioned in the race to become the hardware of choice on the edge, and much previous work [4] [5] [6] has been done on the design of such accelerators.”
BackgroundVéstias, M. P., Duarte, R. P., de Sousa, J. T., & Neto, H. C. (2022) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/3546182 · Cited by 8
“Dedicated hardware accelerators implemented in FPGAs and ASICs have shown very good performance and power efficiency [6, 20].”
BackgroundImplementation of Fully-Pipelined CNN Inference Accelerator on FPGA and HBM2 Platform
Nguyen, V.-C., & NAKASHIMA, Y. (2023) · IEICE Transactions on Information and Systems · doi:10.1587/transinf.2022edp7155 · Cited by 8
Bodiwala, S., & Nanavati, N. R. (2021) · International Journal of Information Technology · doi:10.1007/s41870-021-00682-2 · Cited by 8
Aridhi, E., Laabidi, K., & Mami, A. (2025) · Array · doi:10.1016/j.array.2025.100622 · Cited by 7
EXPRESS: A Framework for Execution Time Prediction of Concurrent CNNs on Xilinx DPU Accelerator
Goel, S., Kedia, R., Sen, R., & Balakrishnan, M. (2024) · ACM Transactions on Embedded Computing Systems · doi:10.1145/3697835 · Cited by 7
FPGA implementation of an optimized neural network for CFD acceleration
Çevik, G., Sarioglu, B., & Aka, İ. B. (2024) · AEU - International Journal of Electronics and Communications · doi:10.1016/j.aeue.2024.155574 · Cited by 7
Strassen Multisystolic Array Hardware Architectures
Pogue, T. E., & Nicolici, N. (2025) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2025.3530785 · Cited by 7
“This is because the hardware complexity of fixed-point multipliers typically scale quadratically with the input bitwidth compared with linearly for adders and registers [24], [25], [26], causing the hardware footprint of multipliers to dominate that of adders and registers.”
WCET Estimation for CNN Inference on FPGA SoC With Multi-DPU Engines
Zhang, W., Yu, Y., Jiang, X., Guan, N., Zhan, N., & Ju, L. (2025) · IEEE Transactions on Parallel and Distributed Systems · doi:10.1109/tpds.2025.3555968 · Cited by 7
Basco, K. J., Singh, A. R., Nasef, D., Hartnett, C., Ruane, M., Tagliarino, J., Nizich, M., & Toma, M. (2025) · Diagnostics · doi:10.3390/diagnostics15070903 · Cited by 7
“Furthermore, field-programmable gate array (FPGA)-accelerated architectures achieve 8–10× higher energy efficiency than GPUs through reduced-bitwidth computations [30].”
Society for Scientific Data Processing mbH Gottingen (GWDG), G., & Kunkel, J. M. (2024) · Journal of Artificial Intelligence & Robotics · doi:10.52768/3067-7947/1017 · Cited by 7
On the Use of FPGAs to Implement CNNs: A Brief Review
Arshad, M., Shahriar, S., & Sagahyroon, A. A. (2020) · International Conference on Computing, Electronics & Communications Engineering (iCCECE) · doi:10.1109/iccece49321.2020.9231243 · Cited by 7
dsODENet: Neural ODE and Depthwise Separable Convolution for Domain Adaptation on FPGAs
Kawakami, H., Watanabe, H., Sugiura, K., & Matsutani, H. (2022) · Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP) · doi:10.1109/pdp55904.2022.00031 · Cited by 7
Minimal Filtering Algorithms for Convolutional Neural Networks
Cariow, A., & Cariowa, G. (2021) · Studies in computational intelligence · doi:10.1007/978-3-030-74556-1_5 · Cited by 7
Designing A Compact Convolutional Neural Network Processor on Embedded FPGAs
Ling, Y.-C., Chin, H.-H., Wu, H., & Tsay, R. (2020) · IEEE Global Conference on Artificial Intelligence and Internet of Things (GCAIoT) · doi:10.1109/gcaiot51063.2020.9345903 · Cited by 7
“Second, according to the previous survey [10], external memory access can cost 50 ~ 70% of processing time, which should be considered at early design stage.”
BackgroundMethodInfluentialArchitecting for Artificial Intelligence with Emerging Nanotechnology
Kulkarni, S., Bhat, S., & Moritz, C. A. (2021) · ACM Journal on Emerging Technologies in Computing Systems · doi:10.1145/3445977 · Cited by 7
Real-time fast learning hardware implementation
Zhang, M. J., Garcia, S., & Terré, M. (2023) · International Journal for Simulation and Multidisciplinary Design Optimization · doi:10.1051/smdo/2023001 · Cited by 7
Rodríguez, I., Campos, C., & Fdez‐Riverola, F. (2023) · Neurocomputing · doi:10.1016/j.neucom.2023.126835 · Cited by 7
Compilation and Optimizations for Efficient Machine Learning on Embedded Systems
Zhang, X., Chen, Y., Hao, C., Huang, S., Li, Y., & Chen, D. (2023) · Springer eBooks · doi:10.1007/978-3-031-39932-9_3 · Cited by 7
Implementation of convolutional-LSTM network based on CPU, GPU and pynq-zl board
Mahjoub, A. B., & Atri, M. (2019) · IEEE International Conference on Design & Test of Integrated Micro & Nano-Systems (DTS) · doi:10.1109/dtss.2019.8915287 · Cited by 7
PoET-BiN: Power Efficient Tiny Binary NeuronsPreprint
Chidambaram, S., Langlois, J. M. P., & David, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.09794 · Cited by 7
NN2CAM: Automated Neural Network Mapping for Multi-Precision Edge Processing on FPGA-Based CamerasPreprint
Jokic, P., Emery, S., & Benini, L. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.12840 · Cited by 7
Posso, J., Kieffer, H., Menga, N., Hlimi, O., Tarris, S., Guérard, H., Bois, G., Couderc, M., & Jenn, É. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2503.08700 · Cited by 7
“Abdelouahab et al. [1] and Guo et al. [9] review designs for neural network accelerators, with a particular emphasis on enhancing FPGA inference within image classification networks.”
CNN inference acceleration on limited resources FPGA platforms_epilepsy detection case study
Saidi, A., Othman, S. B., Dhouibi, M., & Saoud, S. B. (2023) · International Journal of Informatics and Communication Technology (IJ-ICT) · doi:10.11591/ijict.v12i3.pp251-260 · Cited by 6
“For example, visual geometry group-19 (VGG-19) model needs up to 39 billion floating point operations for the classification of an input image with a size of 224×224 and more than 500 MB of model parameters [7].”
An Approach to the Systematic Characterization of Multitask Accelerated CNN Inference in Edge MPSoCs
Cilardo, A., Maisto, V., Mazzocca, N., & di Torrepadula, F. R. (2023) · ACM Transactions on Embedded Computing Systems · doi:10.1145/3611015 · Cited by 6
“The interested reader can refer to other works [1, 13, 36] for a detailed discussion.”
BackgroundMovahedin, A., Landsmeer, L. P. L., & Strydis, C. (2025) · ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (FPGA) · doi:10.1145/3706628.3708875 · Cited by 6
Improving Long‐Term Glucose Prediction Accuracy with Uncertainty‐Estimated ProbSparse‐Transformer
Huang, W. H., Fan, N., Wang, W., Wang, J., Qi, X., & Zhang, S. (2025) · Advanced Intelligent Systems · doi:10.1002/aisy.202500235 · Cited by 6
Learning-assisted schedulability analysis: opportunities and limitations
Baruah, S., Ekberg, P., & Sudvarg, M. (2025) · Real-Time Systems · doi:10.1007/s11241-025-09450-y · Cited by 6
“The rapid recent increase in size and complexity of NNs has spurred interest in performing DNN inference on specially-deployed FPGA kernels (Guo et al. 2019), often achieving highly-predictable execution times (Huang et al. 2019; Khoda et al. 2023).”
FPGA-Accelerated GNN Pipelines for Real-Time Identity Threat Scoring in Zero-Trust Cloud Systems
Smith, O. J. M. (2025) · Journal of Reconfigurable Hardware Architectures and Embedded Systems · Cited by 6
Characterization of Execution Time Variability in FPGA-Based AI-Accelerators
Kirschner, M., Peccia, F. N., Thömmes, F., Betancourt, V. P., Bringmann, O., & Becker, J. (2023) · IEEE Intl Conf on Dependable, Autonomic and Secure Computing, Intl Conf on Pervasive Intelligence and Computing, Intl Conf on Cloud and Big Data Computing, Intl Conf on Cyber Science and Technology Congress (DASC/PiCom/CBDCom/CyberSciTech) · doi:10.1109/dasc/picom/cbdcom/cy59711.2023.10361287 · Cited by 6
“Convolution layers are of particular interest here, as they often account for a large proportion of the computing time [18] and the here investigated open-source AI accelerators are mainly optimized to accelerate convolution layers.”
BackgroundA Hardware Approach For Accelerating Inductive Learning In Description Logic
Algahtani, E. (2024) · ACM Transactions on Embedded Computing Systems · doi:10.1145/3665277 · Cited by 6
“…hardware approaches were successfully used to improve performance for other (non-ILP) ML algorithms such as for Artificial Neural Networks (ANNs) [20, 21], in this work, we propose (and evaluate) a hardware-based approach that explores the potential for hardware-based approaches to improve the…”
GAME: Gaussian Mixture Model Mapping and Navigation Engine on Embedded FPGA
Xu, Y., Zhang, Z., Yu, J., Cao, J., Dong, H., Huang, Z., Wang, Y., & Yang, H. (2021) · IEEE International Symposium on Field-Programmable Custom Computing Machines (FCCM) · doi:10.1109/fccm51124.2021.00015 · Cited by 6
“For example, the FPGA resource consumption for multiplier and adder with 32-bit fixed-point data is about 15 times more than that with 8-bit fixed-point data [8].”
MethodSelf-citationAnalysis of Fully-Pipelined CNN Implementation on FPGA and HBM2
Nguyen, V., & Nakashima, Y. (2021) · International Symposium on Computing and Networking Workshops (CANDARW) · doi:10.1109/candarw53999.2021.00029 · Cited by 6
Machine Learning Across Network-Connected FPGAs
Diaconu, D., Xie, Y., Gungor, M., Handagala, S., Lin, X., & Leeser, M. E. (2023) · IEEE High Performance Extreme Computing Conference · doi:10.1109/hpec58863.2023.10363454 · Cited by 6
An Efficient Streaming Accelerator for Low Bit-Width Convolutional Neural Networks
Chen, Q., Fu, Y., Song, W., Cheng, K., Lu, Z., Zhang, C., & Li, L. (2019) · Electronics · doi:10.3390/electronics8040371 · Cited by 6
Advancing Compilation of DNNs for FPGAs Using Operation Set Architectures
Ringlein, B., Abel, F., Diamantopoulos, D., Weiss, B., Hagleitner, C., & Fey, D. (2022) · IEEE Computer Architecture Letters · doi:10.1109/lca.2022.3227643 · Cited by 6
“Instead of building such a framework from scratch, we decided to leverage the plentiful existing frameworks and tool flows [8], [9] to avoid re-inventing the wheel.”
Elouaret, T., Colomer, S., de Melo, F., Cuperlier, N., Romain, O., Kessal, L., & Zuckerman, S. (2023) · Sensors · doi:10.3390/s23104631 · Cited by 6
Embedded Image Processing the European Way: A new platform for the future automotive market
Hotfilter, T., Kempf, F. J., Becker, J., Reinhardt, D., & Baili, I. (2020) · IEEE World Forum on Internet of Things (WF-IoT) · doi:10.1109/wf-iot48130.2020.9221396 · Cited by 6
On the Effects of Quantisation on Model Uncertainty in Bayesian Neural NetworksPreprint
Ferianc, M., Maji, P., Mattina, M., & Rodrigues, M. R. D. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2102.11062 · Cited by 6
A Quantized Neural Network Library for Proper Implementation of Hardware Emulation
Kiyama, M., Nakahara, Y., Amagasaki, M., & Iida, M. (2019) · 2019 Seventh International Symposium on Computing and Networking Workshops (CANDARW) · doi:10.1109/candarw.2019.00032 · Cited by 5
Secure Video Streaming Using Dedicated Hardware
Murray-Hill, N., Fontes, L., Machado, P., & Ihianle, I. K. (2023) · Journal of Signal Processing Systems · doi:10.1007/s11265-023-01886-4 · Cited by 5
Kang, M., & Park, D. (2025) · Sensors · doi:10.3390/s25247448 · Cited by 5
“These hardware hot swaps complement software-only methods by enabling structural changes in the accelerator pipeline under latency and power constraints [20].”
Context-Aware Layer Scheduling for Seamless Neural Network Inference in Cloud-Edge Systems
Stammler, M., Sidorenko, V., Kreß, F., Schmidt, P., & Becker, J. (2023) · IEEE International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC) · doi:10.1109/mcsoc60832.2023.00022 · Cited by 5
Hardware/Software Co-design for Convolutional Neural Networks Acceleration: A Survey and Open Issues
Pham‐Quoc, C., Nguyen, X.-Q., & Thinh, T. N. (2021) · Lecture notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering · doi:10.1007/978-3-030-93179-7_13 · Cited by 5
Data and Computation Reuse in CNNs Using Memristor TCAMs
de Moura, R. F., de Lima, J. P. C., & Carro, L. (2022) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/3549536 · Cited by 5
Iterative pruning algorithm for efficient look-up table implementation of binary neural networks
Ebrahimi, A., Pullu, V. N., Langlois, J. M. P., & David, J. (2023) · IEEE Interregional NEWCAS Conference (NEWCAS) · doi:10.1109/newcas57931.2023.10198137 · Cited by 5
EFFECT: An End-to-End Framework for Evaluating Strategies for Parallel AI Anomaly Detection
Stammler, M., Hoefer, J., Kraus, D., Schmidt, P., Hotfilter, T., Harbaum, T., & Becker, J. (2023) · Procedia Computer Science · doi:10.1016/j.procs.2023.08.188 · Cited by 5
O-2A: Outlier-Aware Compression for 8-bit Post-Training Quantization Model
Ho, N.-D., & Chang, I. (2023) · IEEE Access · doi:10.1109/access.2023.3311027 · Cited by 5
“(FPGAs) [34], have proven to be effective in reducing memory footprint through optimized memory access.”
BackgroundLeveraging Mixed-Precision CNN Inference for Increased Robustness and Energy Efficiency
Hotfilter, T., Hoefer, J., Merz, P., Kreß, F., Kempf, F., Harbaum, T., & Becker, J. (2023) · IEEE International System-on-Chip Conference (SOCC) · doi:10.1109/socc58585.2023.10256738 · Cited by 5
PreCog: Near-Storage Accelerator for Heterogeneous CNN Inference
An, J., Aliaj, E., & Jun, S.-W. (2023) · IEEE International Conference on Application-specific Systems, Architectures and Processors (ASAP) · doi:10.1109/asap57973.2023.00021 · Cited by 5
Ghavami, B., Sajadi, M., Shannon, L., & Wilton, S. J. E. (2024) · IEEE Computer Society Annual Symposium on VLSI · doi:10.1109/isvlsi61997.2024.00049 · Cited by 5
Metadata
- Authors
- Guo, K., Zeng, S., Yu, J., Wang, Y., & Yang, H.
- Affiliations
- Tsinghua University (CN)
- Venue
- ACM Transactions on Reconfigurable Technology and Systems
- Published
- 2019-03-28
- Volume
- 12
- Issue
- 1
- Pages
- 1-26
- ISSN
- 1936-7406 (ISSN-L)
- 1936-7414
- OpenAlex
- W2997106510
- Cited by
- 292
- References
- 65
- Citations by year
- 2026: 43 · 2025: 37 · 2024: 29 · 2023: 44 · 2022: 42 · 2021: 54 · 2020: 34 · 2019: 9
- FWCI
- 35.21
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- ACM Transactions on Reconfigurable Technology and Systems (publishedVersion)
- Funders
- National Key R&D Program of China (2018YFB0105005, 2017YFA0207600)
- National Natural Science Foundation of China (61622403, 61621091)
- Topics
- Physical Sciences › Computer Science › Hardware and Architecture › Embedded Systems Design Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Engineering › Electrical and Electronic Engineering › VLSI and FPGA Design Techniques
- Keywords
- FPGA, neural network inference, hardware acceleration, energy efficiency, program optimization, hardware optimization, GPU comparison, machine learning hardware
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.