Fused-layer CNN accelerators
The DOI matches its Crossref record; 20 of the 20 references it cites were found in a registry or an index.
Abstract
Deep convolutional neural networks (CNNs) are rapidly becoming the dominant approach to computer vision and a major component of many other pervasive machine learning tasks, such as speech recognition, natural language processing, and fraud detection. As a result, accelerators for efficiently evaluating CNNs are rapidly growing in popularity. The conventional approaches to designing such CNN accelerators is to focus on creating accelerators to iteratively process the CNN layers. However, by processing each layer to completion, the accelerator designs must use off-chip memory to store intermediate data between layers, because the intermediate data are too large to fit on chip. In this work, we observe that a previously unexplored dimension exists in the design space of CNN accelerators that focuses on the dataflow across convolutional layers. We find that we are able to fuse the processing of multiple CNN layers by modifying the order in which the input data are brought on chip, enabling caching of intermediate data between the evaluation of adjacent CNN layers. We demonstrate the effectiveness of our approach by constructing a fused-layer CNN accelerator for the first five convolutional layers of the VGGNet-E network and comparing it to the state-of-the-art accelerator implemented on a Xilinx Virtex-7 FPGA. We find that, by using 362KB of on-chip storage, our fused-layer accelerator minimizes off-chip feature map data transfer, reducing the total transfer by 95%, from 77MB down to 3.6MB per image.
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
“When the number of worker nodes increases from 2 to 4, MoDNN can speedup the DNN computation by 2.17-4.28×[130]”
From Convergence of Edge Computing and Deep Learning: A Comprehensive Survey (2020) · cites this paper as [130]
In the citing paper
“With regard to specific DNN structures, e.g., CNN, a finer grid partitioning can be applied to minimize communication, synchronization, and memory overhead [130].”
From Convergence of Edge Computing and Deep Learning: A Comprehensive Survey (2020) · cites this paper as [130]
Citations
Cites
Very 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×
“Table II shows the results of a similar comparison, but for the larger and newer VGGNet-E network [17].”
BackgroundMethodResultInfluentialImageNet 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 4×
“(We observe a similar pattern in other common CNN structures, such as AlexNet [8].”
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 3×
“Large Scale Visual Recognition Challenge (ILSVRC) [15], uses five convolutional layers, while VGG [17] (winner of a portion of ILSVRC 2014) uses up to 17 convolutional layers.”
MethodOptimizing FPGA-based Accelerator Design for Deep Convolutional Neural Networks
Zhang, C., Li, P., Sun, G., Guan, Y., Xiao, B., & Cong, J. (2015) · Proceedings of the 2015 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays · doi:10.1145/2684746.2689060 · Cited by 2,158 · cited 10×
“The computational dominance of the convolutional layers, coupled with the need for larger and deeper networks, has sparked significant interest in the design and optimization of accelerator structures for these layers [19], [10].”
BackgroundMethodResultInfluentialDaDianNao: A Machine-Learning Supercomputer
Chen, Y., Luo, T., Liu, S., Zhang, S., He, L., Wang, J., Li, L., Chen, T., Xu, Z., Sun, N., et al. (2014) · 2014 47th Annual IEEE/ACM International Symposium on Microarchitecture · doi:10.1109/micro.2014.58 · Cited by 1,492 · cited 3×
“As a result, there has been significant interest in developing and adapting hardware accelerators for CNNs [7] such as GPUs [3], FPGAs [13], [19], [10], [1], and ASICs [2].”
BackgroundMethodTorch7: A Matlab-like Environment for Machine Learning
Collobert, R., Kavukcuoglu, K., & Farabet, C. (2011) · Infoscience (Ecole Polytechnique Fédérale de Lausanne) · Cited by 1,186
cuDNN: Efficient Primitives for Deep Learning
Chetlur, S., Woolley, C., Vandermersch, P., Cohen, J., Tran, J. J., Catanzaro, B., & Shelhamer, E. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1410.0759 · Cited by 943 · cited 2×
“As a result, there has been significant interest in developing and adapting hardware accelerators for CNNs [7] such as GPUs [3], FPGAs [13], [19], [10], [1], and ASICs [2].”
MethodA reconfigurable fabric for accelerating large-scale datacenter services
Putnam, A., Caulfield, A. M., Chung, E. S., Chiou, D., Constantinides, K., Demme, J., Esmaeilzadeh, H., Fowers, J., Gopal, G. P., Gray, J., et al. (2014) · 2014 ACM/IEEE 41st International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca.2014.6853195 · Cited by 576
NeuFlow: A runtime reconfigurable dataflow processor for vision
Farabet, C., Martini, B., Corda, B., Akselrod, P., Culurciello, E., & LeCun, Y. (2011) · CVPR 2011 WORKSHOPS · doi:10.1109/cvprw.2011.5981829 · Cited by 369 · cited 1×
“In [6], [5], [16], systolic implementations were proposed.”
Cited by
Efficient Processing of Deep Neural Networks: A Tutorial and Survey
Sze, V., Chen, Y.-H., Yang, T.-J., & Emer, J. S. (2017) · Proceedings of the IEEE · doi:10.1109/jproc.2017.2761740 · Cited by 4,215
Convergence of Edge Computing and Deep Learning: A Comprehensive Survey
Wang, X., Han, Y., Leung, V. C. M., Niyato, D., Yan, X., & Chen, X. (2020) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2020.2970550 · Cited by 1,498
“, CNN, a finer grid partitioning can be applied to minimize communication, synchronization, and memory overhead [130].”
BackgroundMethodSCNN: An Accelerator for Compressed-sparse Convolutional Neural Networks
Parashar, A., Rhu, M., Mukkara, A., Puglielli, A., Venkatesan, R., Khailany, B., Emer, J., Keckler, S. W., & Dally, W. J. (2017) · Proceedings of the 44th Annual International Symposium on Computer Architecture · doi:10.1145/3079856.3080254 · Cited by 896
“propose to fuse adjacent layers in a dense CNN accelerator so that intermediate activations between layers can be kept on chip [2].”
BackgroundTimeloop: A Systematic Approach to DNN Accelerator Evaluation
Parashar, A., Raina, P., Shao, Y. S., Chen, Y.-H., Ying, V. A., Mukkara, A., Venkatesan, R., Khailany, B., Keckler, S. W., & Emer, J. (2019) · 2019 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS) · doi:10.1109/ispass.2019.00042 · Cited by 571
“Future work includes modeling inter-layer relationships to find globally-optimal solutions for full networks (as in [2]).”
BackgroundBit Fusion: Bit-Level Dynamically Composable Architecture for Accelerating Deep Neural Network
Sharma, H., Park, J., Suda, N., Lai, L., Chau, B., Kim, J. K., Chandra, V., & Esmaeilzadeh, H. (2018) · 2018 ACM/IEEE 45th Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca.2018.00069 · Cited by 543
“Due to their insatiable demand for compute power, naturally, both the research community [1–28] as well the industry [29–31] have turned to accelerators to accommodate modern DNN computation.”
BackgroundA Configurable Cloud-Scale DNN Processor for Real-Time AI
Fowers, J., Ovtcharov, K., Papamichael, M., Massengill, T., Liu, M., Lo, D., Alkalay, S., Haselman, M., Adams, L., Ghandi, M., et al. (2018) · 2018 ACM/IEEE 45th Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca.2018.00012 · Cited by 520
“There has been a large body of research in DNN accelerators targeting FPGAs, in particular for accelerating Convolutional Neural Networks (CNNs), such as [42], [63]–[94].”
BackgroundDeepThings: Distributed Adaptive Deep Learning Inference on Resource-Constrained IoT Edge Clusters
Zhao, Z., Barijough, K. M., & Gerstlauer, A. (2018) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2018.2858384 · Cited by 502
“Layer fusion techniques have been previously applied to CNN accelerators [17].”
MethodA High-Throughput and Power-Efficient FPGA Implementation of YOLO CNN for Object Detection
Nguyen, D. T., Nguyen, T. N., Kim, H., & Lee, H.-J. (2019) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2019.2905242 · Cited by 435
“Moreover, designs in [6], [11], and [14] use doubled buffer to pipeline the computation.”
BackgroundInfluentialRecent advances in convolutional neural network acceleration
Zhang, Q., Zhang, M., Chen, T., Sun, Z., Ma, Y., & Yu, B. (2019) · Neurocomputing · doi:10.1016/j.neucom.2018.09.038 · Cited by 434
A survey of FPGA-based accelerators for convolutional neural networks
Mittal, S. (2020) · Neural Computing and Applications · doi:10.1007/s00521-018-3761-1 · Cited by 429
“To exploit this, they use circular line buffer for every layer, which achieves data reuse with lower complexity and less BRAM resources than the tile-based buffer used by Alwani et al. [78].”
BackgroundMethodPUMA: A Programmable Ultra-efficient Memristor-based Accelerator for Machine Learning Inference
Ankit, A., Hajj, I. E., Chalamalasetti, S. R., Ndu, G., Foltin, M., Williams, R. S., Faraboschi, P., Hwu, W.-M. W., Strachan, J. P., Roy, K., et al. (2019) · Proceedings of the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Systems · doi:10.1145/3297858.3304049 · Cited by 424
Hoefler, T., Alistarh, D., Ben-Nun, T., Dryden, N., & Peste, A. (2021) · ETH Zurich · doi:10.3929/ethz-b-000519202 · Cited by 423
SCNN: An Accelerator for Compressed-sparse Convolutional Neural Networks
Parashar, A., Rhu, M., Mukkara, A., Puglielli, A., Venkatesan, R., Khailany, B., Emer, J., Keckler, S. W., & Dally, W. J. (2017) · ACM SIGARCH Computer Architecture News · doi:10.1145/3140659.3080254 · Cited by 413
Simba: Scaling Deep-Learning Inference with Multi-Chip-Module-Based Architecture
Shao, Y. S., Clemons, J., Venkatesan, R., Zimmer, B., Fojtik, M., Jiang, N., Keller, B., Klinefelter, A., Pinckney, N., Raina, P., et al. (2019) · Proceedings of the 52nd Annual IEEE/ACM International Symposium on Microarchitecture · doi:10.1145/3352460.3358302 · Cited by 406
“Tile-based architectures have frequently been proposed for deeplearning accelerator designs [1, 2, 13, 15, 22, 25, 30, 42, 51, 53, 56, 58, 61, 75].”
BackgroundHarDNet: A Low Memory Traffic Network
Chao, P., Kao, C.-Y., Ruan, Y., Huang, C.-H., & Lin, Y.-L. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00365 · Cited by 324
“In contrast, fused-layer computation for multiple convolutional layers has been proposed [1], in which intermediate layers in a fused-layer group will not produce any memory traffic for feature maps.”
BackgroundTang, W., Yang, Q., Xiong, K., & Yan, W. (2020) · Solar Energy · doi:10.1016/j.solener.2020.03.049 · Cited by 296
[DL] A Survey of FPGA-based Neural Network Inference Accelerators
Guo, K., Zeng, S., Yu, J., Wang, Y., & Yang, H. (2019) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/3289185 · Cited by 292
“[5] address the external memory access problem by fusing two neighboring layers together to avoid the intermediate result transfer between the two layers.”
BackgroundMaximizing CNN Accelerator Efficiency Through Resource Partitioning
Shen, Y., Ferdman, M., & Milder, P. (2017) · Proceedings of the 44th Annual International Symposium on Computer Architecture · doi:10.1145/3079856.3080221 · Cited by 284
“[2] uses a fused-layer technique to reduce bandwidth use of convolutional layers.”
MethodSelf-citationSparTen: A Sparse Tensor Accelerator for Convolutional Neural Networks
Gondimalla, A., Chesnut, N., Thottethodi, M., & Vijaykumar, T. N. (2019) · Proceedings of the 52nd Annual IEEE/ACM International Symposium on Microarchitecture · doi:10.1145/3352460.3358291 · Cited by 278
“Architectures for dense CNNs have optimized compute [17, 23, 25, 28, 37], memory [7, 14, 30], and reuse [3, 8].”
BackgroundMAERI: Enabling Flexible Dataflow Mapping over DNN Accelerators via Reconfigurable Interconnects
Kwon, H., Samajdar, A., & Krishna, T. (2018) · Proceedings of the Twenty-Third International Conference on Architectural Support for Programming Languages and Operating Systems · doi:10.1145/3173162.3173176 · Cited by 270
“It also contains comparators for pooling operations.”
ZigZag: Enlarging Joint Architecture-Mapping Design Space Exploration for DNN Accelerators
Mei, L., Houshmand, P., Jain, V., Giraldo, J. S. P., & Verhelst, M. (2021) · IEEE Transactions on Computers · doi:10.1109/tc.2021.3059962 · Cited by 247
Yang, T.-H., Cheng, H.-Y., Yang, C.-L., Tseng, I.-C., Hu, H.-W., Chang, H.-S., & Li, H.-P. (2019) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1145/3307650.3322271 · Cited by 244
“Modern DNN models [26, 39, 41] are significantly larger than those used in the 1990s [10], requiring 138MB [41] for filter weight storage, 77MB [2, 41] memory accesses, and thousands to millions of computation operations per input activation.”
MethodBenchmarking and Analyzing Deep Neural Network Training
Zhu, H., Akrout, M., Zheng, B., Pelegris, A., Jayarajan, A., Phanishayee, A., Schroeder, B., & Pekhimenko, G. G. (2018) · IEEE International Symposium on Workload Characterization · doi:10.1109/iiswc.2018.8573476 · Cited by 236
Interstellar: Using Halide's Scheduling Language to Analyze DNN Accelerators
Yang, X., Gao, M., Liu, Q., Setter, J., Pu, J., Nayak, A., Bell, S., Cao, K., Ha, H., Raina, P., et al. (2020) · Proceedings of the Twenty-Fifth International Conference on Architectural Support for Programming Languages and Operating Systems · doi:10.1145/3373376.3378514 · Cited by 231
“As an example, C | K is a widely adopted data � ow (Figure 1, and [5, 36, 37, 40]) due to its � exibility to also map matrix multiplications in MLPs and LSTMs.”
MethodRecent advances in efficient computation of deep convolutional neural networks
Cheng, J., Wang, P.-S., Li, G., Hu, Q.-H., & Lu, H.-Q. (2018) · Frontiers of Information Technology & Electronic Engineering · doi:10.1631/fitee.1700789 · Cited by 230
“(2017) designed a flexible data buffing scheme to reduce bandwidth requirements, and Alwani et al. (2016) and Xiao et al.”
BackgroundMethodInfluentialMAERI: Enabling Flexible Dataflow Mapping over DNN Accelerators via Reconfigurable Interconnects
Kwon, H., Samajdar, A., & Krishna, T. (2018) · ACM SIGPLAN Notices · doi:10.1145/3296957.3173176 · Cited by 217
Full Stack Optimization of Transformer Inference: a SurveyPreprint
Kim, S., Hooper, C. R. C., Wattanawong, T., Kang, M., Yan, R., Genc, H., Dinh, G., Huang, Q., Keutzer, K., Mahoney, M. W., et al. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2302.14017 · Cited by 215
“A large number of operation fusion techniques [10, 247, 252] have been developed to optimize the mapping for more data reuse across DNN layers.”
MethodIntelligence Beyond the Edge: Inference on Intermittent Embedded Systems
Gobieski, G., Lucia, B., & Beckmann, N. (2019) · Proceedings of the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Systems · doi:10.1145/3297858.3304011 · Cited by 203
“of the neural network. SONIC optimizes DNN inference by ensuring that these loops execute correctly on intermittent power while adding much less overhead than prior task-based systems. 7 SONIC TAILS a[7] += b[7] !c a[4] += b[4] !c Tile -5 Tile -12 a[0] += b[0] !c a[9] += b[9] !c a[0] += b[0] !c ! ! a[5] += b[5] !c ! E…”
BackgroundExploring Heterogeneous Algorithms for Accelerating Deep Convolutional Neural Networks on FPGAs
Xiao, Q., Liang, Y., Lu, L., Yan, S., & Tai, Y.-W. (2017) · Proceedings of the 54th Annual Design Automation Conference 2017 · doi:10.1145/3061639.3062244 · Cited by 202
“7.3 Case Study of AlexNet AlexNet [13] is composed of five convolutional layers (integrated with ReLU), three pooling layers, two LRN layers and three final fully connected layers.”
MethodInfluentialDeep Learning Based Module Defect Analysis for Large-Scale Photovoltaic Farms
Li, X., Yang, Q., Lou, Z., & Yan, W. (2019) · IEEE Transactions on Energy Conversion · doi:10.1109/tec.2018.2873358 · Cited by 202
“In CNN [29], [30], the alternating convolution and sub-sampling operations are firstly carried out, and the finally a generic multi-layer network is adopted.”
MethodToolflows for Mapping Convolutional Neural Networks on FPGAs: A Survey and Future Directions
Venieris, S. I., Kouris, A., & Bouganis, C.-S. (2019) · ACM Computing Surveys · doi:10.1145/3186332 · Cited by 189
“and-tuned accelerators for particular CNN-FPGA pairs in RTL [18][16][49], HLS [6][44] and mixed RTL-HLS [101], together with designs that focus on optimising the external memory bandwidth utilisation [5][77]. A number of existing works lie close to the presented CNN-to-FPGA toolflows, but lack essential components tha…”
BackgroundTiny Machine Learning: Progress and Futures [Feature]
Lin, J., Zhu, L., Chen, W.-M., Wang, W.-C., & Han, S. (2023) · IEEE Circuits and Systems Magazine · doi:10.1109/mcas.2023.3302182 · Cited by 186
“With a similar goal of reducing peak memory, other researchers further propose computing partial spatial regions across multiple layers [103, 104, 105].”
Efficient Processing of Deep Neural Networks
Sze, V., Chen, Y., Yang, T.-J., & Emer, J. (2020) · Synthesis lectures on computer architecture · doi:10.2200/s01004ed1v01y202004cac050 · Cited by 183
MAGNet: A Modular Accelerator Generator for Neural Networks
Venkatesan, R., Shao, Y. S., Wang, M., Clemons, J., Dai, S. X., Fojtik, M. R., Keller, B., Klinefelter, A., Pinckney, N., Raina, P., et al. (2019) · doi:10.1109/iccad45719.2019.8942127 · Cited by 180
Efficient Hardware Architectures for Accelerating Deep Neural Networks: Survey
Pudi, D., Boppu, S., Manikandan, M. S., & Cenkeramaddi, L. R. (2022) · IEEE Access · doi:10.1109/access.2022.3229767 · Cited by 180
“Layer fusion allows the processing of multiple layers of CNN in the overlapped fashion [33], which was exploited by TCPA to save the intermediate memory needed between the layers.”
MethodTANGRAM: Optimized Coarse-Grained Dataflow for Scalable NN Accelerators
Gao, M., Yang, X., Pu, J., Horowitz, M., & Kozyrakis, C. (2019) · Proceedings of the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Systems · doi:10.1145/3297858.3304014 · Cited by 178
“Alternatively, we can use multiple engines to process multiple layers in a pipelined manner by spatially mapping the NN DAG structures [2, 25, 36, 38, 40, 44].”
BackgroundMethodAccelerating Convolutional Neural Network With FFT on Embedded Hardware
Abtahi, T., Shea, C., Kulkarni, A., & Mohsenin, T. (2018) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2018.2825145 · Cited by 171
“[30] proposed fused-layer CNN accelerators which focus on reducing data flow across layers and fusing multiple convolutional layer computation together and achieved 95% reduction in total data transfer.”
BackgroundEfficient Hardware Architectures for Deep Convolutional Neural Network
Wang, J., Lin, J. T., & Wang, Z. (2017) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2017.2767204 · Cited by 166
“Reference [25] proposed an efficient dataflow across convolutional layers, which employed a pyramid-shaped multi-layer”
MethodDeep Learning on Mobile and Embedded Devices: State-of-the-art, Challenges, and Future Directions
Chen, Y., Zheng, B., Zhang, Z., Wang, Q., Shen, C., & Zhang, Q. (2021) · ACM Computing Surveys · doi:10.1145/3398209 · Cited by 159
“Alwani et al. [4] presented a fused-layer CNN accelerator to reuse inter-layer intermediate data on chip by fusing the computation of multiple CNN layers through the pyramid-shaped multilayer sliding window.”
MethodProcessing-in-Memory for Energy-Efficient Neural Network Training: A Heterogeneous Approach
Liu, J., Zhao, H., Ogleari, M., Li, D., & Zhao, J. (2018) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro.2018.00059 · Cited by 156
Boroumand, A., Ghose, S., Akin, B., Narayanaswami, R., Junior, G. F. D. O., Ma, X., Shiu, E., & Mutlu, O. (2021) · International Conference on Parallel Architectures and Compilation Techniques (PACT) · doi:10.1109/pact52795.2021.00019 · Cited by 153
“Many prior works [1, 2, 7, 8, 9, 17, 20, 35, 36, 55, 68, 73, 82, 83, 85, 91, 92, 102] examine a specific model type (predominantly CNNs).”
BackgroundA Multi-Neural Network Acceleration Architecture
Baek, E., Kwon, D., & Kim, J. (2020) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca45697.2020.00081 · Cited by 148
“For example, weight-stationary [7], [13], [16], [33], [43], [46], [47], [51], [52], [59], output-stationary [22], [51], row-stationary [14], [15], [26] dataflows determine spatial and temporal data mappings to PEs differently.”
Heterogeneous Dataflow Accelerators for Multi-DNN Workloads
Kwon, H., Lai, L., Pellauer, M., Krishna, T., Chen, Y.-H., & Chandra, V. (2021) · 2021 IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca51647.2021.00016 · Cited by 144
“MAERI [18] is an RDA that efficiently supports irregular mappings resulting from sparsity, cross-layer mapping [33], and so on.”
BackgroundMethodLaconic deep learning inference acceleration
Sharify, S., Lascorz, A. D., Mahmoud, M., Nikolić, M., Siu, K., Stuart, D. M., Poulos, Z., & Moshovos, A. (2019) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1145/3307650.3322255 · Cited by 143
“It also includes techniques that fuse computation across multiple levels [5].”
BackgroundFLAT: An Optimized Dataflow for Mitigating Attention Bottlenecks
Kao, S.-C., Subramanian, S., Agrawal, G., Yazdanbakhsh, A., & Krishna, T. (2023) · International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) · doi:10.1145/3575693.3575747 · Cited by 143
“Arbitrary inter-loop tiling as employed by prior CONV/FC fusion techniques [8, 91] violates this data dependency constraint.”
BackgroundMethodInfluentialCentaur: A Chiplet-based, Hybrid Sparse-Dense Accelerator for Personalized Recommendations
Hwang, R., Kim, T., Kwon, Y., & Rhu, M. (2020) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca45697.2020.00083 · Cited by 142
“potential for flexible acceleration of the constantly evolving ML applications [4], [20], [44], [45], [48], [59], [64]–[66].”
BackgroundDave, S., Baghdadi, R., Nowatzki, T., Avancha, S., Shrivastava, A., & Li, B. (2021) · Proceedings of the IEEE · doi:10.1109/jproc.2021.3098483 · Cited by 140
“3) Fused-layer execution: Fused-layer CNNs [173] lever-aged cross-layer reuse by processing a small tile of activations such that outputs for few layers can be computed alongside while retaining the corresponding data in the on-chip memory.”
BackgroundHyPar: Towards Hybrid Parallelism for Deep Learning Accelerator Array
Song, L., Mao, J., Zhuo, Y., Qian, X., Li, H. H., & Chen, Y. (2019) · IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca.2019.00027 · Cited by 139
Koppula, S., Orosa, L., Yağlıkçı, A. G., Azizi, R., Shahroodi, T., Kanellopoulos, K., & Mutlu, O. (2019) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1145/3352460.3358280 · Cited by 137
Chang, J.-W., Kang, K.-W., & Kang, S. (2018) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2018.2888898 · Cited by 136
Adaptive parallel execution of deep neural networks on heterogeneous edge devices
Zhou, L., Samavatian, M. H., Bacha, A., Majumdar, S., & Teodorescu, R. (2019) · ACM/IEEE Symposium on Edge Computing (SEC) · doi:10.1145/3318216.3363312 · Cited by 134
“The concept of layer fusion was first proposed in [2] as a method to reduce off-chip data movement in a CNN accelerator.”
MethodMCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep LearningPreprint
Lin, J., Chen, W.-M., Cai, H., Gan, C., & Han, S. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.15352 · Cited by 132
“Computing partial spatial regions across multiple layers can reduce the peak memory [15, 3, 40].”
BackgroundGong, L., Wang, C., Li, X., Chen, H., & Zhou, X. (2018) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2018.2857078 · Cited by 131
“…find the optimal tiling mode directly for each PE, and another one is layer-fusion-based strategy which is introduced from the fused-layer technique [28], and its main purpose is to decrease the data transmission amount between successive CONV layers and on this basis, finding the optimal tiling…”
BackgroundSelf-citationZhang, J. J., Rangineni, K., Ghodsi, Z., & Garg, S. (2018) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1145/3195970.3196129 · Cited by 127
“As in [2, 23, 24], Thundervolt is only used to accelerate ImageNet’s convolutional layers, which make up more than 90% of its computational cost.”
MethodPointAcc: Efficient Point Cloud Accelerator
Lin, Y., Zhang, Z., Tang, H., Wang, H., & Han, S. (2021) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1145/3466752.3480084 · Cited by 127
“The conventional layer fusion [2] spatially pipelines the computation, and thus fixes the number of fused layers and requires matching throughput in between.”
BackgroundREQ-YOLO: A Resource-Aware, Efficient Quantization Framework for Object Detection on FPGAs
Ding, C., Wang, S., Liu, N., Xu, K., Wang, Y., & Liang, Y. E. (2019) · ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (FPGA) · doi:10.1145/3289602.3293904 · Cited by 126
“The second one is efficient hardware implementations, including FPGAs and ASICs [1, 7, 31, 32, 36, 49, 50, 53, 54].”
MethodSparsity-Aware and Re-configurable NPU Architecture for Samsung Flagship Mobile SoC
Jang, J.-W., Lee, S., Kim, D., Park, H., Ardestani, A. S., Choi, Y., Kim, C., Kim, Y., Yu, H., Abdel-Aziz, H. A., et al. (2021) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca52012.2021.00011 · Cited by 126
“In addition, we applied specific compiler optimization techniques for our architecture, such as determining the configurations of the re-configurable MAC arrays (Section V-B) and decomposing stages for mixed-precision support (Section V-D) as well as common optimization techniques including tiling and layer fusion [38…”
MethodMaximizing CNN Accelerator Efficiency Through Resource Partitioning
Shen, Y., Ferdman, M., & Milder, P. A. (2017) · ACM SIGARCH Computer Architecture News · doi:10.1145/3140659.3080221 · Cited by 124
Self-citationPermDNN: Efficient Compressed DNN Architecture with Permuted Diagonal Matrices
Deng, C., Liao, S., Xie, Y., Parhi, K. K., Qian, X., & Yuan, B. (2018) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro.2018.00024 · Cited by 120
“Among the research effort on data flow in DNN, [47]– [49] propose optimized data flow to provide high flexibility”
BackgroundComputation Reuse in DNNs by Exploiting Input Similarity
Riera, M., Arnau, J.-M., & González, A. M. (2018) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca.2018.00016 · Cited by 118
“Some of the most popular applications of MLPs are speech recognition and machine translation [12], [33], whereas CNNs are commonly used for object recognition, image classification [3] and, more recently, video classification and labeling [34], [35].”
BackgroundDesign Flow of Accelerating Hybrid Extremely Low Bit-Width Neural Network in Embedded FPGA
Wang, J., Lou, Q., Zhang, X., Zhu, C., Lin, Y., & Chen, D. (2018) · International Conference on Field Programmable Logic and Applications (FPL) · doi:10.1109/fpl.2018.00035 · Cited by 118
EVA²: Exploiting Temporal Redundancy in Live Computer Vision
Buckler, M., Bedoukian, P., Jayasuriya, S., & Sampson, A. (2018) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca.2018.00051 · Cited by 116
“In research, many recent ASICs target convolutional and fullyconnected layers [43, 17, 1, 44, 2, 45, 46, 39], as do some FPGA designs [47, 48, 49, 50].”
BackgroundSmartShuttle: Optimizing off-chip memory accesses for deep learning accelerators
Li, J., Yan, G., Lü, W., Jiang, S., Gong, S., Wu, J., & Li, X. (2018) · Design, Automation, and Test in Europe · doi:10.23919/date.2018.8342033 · Cited by 115
“Previous work simply maximize the reuse of one single data type for all convolutional layers, e.g. Zhang et al. [3] for ofm , Alwani et al. [10] for wght .”
MethodEscher: A CNN Accelerator with Flexible Buffering to Minimize Off-Chip Transfer
Shen, Y., Ferdman, M., & Milder, P. A. (2017) · IEEE International Symposium on Field-Programmable Custom Computing Machines (FCCM) · doi:10.1109/fccm.2017.47 · Cited by 113
“2, because the early convolutional layers have very few weights [19], thus reusing them makes little difference.”
BackgroundSelf-citationA pseudo-softmax function for hardware-based high speed image classification
Cardarilli, G. C., Nunzio, L. D., Fazzolari, R., Giardino, D., Nannarelli, A., Re, M., & Spanò, S. (2021) · Scientific Reports · doi:10.1038/s41598-021-94691-7 · Cited by 108
“The results are given as the global accuracy of the overall Machine Learning system, not focusing on the softmax computation.”
MethodInfluentialReview of ASIC accelerators for deep neural network
Machupalli, R., Hossain, M., & Mandal, M. K. (2022) · Microprocessors and Microsystems · doi:10.1016/j.micpro.2022.104441 · Cited by 108
Accelerating CNN inference on FPGAs: A Survey
Abdelouahab, K., Pelcat, M., Serot, J., & Berry, F. (2018) · arXiv · doi:10.48550/arxiv.1806.01683 · Cited by 107
“In FPGAs, implementations such [39, 79, 77] employ this data representation.”
Kim, Y., Choi, J.-S., & Kim, M. (2018) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2018.2864321 · Cited by 107
“Many hardware architectures have been proposed to support CNN [26]–[31], [60]–[62] in various areas of research.”
BackgroundAccelerating CNN Inference on ASICs: A Survey
Moolchandani, D., Kumar, A., & Sarangi, S. R. (2020) · Journal of Systems Architecture · doi:10.1016/j.sysarc.2020.101887 · Cited by 107
“If multiple layers are executed on a PE array concurrently as in [10, 103], a synchronization mechanism among the PEs is necessary to support the pipelining of the layers.”
BackgroundTowards a Uniform Template-based Architecture for Accelerating 2D and 3D CNNs on FPGA
Shen, J., Huang, Y. R., Wang, Z., Qiao, Y., Wen, M., & Zhang, C. (2018) · ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (FPGA) · doi:10.1145/3174243.3174257 · Cited by 106
“The works in [1, 11] take opposite approaches by making all layers working concurrently in a pipelined structure.”
BackgroundMethodNguyen, D. T., Kim, H., & Lee, H. (2020) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2020.3020569 · Cited by 104
“Similar to the fused layer design [31], the functionality of pipelined layers is to reduce off-chip access for feature-maps completely.”
BackgroundMethodResultInter-layer Scheduling Space Definition and Exploration for Tiled Accelerators
Cai, J.-W., Wei, Y.-C., Wu, Z., Peng, S., & Ma, K.-S. (2023) · International Symposium on Computer Architecture · doi:10.1145/3579371.3589048 · Cited by 104
“Although inter-layer scheduling plays an increasingly important role in keeping tiled accelerators highly utilized and energy-efficient,thereisasignificantdeficiency initsresearch: Mostworks[3, 6, 17, 38, 39, 55, 62] continue to optimize existing heuristic patterns, LP and LS, but do not propose new patterns, let alon…”
BackgroundInfluentialGemini: Mapping and Architecture Co-exploration for Large-scale DNN Chiplet Accelerators
Cai, J.-W., Wu, Z., Peng, S., Wei, Y.-C., Tan, Z.-H., Shi, G., Gao, M., & Ma, K.-S. (2023) · International Symposium on High-Performance Computer Architecture · doi:10.1109/hpca57654.2024.00022 · Cited by 91
“However, with the increase in accelerator scale, LS mapping demonstrates limited scalability [3], [15], [46], [66], whereas LP mapping exhibits a higher potential.”
BackgroundMethodLi, Y., Park, J., Alian, M., Yuan, Y., Qu, Z., Pan, P., Wang, R., Schwing, A. G., Esmaeilzadeh, H., & Kim, N. S. (2018) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro.2018.00023 · Cited by 102
TGPA: tile-grained pipeline architecture for low latency CNN inference
Wei, X., Liang, Y. E., Li, X., Yu, C. H., Zhang, P., & Cong, J. (2018) · doi:10.1145/3240765.3240856 · Cited by 100
Melon: breaking the memory wall for resource-efficient on-device machine learning
Wang, Q., Xu, M., Jin, C., Dong, X., Yuan, J., Jin, X., Huang, G., Liu, Y., & Liu, X. (2022) · International Conference on Mobile Systems, Applications, and Services · doi:10.1145/3498361.3538928 · Cited by 100
“A fewmemory saving techniques [7] are also proposed for hardware accelerators, but they are not compatible with commodity mobile devices.”
BackgroundPrediction Based Execution on Deep Neural Networks
Song, M., Zhao, J., Hu, Y., Zhang, J., & Li, T. (2018) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca.2018.00068 · Cited by 100
The Case for FPGA-Based Edge Computing
Xu, C., Jiang, S., Luo, G., Sun, G., An, N., Huang, G., & Liu, X. (2020) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2020.3041781 · Cited by 97
“As a generic example for data modeling and representation, FPGA-based accelerators for convolutional neural networks (CNN) can achieve 2.0x (cid:1) 2.5x speedup on AlexNet for image classifi-cation over CPUs [7], [8], [9].”
Zhang, J. J., Rangineni, K., Ghodsi, Z., & Garg, S. (2018) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1109/dac.2018.8465918 · Cited by 97
Accelerating Convolutional Neural Networks via Activation Map Compression
Georgiadis, G. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.00725 · Cited by 96
“them, which can reduce the number of MAC’s performed. However, their method also modifies the network structure and in addition, it does not increase the sparsity of the activation maps. Alwani et al. [3] reduce the memory requirement of the network by recomputing activation maps instead of storing them, which obviousl…”
BackgroundMobile Foundation Model as Firmware
Yuan, J., Yang, C., Cai, D., Wang, S., Yuan, X., Zhang, Z., Li, X., Zhang, D., Mei, H., Jia, X., et al. (2024) · ACM Annual International Conference on Mobile Computing and Networking (MobiCom) · doi:10.1145/3636534.3649361 · Cited by 96
“This fragmentation, which often results in ad-hoc optimization efforts [40, 99, 107], seems unavoidable.”
An Overview of Efficient Interconnection Networks for Deep Neural Network Accelerators
Nabavinejad, S. M., Baharloo, M. M., Chen, K.-C. J., Palesi, M., Kogel, T., & Ebrahimi, M. (2020) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2020.3022920 · Cited by 95
“MAERI can also support approaches such as Fused CNN [70] that aim to improve the power efficiency of accelerators.”
MethodDNNVM: End-to-End Compiler Leveraging Heterogeneous Optimizations on FPGA-Based CNN Accelerators
Xing, Y., Liang, S., Sui, L., Jia, X., Qiu, J., Liu, X., Wang, Y., Shan, Y., & Wang, Y. (2019) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2019.2930577 · Cited by 93
“Fuse1 [33] focuses on optimizing the external memory bandwidth utilization.”
BackgroundDeepSZ: A Novel Framework to Compress Deep Neural Networks by Using Error-Bounded Lossy Compression
Jin, S., Di, S., Liang, X., Tian, J., Tao, D., & Cappello, F. (2019) · ACM International Symposium on High-Performance Parallel and Distributed Computing (HPDC) · doi:10.1145/3307681.3326608 · Cited by 91
“performed for testing. In the following discussion, test refers to the forward pass process on the test dataset for generating the inference accuracy of the neural network. According to prior studies [4], although convolutional layers occupy most of the computation time (∼95%) because of the expensive convolution oper…”
MethodAccPar: Tensor Partitioning for Heterogeneous Deep Learning Accelerators
Song, L., Chen, F., Zhuo, Y., Qian, X., Li, H. H., & Chen, Y. (2020) · IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca47549.2020.00036 · Cited by 91
“Neuflow [51], Eyeriss [52–54], Flexflow [55], Fused-CNN [56], CNN-Paritition [57], GANAX [58], UCNN [59], TANGRAM [131], Sparse-Systolic [132], MAERI [133]”
E-RNN: Design Optimization for Efficient Recurrent Neural Networks in FPGAs
Li, Z., Ding, C., Wang, S., Wen, W., Zhuo, Y., Liu, C., Qiu, Q., Xu, W., Lin, X., Qian, X., et al. (2019) · IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca.2019.00028 · Cited by 91
TIE: energy-efficient tensor train-based inference engine for deep neural network
Deng, C., Sun, F., Qian, X., Lin, J. T., Wang, Z., & Yuan, B. (2019) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1145/3307650.3322258 · Cited by 89
A Survey and Taxonomy of FPGA-based Deep Learning Accelerators
Blaiech, A. G., Ben Khalifa, K., Valderrama, C., Fernandes, M. A. C., & Bedoui, M. H. (2019) · Journal of Systems Architecture · doi:10.1016/j.sysarc.2019.01.007 · Cited by 88
SMAUG: End-to-End Full-Stack Simulation Infrastructure for Deep Learning Workloads
Xi, S. L., Yao, Y., Bhardwaj, K., Whatmough, P. N., Wei, G.-Y., & Brooks, D. M. (2020) · ACM Transactions on Architecture and Code Optimization · doi:10.1145/3424669 · Cited by 87
“Broadly speaking, the architecture community has focused on designing efficient dataflows to maximize local reuse of data and functional unit utilization [4, 10, 11, 15, 28, 34, 37, 39], explore the space of possible dataflows and mappings [26, 45, 74], exploit model sparsity and data quantization [17, 21, 29, 38, 46,…”
BackgroundAccelerating Deep Learning Inference via Model Parallelism and Partial Computation Offloading
Zhou, H., Li, M., Wang, N., Min, G., & Wu, J. (2022) · IEEE Transactions on Parallel and Distributed Systems · doi:10.1109/tpds.2022.3222509 · Cited by 85
“In this paper, we first leverage the Fused-Layer (FL) technique to convert a single sequence of DNN layers into multiple sequences, called paths, where each FL path consists of a sequence of small layers without modifying the inference results [26].”
MethodStride 2 1-D, 2-D, and 3-D Winograd for Convolutional Neural Networks
Yépez, J., & Ko, S. (2020) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2019.2961602 · Cited by 85
“In [20], an instruction-driven CNN accelerator is proposed which supports the Wino-grad algorithm and cross-layer scheduling; their accelerator achieves 7 × speedup compared to another cross-layer accelerator [21] on the same platform.”
MethodA full-stack search technique for domain optimized deep learning accelerators
Zhang, D., Huda, S., Songhori, E. M., Prabhu, K., Le, Q. V., Goldie, A., & Mirhoseini, A. (2021) · International Conference on Architectural Support for Programming Languages and Operating Systems · doi:10.1145/3503222.3507767 · Cited by 68
“There is also a growing body of work on more elaborate fusion [7, 11, 43, 60, 80, 111, 113].”
BackgroundAn Efficient and Flexible Accelerator Design for Sparse Convolutional Neural Networks
Xie, X., Lin, J. T., Wang, Z., & Wei, J. (2021) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2021.3074300 · Cited by 81
“To implement fusion dataflow on resource-limited hardware platforms, accelerators are always designed to be tile-based [4], [14].”
BackgroundMethodInfluentialDECORE: Deep Compression with Reinforcement Learning
Alwani, M., Wang, Y., & Madhavan, V. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01203 · Cited by 80
“For example, this approach could be used to find important input channels for network training [55] or removing channels only from early layers which have a high memory consumption and DRAM access time [2].”
MethodSelf-citationDIANA: An End-to-End Hybrid DIgital and ANAlog Neural Network SoC for the Edge
Houshmand, P., Sarda, G. M., Jain, V., Ueyoshi, K., Papistas, I. A., Shi, M., Zheng, Q., Bhattacharjee, D., Mallik, A., Debacker, P., et al. (2022) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2022.3214064 · Cited by 80
“Parallelization can be coupled with layer fusion [33]: instead of waiting until one layer has completely finished execution to start the next layer (on the same core, or the other core), it is possible to start execution of a layer as soon as a part of a layer has been processed.”
BackgroundProcrustes: a Dataflow and Accelerator for Sparse Deep Neural Network Training
Yang, D., Ghasemazar, A., Ren, X., Golub, M., Lemieux, G. G. F., & Lis, M. (2020) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro50266.2020.00064 · Cited by 79
Low Latency Deep Learning Inference Model for Distributed Intelligent IoT Edge Clusters
Naveen, S., Kounte, M. R., & Ahmed, M. R. (2021) · IEEE Access · doi:10.1109/access.2021.3131396 · Cited by 77
“In this work, we employ fused layer partitioning [63] for CNN inference models for multiple resource constraints edge devices.”
MethodMo, H., Zhu, W., Hu, W., Wang, G., Li, Q., Li, A., Yin, S., Wei, S., & Liu, L. (2021) · Digest of technical papers/Digest of technical papers - IEEE International Solid-State Circuits Conference · doi:10.1109/isscc42613.2021.9365943 · Cited by 76
“Without a fixed pyramid structure in RP mode, 38.6% and 23.2% less on-chip memory is required for feature maps to implement a canonical residual block compared with Fused [10] and Multi-Face [11], respectively.”
MethodInfluentialTender: Accelerating Large Language Models via Tensor Decomposition and Runtime Requantization
Lee, J., Lee, W., & Sim, J. (2024) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca59077.2024.00080 · Cited by 76
“Domain-specific accelerators for DNNs have been extensively studied over the past decade [3], [7], [8], [14], [15], [23], [24], [28], [39], [45], [47].”
DeeperThings: Fully Distributed CNN Inference on Resource-Constrained Edge Devices
Stahl, R., Hoffman, A., Mueller-Gritschneder, D., Gerstlauer, A., & Schlichtmann, U. (2021) · International Journal of Parallel Programming · doi:10.1007/s10766-021-00712-3 · Cited by 74
“With model distribution being an actively researched field, [1] presented work applicable to the distribution of CNN models targeting hardware accelerators.”
Method
Metadata
- Authors
- Alwani, M., Chen, H., Ferdman, M., & Milder, P.
- Affiliations
- Stony Brook University (US)
- Venue
- 2016 49th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO)
- Publisher
- IEEE
- Published
- 2016-10
- Pages
- 1-12
- ISSN
- 1072-4451 (ISSN-L)
- OpenAlex
- W4249932213
- Cited by
- 662
- References
- 20
- Citations by year
- 2026: 20 · 2025: 51 · 2024: 72 · 2023: 67 · 2022: 87 · 2021: 103 · 2020: 72 · 2019: 90 · 2018: 76 · 2017: 23 · 2016: 1
- FWCI
- 21.98
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- IEEE/ACM International Symposium on Microarchitecture (MICRO) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Hardware and Architecture › Embedded Systems Design Techniques
- Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
- Keywords
- CNN accelerator, layer fusion, data flow design, FPGA implementation, convolutional neural network inference
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.