In-Datacenter Performance Analysis of a Tensor Processing Unit
The DOI matches its Crossref record; 52 of the 54 references it cites were found in a registry or an index.
Abstract
Many architects believe that major improvements in cost-energy-performance must now come from domain-specific hardware. This paper evaluates a custom ASIC---called a Tensor Processing Unit (TPU) --- deployed in datacenters since 2015 that accelerates the inference phase of neural networks (NN). The heart of the TPU is a 65,536 8-bit MAC matrix multiply unit that offers a peak throughput of 92 TeraOps/second (TOPS) and a large (28 MiB) software-managed on-chip memory. The TPU's deterministic execution model is a better match to the 99th-percentile response-time requirement of our NN applications than are the time-varying optimizations of CPUs and GPUs that help average throughput more than guaranteed latency. The lack of such features helps explain why, despite having myriad MACs and a big memory, the TPU is relatively small and low power. We compare the TPU to a server-class Intel Haswell CPU and an Nvidia K80 GPU, which are contemporaries deployed in the same datacenters. Our workload, written in the high-level TensorFlow framework, uses production NN applications (MLPs, CNNs, and LSTMs) that represent 95% of our datacenters' NN inference demand. Despite low utilization for some applications, the TPU is on average about 15X -- 30X faster than its contemporary GPU or CPU, with TOPS/Watt about 30X -- 80X higher. Moreover, using the CPU's GDDR5 memory in the TPU would triple achieved TOPS and raise TOPS/Watt to nearly 70X the GPU and 200X the CPU.
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
“In the past few years, some companies and universities have presented application specific integrated circuits (ASICs) based on the complementary metal oxide semiconductor (CMOS) technology that are capable to compute and store information in the same unit.”
From Hardware implementation of memristor-based artificial neural networks (2024) · cites this paper as [19]
In the citing paper
“Additional optimization techniques, such as model compression [165] and hardware acceleration using graphics processing units (GPUs) or tensor processing units (TPUs) [166], further enhance efficiency.”
From Multimodal Large Language Models in Health Care: Applications, Challenges, and Future Outlook. (2024) · cites this paper as [166]
In the citing paper
“Tensor Processing Units are a class of proprietary AI accelerator hardware developed by Google, and used in their internal computing projects and employed in Google Cloud datacenters [17].”
From The rising costs of training frontier AI models (2024) · cites this paper as [17]
In the citing paper
“Resource-efficient computing devices that require less natural resources (energy, water, raw materials) to build and use can be seen as an obvious gain in terms of environmental impacts, and different hardware platforms are being developed towards this end as an alternative to GPUs and CPUs, which represent the current status quo of AI computing hardware [61, 62].”
From From Efficiency Gains to Rebound Effects: The Problem of Jevons' Paradox in AI's Polarized Environmental Debate (2025) · cites this paper as [62]
Citations
Cites
ImageNet 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
Going 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
Mastering the game of Go with deep neural networks and tree search
Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016) · Nature · doi:10.1038/nature16961 · Cited by 15,612
Identity mappings in deep residual networks
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46493-0_38 · Cited by 10,741
Identity Mappings in Deep Residual Networks
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · arXiv · doi:10.48550/arxiv.1603.05027 · Cited by 10,665
TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G. S., Davis, A., Dean, J., Devin, M., et al. (2016) · arXiv · doi:10.48550/arxiv.1603.04467 · Cited by 9,408
Self-citationComputer Architecture: A Quantitative Approach
Hennessy, J. L., & Patterson, D. A. (1989) · UCLA - Biblioteca de Ciencias y Tecnologia Felix Morales Bueno · Cited by 9,375
Self-citationGoogle's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
Wu, Y., Schuster, M., Chen, Z., Le, Q. V., Norouzi, M., Macherey, W., Krikun, M., Cao, Y., Gao, Q., Macherey, K., et al. (2016) · arXiv · doi:10.48550/arxiv.1609.08144 · Cited by 5,319
Self-citationWilliams, S., Waterman, A., & Patterson, D. A. (2009) · Communications of the ACM · doi:10.1145/1498765.1498785 · Cited by 2,798
Self-citation
Research works printed but not linked: 10
- Asanović, K. 1998. Asanović, K., Beck, Johnson, J., Wawrzynek, J., Kingsbury, B. and Morgan, N., November 1998. Training Neural Networks with Spert-II. Chapter 11 in Parallel Architectures for Artificial Networks: Paradigms and Implementations, N. Sundararajan and P. Saratchandran (Eds.), IEEE Computer Society Press, ISBN 0-8186-8399-6. https://people.eecs.berkeley.edu/~krste/papers/annbook.pdf
- Han S.; Pool J.; Tran J.; and Dally W. 2015. Learning both weights and connections for efficient neural networks. In Advances in Neural Information Processing Systems. · doi:10.5555/2969239.2969366
- Ross, J., Phelps, A., 2015. Computing Convolutions Using a Neural Network Processor, Patent Application No. 62/164,902.
- Ross, J., 2015. Prefetching Weights for a Neural Network Processor, Patent Application No. 62/164,981.
- Ross, J., Thorson, G., 2015. Rotating Data for Neural Network Computations, Patent Application No. 62/164,908.
- Schurman, E. and Brutlag, J., 2009, June. The user and business impact of server delays, additional bytes, and HTTP chunking in web search. In Velocity Web Performance and Operations Conference.
- 10.5555/800048.801719 · doi:10.5555/800048.801719
- Steinberg, D., 2015. Full-Chip Simulations, Keys to Success. Proc. Synopsys Users Group (SNUG) Silicon Valley 2015.
- Thorson, G., Clark, C., Luu, D., 2015. Vector Computation Unit in a Neural Network Processor, Patent Application No. 62/165,022.
- Young, C., 2015. Batch Processing in a Neural Network Processor, Patent Application No. 62/165,020.
Cited by
Items 201–300 of 1,031
A system hierarchy for brain-inspired computing
Zhang, Y., Qu, P., Ji, Y., Zhang, W., Gao, G., Wang, G., Song, S., Li, G., Chen, W., Zheng, W., et al. (2020) · Nature · doi:10.1038/s41586-020-2782-y · Cited by 208
“…systems 5–14 , approximation is a way to achieve low power and high performance, implemented using either low-precision digital calculations 5,6,8–10, 14 or analogue Neuromorphic computing is also distinct from traditional computing 2 in that it uses colocated computing and storage, uses…”
BackgroundNewton: A DRAM-maker’s Accelerator-in-Memory (AiM) Architecture for Machine Learning
He, M., Song, C., Kim, I., Jeong, C., Kim, S., Park, I., Thottethodi, M., & Vijaykumar, T. N. (2020) · 2020 53rd Annual IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro50266.2020.00040 · Cited by 208
“However, supporting small-batch inference is important (say 8-way batches), as also argued in [20], especially for acceleration at the edge (Section I).”
BackgroundKwon, Y.-C., Lee, S. H., Lee, J., Kwon, S.-H., Ryu, J. M., Son, J.-P., Seongil, O., Yu, H.-S., Lee, H., Kim, S. Y., et al. (2021) · 2021 IEEE International Solid-State Circuits Conference (ISSCC) · doi:10.1109/isscc42613.2021.9365862 · Cited by 204
“However, these accelerators can only operate at their peak performance when they get the necessary data from memory as quickly as it is processed: requiring off-chip memory with a high bandwidth and a large capacity [1].”
BackgroundAnalyzing and Increasing the Reliability of Convolutional Neural Networks on GPUs
Santos, F. F. D., Pimenta, P. F., Lunardi, C., Draghetti, L., Carro, L., Kaeli, D., & Rech, P. (2019) · IEEE Transactions on Reliability · doi:10.1109/tr.2018.2878387 · Cited by 226
“As GPUs are highly efficient for accelerating matrix multiplication, they are a great target for CNNs. Specific processors can be designed for DNN, achieving good performance and low power consumption [42], [43].”
BackgroundNovel frontier of photonics for data processing—Photonic accelerator
Kitayama, K.-I., Notomi, M., Naruse, M., Inoue, K., Kawakami, S., & Uchida, A. (2019) · APL Photonics · doi:10.1063/1.5108912 · Cited by 216
Enhancing Quality for HEVC Compressed Videos
Yang, R., Xu, M., Liu, T., Wang, Z., & Guan, Z. (2019) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2018.2867568 · Cited by 214
“cceleration hardwares to speed up the CNN networks. For example, the Cambricon-X accelerator has more than 7 times speed of GPU with much lower energy [41]. Recently, the Tensor Processing Unit (TPU) [42] developed by Google achieves 15 to 30 times speed compared with GPU, with smaller size and lower power consumption…”
MethodA Survey on Deep Learning Hardware Accelerators for Heterogeneous HPC Platforms
Silvano, C., Ielmini, D., Ferrandi, F., Fiorin, L., Curzel, S., Benini, L., Conti, F., Garofalo, A., Zambelli, C., Calore, E., et al. (2025) · ACM Computing Surveys · doi:10.1145/3729215 · Cited by 262
“The architecture of the TPU presented in [96, 98] is centered on a large (256 × 256) systolic array operating on 8-bit integers and targeting exclusively data center inference applications; this is coupled with a large amount of on-chip SRAM for activations (24 MiB) and a high-bandwidth (30 GiB/s)…”
A Retrospective and Prospective View of Approximate Computing [Point of View}
Liu, W., Lombardi, F., & Shulte, M. (2020) · Proceedings of the IEEE · doi:10.1109/jproc.2020.2975695 · Cited by 210
“For example, the tensor processing units (TPUs) by Google use reduced precision, a common approximate computing technique to reduce its power consumption [1].”
MethodPerspective: A review on memristive hardware for neuromorphic computation
Sung, C., Hwang, H., & Yoo, I. K. (2018) · Journal of Applied Physics · doi:10.1063/1.5037835 · Cited by 217
Amirsoleimani, A., Alibart, F., Yon, V., Xu, J., Pazhouhandeh, M. R., Ecoffey, S., Beilliard, Y., Genov, R., & Drouin, D. (2020) · Advanced Intelligent Systems · doi:10.1002/aisy.202000115 · Cited by 216
A Survey on Efficient Convolutional Neural Networks and Hardware Acceleration
Ghimire, D., Kil, D., & Kim, S.-H. (2022) · Electronics · doi:10.3390/electronics11060945 · Cited by 198
Homecare robotic systems for healthcare 4.0: visions and enabling technologies
Yang, G., Pang, Z., Jamal Deen, M., Dong, M., Zhang, Y.-T., Lovell, N., & Rahmani, A. M. (2020) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2020.2990529 · Cited by 213
“Two solutions are adopted for scheduling hardware resources, including hardware virtualization and small size hardware [161], [162].”
BackgroundYin, S., Ouyang, P., Tang, S., Tu, F., Li, X., Zheng, S., Lu, T., Gu, J., Liu, L., & Wei, S. (2018) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2017.2778281 · Cited by 212
“performance and energy efficiency, several NN processors are proposed [12]–[19].”
BackgroundDeep learning with coherent VCSEL neural networks
Chen, Z., Sludds, A., Davis, R., Christen, I., Bernstein, L., Ateshian, L., Heuser, T., Heermeier, N., Lott, J. A., Reitzenstein, S., et al. (2023) · Nature Photonics · doi:10.1038/s41566-023-01233-w · Cited by 191
A Machine Learning-Oriented Survey on Tiny Machine Learning
Capogrosso, L., Cunico, F., Cheng, D. S., Fummi, F., & Cristani, M. (2024) · IEEE Access · doi:10.1109/access.2024.3365349 · Cited by 248
“A specialized processor, known as the TPU, has been developed by Google explicitly for ML tasks, with a specific emphasis on tensor operations [160].”
MethodDo Switches Dream of Machine Learning?: Toward In-Network Classification
Xiong, Z., & Zilberman, N. (2019) · Proceedings of the 18th ACM Workshop on Hot Topics in Networks · doi:10.1145/3365609.3365864 · Cited by 198
“These challenges have driven innovation in hardware design for ML, including CPU optimization (e.g., [6, 53]), GPU (e.g., [14]) and FPGA (e.g., [18, 21, 51]) solutions, and specialized processing ASIC [12, 26].”
BackgroundInfluentialTensor Methods in Computer Vision and Deep Learning
Panagakis, Y., Kossaifi, J., Chrysos, G. G., Oldfield, J., Nicolaou, M. A., Anandkumar, A., & Zafeiriou, S. (2021) · Proceedings of the IEEE · doi:10.1109/jproc.2021.3074329 · Cited by 180
“Examples include application-specific integrated circuits (ASICs) designed specifically for running a specialized set of instructions (e.g., TPUs [54]), or field-programmable gate arrays (FPGAs) that conversely can be dynamically programmed depending on the application.”
BackgroundFPGA HLS Today: Successes, Challenges, and Opportunities
Cong, J., Lau, J., Liu, G., Neuendorffer, S., Pan, P., Vissers, K., & Zhang, Z. (2022) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/3530775 · Cited by 187
“Such optimizations are required on any computing platform, including multi-core CPUs, many-core GPUs, and TPUs, and they are not FPGA-specific.”
MethodHooker, S. (2021) · Communications of the ACM · doi:10.1145/3467017 · Cited by 211
“The first examples of domain specialized hardware released over the last few years – TPUs (Jouppi et al., 2017), edge-TPUs (Gupta & Tan, 2019), Arm Cortex-M55 (ARM, 2020), Facebook’s big sur (Lee & Wang, 2018) – optimize explicitly for costly operations common to deep neural networks like matrix multiplies.”
BackgroundInfluentialTianjic: A Unified and Scalable Chip Bridging Spike-Based and Continuous Neural Computation
Deng, L., Wang, G., Li, G., Li, S., Liang, L., Zhu, M., Wu, Y., Yang, Z., Zou, Z., Pei, J., et al. (2020) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2020.2970709 · Cited by 201
“Actually, many neural network hardwares also adopt this precision level or even lower [30], [35], [37], [42], [46], [75], [80], [87], [88].”
BackgroundMethodKim, H., Yoo, T., Kim, T. T.-H., & Kim, B. (2021) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2021.3061508 · Cited by 199
Analog architectures for neural network acceleration based on non-volatile memory
Xiao, T. P., Bennett, C. H., Feinberg, B., Agarwal, S., & Marinella, M. J. (2020) · Applied Physics Reviews · doi:10.1063/1.5143815 · Cited by 193
Taskmaster-1: Toward a Realistic and Diverse Dialog Dataset
Byrne, B., Krishnamoorthi, K., Sankar, C., Neelakantan, A., Goodrich, B., Duckworth, D., Yavuz, S., Dubey, A., Kim, K.-Y., & Cedilnik, A. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1459 · Cited by 139
“…is generally attributed to major breakthroughs in speech recognition and text-to-speech (TTS) technologies aided by recent progress in deep learning (Lecun et al., 2015), exponential gains in compute power (Steinkrau et al., 2005; Jouppi et al., 2017), and the ubiquity of powerful mobile devices.”
BackgroundExploring the Granularity of Sparsity in Convolutional Neural Networks
Mao, H., Han, S., Pool, J., Li, W., Liu, X., Wang, Y., & Dally, W. J. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2017.241 · Cited by 198
UNet-based model for crack detection integrating visual explanations
Liu, F., & Wang, L. (2022) · Construction and Building Materials · doi:10.1016/j.conbuildmat.2021.126265 · Cited by 191
Mattson, P., Cheng, C., Coleman, C., Diamos, G., Micikevicius, P., Patterson, D., Tang, H., Wei, G.-Y., Bailis, P., Bittorf, V., et al. (2019) · arXiv · doi:10.48550/arxiv.1910.01500 · Cited by 206
“As the number of hardware and software systems for DL training increases (Paszke et al., 2017; Abadi et al., 2016; Chen et al., 2015; Jia et al., 2014; Jouppi et al., 2017; Chen et al., 2018; Markidis et al., 2018; Intel, 2019), so does the need for a comprehensive benchmark.”
BackgroundSelf-citationZigZag: 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 248
Scalable Second Order Optimization for Deep LearningPreprint
Anil, R., Gupta, V., Koren, T., Regan, K., & Singer, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.09018 · Cited by 237
“Our method is designed to run effectively on modern neural network accelerators such as TPUs (Jouppi et al., 2017) or GPUs.”
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 238
“Image Classification Only Broader (include non-CNN workloads) Training [27] [32] [34] [48] [54] [55] [72] [78] [83] [8] [20] [51] [58] [67] [69] [86] Inference [10] [11] [12] [23] [26] [34] [36] [38] [54] [59] [60] [66] [70] [74] [75] [76] [78] [90] [91] [8] [35] [41] [43] [53] [67]”
Capra, M., Bussolino, B., Marchisio, A., Shafique, M., Masera, G., & Martina, M. (2020) · Future Internet · doi:10.3390/fi12070113 · Cited by 195
“Popular WS accelerators are [59,60] In recent years, to further improve the performance and energy efficiency of accelerators, attention has been focused on new strategies.”
BackgroundIn-memory Learning with Analog Resistive Switching Memory: A Review and Perspective
Xi, Y., Gao, B., Tang, J., Chen, A., Chang, M.-F., Hu, X. S., Spiegel, J. V. D., Qian, H., & Wu, H. (2021) · Proceedings of the IEEE · doi:10.1109/jproc.2020.3004543 · Cited by 196
“Also, custom-designed processors and accelerators have been used, for example, Manchester’s SpiNNaker [4], Zurich’s ROLLS [5], Heidelberger’s BrainScaleS [6], Google’s tensor processing unit (TPU) [7], and Tsinghua’s Tianjic [8].”
MethodEnergy-Efficient Neural Network Accelerator Based on Outlier-Aware Low-Precision Computation
Park, E., Kim, D., & Yoo, S. (2018) · 2018 ACM/IEEE 45th Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca.2018.00063 · Cited by 201
Machine Learning With Neuromorphic Photonics
de Lima, T. F., Peng, H.-T., Tait, A. N., Nahmias, M. A., Miller, H. B., Shastri, B. J., & Prucnal, P. R. (2019) · Journal of Lightwave Technology · doi:10.1109/jlt.2019.2903474 · Cited by 200
“IBM [1], [2], HP [3], Intel [4], and Google [5], [6], have all shifted their core technological strategies from “mobile first” to “AI first”.”
BackgroundMagalhães, S. A., Castro, L., Moreira, G., dos Santos, F. N., Cunha, M., Dias, J., & Moreira, A. P. (2021) · Sensors · doi:10.3390/s21103569 · Cited by 184
Value-Aware Quantization for Training and Inference of Neural Networks
Park, E., Yoo, S., & Vajda, P. (2018) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-01225-0_36 · Cited by 181
Nowatzki, T., Gangadhar, V., Ardalani, N., & Sankaralingam, K. (2017) · Proceedings of the 44th Annual International Symposium on Computer Architecture · doi:10.1145/3079856.3080255 · Cited by 176
UCNN: Exploiting Computational Reuse in Deep Neural Networks via Weight Repetition
Hegde, K., Yu, J., Agrawal, R., Yan, M., Pellauer, M., & Fletcher, C. (2018) · 2018 ACM/IEEE 45th Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca.2018.00062 · Cited by 191
Efficient and Effective Sparse LSTM on FPGA with Bank-Balanced Sparsity
Cao, S., Zhang, C., Yao, Z., Xiao, W., Nie, L., Zhan, D., Liu, Y., Wu, M., & Zhang, L. (2019) · Proceedings of the 2019 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays · doi:10.1145/3289602.3293898 · Cited by 178
Prospects and applications of photonic neural networks
Huang, C., Sorger, V. J., Miscuglio, M., Al-Qadasi, M., Mukherjee, A., Lampe, L., Nichols, M., Tait, A. N., Ferreira de Lima, T., Marquez, B. A., et al. (2022) · Advances in Physics: X · doi:10.1080/23746149.2021.1981155 · Cited by 180
ScaleDeep: A Scalable Compute Architecture for Learning and Evaluating Deep Networks
Venkataramani, S., Ranjan, A., Banerjee, S., Das, D., Avancha, S., Jagannathan, A., Durg, A., Nagaraj, D., Kaul, B., Dubey, P., et al. (2017) · Proceedings of the 44th Annual International Symposium on Computer Architecture · doi:10.1145/3079856.3080244 · Cited by 188
Compute Solution for Tesla's Full Self-Driving Computer
Talpes, E., Sarma, D. D., Venkataramanan, G., Bannon, P., McGee, B., Floering, B., Jalote, A., Hsiong, C., Arora, S., Gorti, A., et al. (2020) · IEEE Micro · doi:10.1109/mm.2020.2975764 · Cited by 193
“Thus, each MAC unit computes the dotproduct locally with no data movement within the MAC array, unlike the systolic array computations implemented in many other processors.(3,4) This results in lower power and less cell area than systolic array implementations.”
Ham, T. J., Lee, Y., Seo, S. H., Kim, S., Choi, H., Jung, S. J., & Lee, J. W. (2021) · 2021 ACM/IEEE 48th Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca52012.2021.00060 · Cited by 192
“Various hardware accelerators [6], [8]–[10], [19], [21], [29], [41], [42], [69] have been proposed to accelerate key neural network operations represented as matrix multiplications.”
BackgroundPark, J., Naumov, M., Basu, P., Deng, S., Kalaiah, A., Khudia, D. S., Law, J., Malani, P., Malevich, A., Nadathur, S., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.09886 · Cited by 215
“In order to tackle this trend, a lot of research has been done on optimizing computing platforms for DL, including but not limited to [1, 18, 25, 33, 49, 50, 61, 62].”
BackgroundMethodSelf-citationChasing Carbon: The Elusive Environmental Footprint of Computing
Gupta, U., Kim, Y. G., Lee, S., Tse, J., Lee, H.-H. S., Wei, G.-Y., Brooks, D. M., & Wu, C.-J. (2022) · IEEE Micro · doi:10.1109/mm.2022.3163226 · Cited by 250
Duan, Q., Huang, J., Hu, S., Deng, R., Lu, Z., & Yu, S. (2023) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2023.3316615 · Cited by 159
TANGRAM: 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
DeepRecSys: A System for Optimizing End-To-End At-Scale Neural Recommendation Inference
Gupta, U., Hsia, S., Saraph, V., Wang, X., Reagen, B., Wei, G.-Y., Lee, H.-H. S., Brooks, D., & Wu, C.-J. (2020) · 2020 ACM/IEEE 47th Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca45697.2020.00084 · Cited by 173
“Similar capacity demands can be found at Google [11], Amazon [8], [12], and Alibaba [5], [6].”
BackgroundInfluentialOliVe: Accelerating Large Language Models via Hardware-friendly Outlier-Victim Pair Quantization
Guo, C., Tang, J., Hu, W., Leng, J., Zhang, C., Yang, F., Liu, Y., Guo, M., & Zhu, Y. (2023) · Proceedings of the 50th Annual International Symposium on Computer Architecture · doi:10.1145/3579371.3589038 · Cited by 152
HPC Cloud for Scientific and Business Applications: Taxonomy, Vision, and Research Challenges
Netto, M. A. S., Calheiros, R. N., Rodrigues, E. R., Cunha, R. L. F., & Buyya, R. (2018) · ACM Computing Surveys · doi:10.1145/3150224 · Cited by 227
Hooker, S. (2020) · arXiv · doi:10.48550/arxiv.2009.06489 · Cited by 185
“The first examples of domain specialized hardware released over the last few years – TPUs (Jouppi et al., 2017), edge-TPUs (Gupta & Tan, 2019), Arm Cortex-M55 (ARM, 2020), Facebook’s big sur (Lee & Wang, 2018) – optimize explicitly for costly operations common to deep neural networks like matrix multiplies.”
BackgroundInfluentialIntelligent and connected vehicles: Current status and future perspectives
Yang, D., Jiang, K., Zhao, D., Yu, C., Cao, Z., Xie, S., Xiao, Z., Jiao, X., Wang, S., & Zhang, K. (2018) · Science China Technological Sciences · doi:10.1007/s11431-017-9338-1 · Cited by 176
Zhang, J. J., Gu, T., Basu, K., & Garg, S. (2018) · 2018 IEEE 36th VLSI Test Symposium (VTS) · doi:10.1109/vts.2018.8368656 · Cited by 173
Marinella, M. J., Agarwal, S., Hsia, A., Richter, I., Jacobs-Gedrim, R., Niroula, J., Plimpton, S. J., Ipek, E., & James, C. D. (2018) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2018.2796379 · Cited by 183
“Google has recently provided a performance analysis of their tensor processing unit (TPU) being used for inference with deep MLPs, convolutional neural nets (CNNs), and long short-term memory deep networks with as many as 108 weights [7].”
MethodKohn-Sham Equations as Regularizer: Building Prior Knowledge into Machine-Learned Physics
Li, L., Hoyer, S., Pederson, R., Sun, R., Cubuk, E. D., Riley, P., & Burke, K. (2021) · Physical Review Letters · doi:10.1103/physrevlett.126.036401 · Cited by 169
Device and Circuit Architectures for In‐Memory Computing
Ielmini, D., & Pedretti, G. (2020) · Advanced Intelligent Systems · doi:10.1002/aisy.202000040 · Cited by 170
Elf: accelerate high-resolution mobile deep vision with content-aware parallel offloading
Zhang, W., He, Z., Liu, L., Jia, Z., Liu, Y., Gruteser, M., Raychaudhuri, D., & Zhang, Y. (2021) · Proceedings of the 27th Annual International Conference on Mobile Computing and Networking · doi:10.1145/3447993.3448628 · Cited by 172
“On the other hand, people have proposed to increase the computing resources by using massive accelerators, such as GPU, FPGA [14] and ASIC [15].”
MethodInfluentialTraining Compute-Optimal Large Language Models
Hoffmann, J., Borgeaud, S., Mensch, A., Buchatskaya, E., Cai, T., Rutherford, E., Casas, D. D. L., Hendricks, L. A., Welbl, J., Clark, A., et al. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-2176 · Cited by 932
“All models in this analysis have been trained on TPUv3/TPUv4 (Jouppi et al., 2017) with JAX (Bradbury et al., 2018) and Haiku (Hennigan et al., 2020).”
Channel Gating Neural Networks
Hua, W., Zhou, Y. J., De, C., Zhang, Z., & Suh, G. E. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.12549 · Cited by 124
“, small comparators for gating functions) to support fine-grained pruning decisions on a dense systolic array; (2) channel gating maintains the locality and the regularity in both computations and data accesses, and can be efficiently implemented with small changes to a systolic array architecture similar to the Googl…”
MethodCloudscout: a deep neural network for on-board cloud detection on hyperspectral images
Giuffrida, G., Diana, L., de Gioia, F., Benelli, G., Meoni, G., Donati, M., & Fanucci, L. (2020) · Remote Sensing · doi:10.3390/rs12142205 · Cited by 176
“TPU : TPU is an innovative hardware accelerator dedicated to a particular data structure: Tensors [16].”
Deep Neural Network Approximation for Custom Hardware: Where We've Been, Where We're Going
Wang, E., Davis, J. J., Zhao, R., Ng, H.-C., Niu, X., Luk, W., Cheung, P. Y. K., & Constantinides, G. A. (2020) · ACM Computing Surveys · doi:10.1145/3309551 · Cited by 166
“Their timeconsuming and resource-demanding design and fabrication processes, however, make it hard for them to keep up with the rapid evolution of DNN algorithms [28, 66].”
BackgroundMethodInfluentialCommunication-Efficient Distributed Deep Learning: A Comprehensive Survey
Tang, Z., Shi, S., Wang, W., Li, B., & Chu, X. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.06307 · Cited by 86
“2D-Torus All-Reduce [84, 127] can also massively reduce the communication latency using a 2D-Torus topology network.”
MethodSelf-citationMaking AI Forget You: Data Deletion in Machine Learning
Ginart, A., Guan, M. Y., Valiant, G., & Zou, J. (2019) · arXiv · doi:10.48550/arxiv.1907.05012 · Cited by 175
“Deep Neural Networks and Stochastic Gradient Descent A line of research has observed the robustness of neural network training robustness to quantization and pruning [84, 46, 40, 71, 25, 52].”
BackgroundA domain-specific architecture for deep neural networks
Jouppi, N. P., Young, C., Patil, N., & Patterson, D. (2018) · Communications of the ACM · doi:10.1145/3154484 · Cited by 166
Self-citationSilicon microring synapses enable photonic deep learning beyond 9-bit precision
Zhang, W., Huang, C., Peng, H.-T., Bilodeau, S., Jha, A., Blow, E., de Lima, T. F., Shastri, B. J., & Prucnal, P. (2022) · Optica · doi:10.1364/optica.446100 · Cited by 163
“The state-of-the-art DNNs implemented in electronics, like the TPU [8], have a precision of 8 bits or higher.”
Machine Learning on DNA-Encoded Libraries: A New Paradigm for Hit Finding
McCloskey, K., Sigel, E. A., Kearnes, S., Xue, L., Tian, X., Moccia, D., Gikunju, D., Bazzaz, S., Chan, B., Clark, M. A., et al. (2020) · Journal of Medicinal Chemistry · doi:10.1021/acs.jmedchem.0c00452 · Cited by 158
Yin, S., Sun, X., Yu, S., & Seo, J.-S. (2020) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2020.3015178 · Cited by 171
A Berkeley View of Systems Challenges for AI
Stoica, I., Song, D., Popa, R. A., Patterson, D., Mahoney, M. W., Katz, R., Joseph, A. D., Jordan, M., Hellerstein, J. M., Gonzalez, J. E., et al. (2017) · arXiv · doi:10.48550/arxiv.1712.05855 · Cited by 171
“ing and scienti•c •elds and cuts across many disciplines in computing. In particular, computer systems have already proved essential in catalyzing recent progress in AI. Advances in parallel hardware [31, 58, 90] and scalable so›ware systems [32, 46, 114] have sparked the development of new machine learning frameworks…”
BackgroundSelf-citationBringing ai to edge: from deep learning’s perspective
Liu, D., Kong, H., Luo, X., Liu, W., & Subramaniam, R. (2022) · Neurocomputing · doi:10.1016/j.neucom.2021.04.141 · Cited by 171
“The academia and industry have a consensus that the breakthroughs in hardware architecture will bring DNNs to a new level and boost DNN adoption [15] [205].”
BackgroundTowards Unified INT8 Training for Convolutional Neural Network
Zhu, F., Gong, R., Yu, F., Liu, X., Wang, Y., Li, Z., Yang, X., & Yan, J. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00204 · Cited by 160
Gmail Smart Compose: Real-Time Assisted Writing
Chen, M. X., Lee, B. N., Bansal, G., Cao, Y., Zhang, S., Lu, J., Tsay, J., Wang, Y., Dai, A. M., Chen, Z., et al. (2019) · Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining · doi:10.1145/3292500.3330723 · Cited by 169
“Initially, the first-generation tensor processing unit (TPU) [25] was used to accelerate matrix-multiplications and activation functions.”
MethodPathways to efficient neuromorphic computing with non-volatile memory technologies
Chakraborty, I., Jaiswal, A., Saha, A. K., Gupta, S. K., & Roy, K. (2020) · Applied Physics Reviews · doi:10.1063/1.5113536 · Cited by 168
Kraft, M., Piechocki, M., Ptak, B., & Walas, K. (2021) · Remote Sensing · doi:10.3390/rs13050965 · Cited by 187
A New Golden Age in Computer Architecture: Empowering the Machine-Learning Revolution
Dean, J., Patterson, D., & Young, C. (2018) · IEEE Micro · doi:10.1109/mm.2018.112130030 · Cited by 168
Self-citationDevice and materials requirements for neuromorphic computing
Islam, R., Li, H., Chen, P.-Y., Wan, W., Chen, H.-Y., Gao, B., Wu, H., Yu, S., Saraswat, K., & Philip Wong, H.-S. (2019) · Journal of Physics D: Applied Physics · doi:10.1088/1361-6463/aaf784 · Cited by 158
Verhelst, M., & Moons, B. (2017) · IEEE Solid-State Circuits Magazine · doi:10.1109/mssc.2017.2745818 · Cited by 158
“These functional units are each equipped with a very small SRAM (as in [14]) or even just registers (as in [9]) to store data locally and maximize data reuse within the array.”
BackgroundCapuchin: Tensor-based GPU Memory Management for Deep Learning
Peng, X., Shi, X., Dai, H., Jin, H., Ma, W., Xiong, Q., Yang, F., & Qian, X. (2020) · International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) · doi:10.1145/3373376.3378505 · Cited by 164
“Deep neural networks (DNNs) become very popular nowadays due to the tremendous compute power provided by several kinds of emerging heterogeneous hardware, such as TPU [17], ASIC, and GPU — currently the most prevalent training option.”
BackgroundThe Design Process for Google's Training Chips: TPUv2 and TPUv3
Norrie, T., Patil, N. P., Yoon, D. H., Kurian, G. T., Li, S., Laudon, J. P., Young, C., Jouppi, N. P., & Patterson, D. A. (2021) · IEEE Micro · doi:10.1109/mm.2021.3058217 · Cited by 199
“The motivation to slash ML inference serving costs led to the TPUv1 inference system, deployed successfully in 2015.(5)”
MethodSelf-citationAlgorithm and hardware implementation for visual perception system in autonomous vehicle: A survey
Shi, W., Alawieh, M. B., Li, X., & Yu, H. (2017) · Integration · doi:10.1016/j.vlsi.2017.07.007 · Cited by 175
GCNAX: A Flexible and Energy-efficient Accelerator for Graph Convolutional Neural Networks
Li, J., Louri, A., Karanth, A., & Bunescu, R. (2021) · IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca51647.2021.00070 · Cited by 198
“For dense neural networks, the accelerators mainly focus on leveraging the massive parallelism to improve performance and utilization, such as TPU [55] and Eyeriss [16].”
BackgroundMethodPOSEIDON: Privacy-Preserving Federated Neural Network Learning
Sav, S., Pyrgelis, A., Troncoso-Pastoriza, J. R., Froelicher, D., Bossuat, J.-P., Sousa, J. S., & Hubaux, J.-P. (2021) · Proceedings 2021 Network and Distributed System Security Symposium · doi:10.14722/ndss.2021.24119 · Cited by 162
Meta-optics for spatial optical analog computing
Abdollahramezani, S., Hemmatyar, O., & Adibi, A. (2020) · Nanophotonics · doi:10.1515/nanoph-2020-0285 · Cited by 159
fpgaConvNet: Mapping regular and irregular convolutional neural networks on FPGAs
Venieris, S. I., & Bouganis, C.-S. (2019) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2018.2844093 · Cited by 161
“The increased computational needs of ConvNet models together with the recent hardware advances led to a shift toward customized hardware, with the prominence of graphics processing units (GPUs) [9] and application-specific integrated circuits (ASICs) [10].”
BackgroundAccelerating Deep Learning by Focusing on the Biggest LosersPreprint
Jiang, A. H., Wong, D. L. K., Zhou, G., Andersen, D. G., Dean, J., Ganger, G. R., Joshi, G., Kaminksy, M., Kozuch, M. A., Lipton, Z. C., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.00762 · Cited by 181
Self-citationHoang, L.-H., Hanif, M. A., & Shafique, M. (2020) · Design, Automation, and Test in Europe · doi:10.23919/date48585.2020.9116571 · Cited by 161
“To efficiently process data using these networks, specialized hardware accelerators are utilized which are built using smaller technology nodes, in order to achieve high power and performance efficiency [2], [3], [4].”
FastMoE: A Fast Mixture-of-Expert Training System
He, J., Qiu, J., Zeng, A., Yang, Z., Zhai, J., & Tang, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.13262 · Cited by 37
“However, despite the existence of some naive single-GPU implementation in PyTorch (Rau, 2019), the only present system that supports scalable MoE training is based on Google’s private hardware and software stack — TPU (Jouppi et al., 2017) and Mesh TensorFlow (Shazeer et al.”
MethodInfluentialDeep 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
“TPU [83] is a chip designed specifically to accelerate the computing of deep neural networks.”
BackgroundPark, J., Kumar, A., Zhou, Y., Oh, S., Kim, J. H., Shi, Y., Jain, S., Hota, G., Qiu, E., Nagle, A. L., et al. (2024) · Nature Communications · doi:10.1038/s41467-024-46682-1 · Cited by 171
“While GPUs and tensor processing units excel in parallel computing compared to CPUs, they are still reliant on static random access memory, which demands substantial physical space 2,3 .”
Graviton: trusted execution environments on GPUs
Volos, S., Vaswani, K., & Bruno, R. (2018) · Operating Systems Design and Implementation · doi:10.5555/3291168.3291219 · Cited by 307
TransPIM: A Memory-based Acceleration via Software-Hardware Co-Design for Transformer
Zhou, M., Xu, W., Kang, J., & Rosing, T. (2022) · 2022 IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca53966.2022.00082 · Cited by 143
PREMA: A Predictive Multi-Task Scheduling Algorithm For Preemptible Neural Processing Units
Choi, Y., & Rhu, M. (2020) · 2020 IEEE International Symposium on High Performance Computer Architecture (HPCA) · doi:10.1109/hpca47549.2020.00027 · Cited by 142
“To meet the demands of computation-hungry deep neural network (DNN) based machine learning (ML) algorithms, researchers have put enormous efforts into developing DNN accelerators [1], [2], [3], [4], also known as neural processing units (NPUs).”
BackgroundMethodInfluentialDeep Neural Network Through an InP SOA-Based Photonic Integrated Cross-Connect
Shi, B., Calabretta, N., & Stabile, R. (2019) · IEEE Journal of Selected Topics in Quantum Electronics · doi:10.1109/jstqe.2019.2945548 · Cited by 173
“electronics: among those we name IBM TrueNorth [9], Neurogrid [10], SpiNNaker [11], and BrainDrop [12], for spiking neural networks, FPGA [13] and Google TPU [14] for deep neural networks.”
MethodInfluentialImproving the Accuracy and Hardware Efficiency of Neural Networks Using Approximate Multipliers
Ansari, M. S., Mrázek, V., Cockburn, B. F., Sekanina, L., Vašíček, Z., & Han, J. (2019) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2019.2940943 · Cited by 167
“Through comprehensive simulations, we confirmed that 8-bit multipliers are just wide enough to provide reasonable accuracies in NNs [10], [20].”
BackgroundAnalyzing and mitigating data stalls in DNN training
Mohan, J., Phanishayee, A., Raniwala, A., & Chidambaram, V. (2021) · Proceedings of the VLDB Endowment · doi:10.14778/3446095.3446100 · Cited by 187
“hardware solutions, by efficiently using available hardware. Optimizing DNN training time. A number of solutions have been proposed to reduce the training time for DNNs including specialized hardware [4, 9, 54, 57, 76, 77, 80, 86], parallel training [32, 36, 48, 56, 59, 60, 73], GPU memory optimizations [31, 52, 81],…”
MethodEfficient 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
“Tensor Processing Unit (TPU) is developed by Google in order to implement machine learning algorithms. matrix multiplication unit as a systolic array of 256 × 256 units is used in the TPU architecture [120].”
MethodSolving Fourier ptychographic imaging problems via neural network modeling and TensorFlow
Jiang, S., Guo, K., Liao, J. B., & Zheng, G. (2018) · Biomedical Optics Express · doi:10.1364/boe.9.003306 · Cited by 160
CoSA: Scheduling by Constrained Optimization for Spatial Accelerators
Huang, Q., Kang, M., Dinh, G., Norell, T., Kalaiah, A., Demmel, J. W., Wawrzynek, J., & Shao, Y. S. (2021) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1109/isca52012.2021.00050 · Cited by 196
“Motivated by the high computational requirements of DNNs, there have been exciting developments in both research and commercial spaces in building specialized DNN accelerators for both edge [1], [16], [17], [26], [50], [61], [63], [72] and cloud applications [5], [19], [27], [36], [39], [69].”
BackgroundA Comparison of Techniques for Language Model Integration in Encoder-Decoder Speech Recognition
Toshniwal, S., Kannan, A., Chiu, C.-C., Wu, Y., Sainath, T. N., & Livescu, K. (2018) · 2018 IEEE Spoken Language Technology Workshop (SLT) · doi:10.1109/slt.2018.8639038 · Cited by 149
MAGNet: A Modular Accelerator Generator for Neural Networks
Venkatesan, R., Shao, Y. S., Wang, M., Clemons, J., Dai, S., Fojtik, M., Keller, B., Klinefelter, A., Pinckney, N., Raina, P., et al. (2019) · 2019 IEEE/ACM International Conference on Computer-Aided Design (ICCAD) · doi:10.1109/iccad45719.2019.8942127 · Cited by 138
A comprehensive study of deep learning compiler bugs
Shen, Q., Ma, H., Chen, J., Tian, Y., Cheung, S.-C., & Chen, X. (2021) · ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE) · doi:10.1145/3468264.3468591 · Cited by 179
“Various kinds of hardware (e.g., Google TPU [41], Intel NNP [36], and Apple Bionic [43]) are also designed to accelerate the execution of these models.”
BackgroundGist: Efficient Data Encoding for Deep Neural Network Training
Jain, A., Phanishayee, A., Mars, J., Tang, L., & Pekhimenko, G. (2018) · 2018 ACM/IEEE 45th Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca.2018.00070 · Cited by 140
“While using FPGAs [50] and ASICs [28], [25] is also possible, most of these designs are either proprietary or in relatively early development stages.”
BackgroundMethod
Metadata
- Authors
- Jouppi, N. P., Young, C., Patil, N., Patterson, D., Agrawal, G., Bajwa, R., Bates, S., Bhatia, S., Boden, N., Borchers, A., Boyle, R., Cantin, P.-L., Chao, C., Clark, C., Coriell, J., Daley, M., Dau, M., Dean, J., Gelb, B., Ghaemmaghami, T. V., Gottipati, R., Gulland, W., Hagmann, R., Ho, C. R., Hogberg, D., Hu, J., Hundt, R., Hurt, D., Ibarz, J., Jaffey, A., Jaworski, A., Kaplan, A., Khaitan, H., Killebrew, D., Koch, A., Kumar, N., Lacy, S., Laudon, J., Law, J., Le, D., Leary, C., Liu, Z., Lucke, K., Lundin, A., MacKean, G., Maggiore, A., Mahony, M., Miller, K., Nagarajan, R., Narayanaswami, R., Ni, R., Nix, K., Norrie, T., Omernick, M., Penukonda, N., Phelps, A., Ross, J., Ross, M., Salek, A., Samadiani, E., Severn, C., Sizikov, G., Snelham, M., Souter, J., Steinberg, D., Swing, A., Tan, M., Thorson, G., Tian, B., Toma, H., Tuttle, E., Vasudevan, V., Walter, R., Wang, W., Wilcox, E., & Yoon, D. H.
- Affiliations
- Google (United States) (US)
- ORCID
- Venue
- Proceedings of the 44th Annual International Symposium on Computer Architecture
- Publisher
- ACM
- Published
- 2017-06-24
- Pages
- 1-12
- ISSN
- 1063-6897 (ISSN-L)
- OpenAlex
- W2606722458
- Cited by
- 4,537
- References
- 54
- Citations by year
- 2026: 231 · 2025: 429 · 2024: 409 · 2023: 498 · 2022: 528 · 2021: 741 · 2020: 641 · 2019: 582 · 2018: 396 · 2017: 77
- FWCI
- 329.20
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Proceedings - International Symposium on Computer Architecture/Proceedings (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Low-power high-performance VLSI design
- Keywords
- tensor processing units, neural network inference acceleration, domain-specific hardware, energy efficiency, deterministic execution, on-chip memory, CPU vs GPU comparison, TensorFlow
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.