Matches in SemOpenAlex for { <https://semopenalex.org/work/W4249932213> ?p ?o ?g. }
- W4249932213 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." @default.
- W4249932213 created "2022-05-12" @default.
- W4249932213 creator A5004586541 @default.
- W4249932213 creator A5010051312 @default.
- W4249932213 creator A5012860782 @default.
- W4249932213 creator A5061612911 @default.
- W4249932213 date "2016-10-01" @default.
- W4249932213 modified "2023-10-11" @default.
- W4249932213 title "Fused-layer CNN accelerators" @default.
- W4249932213 cites W1990315422 @default.
- W4249932213 cites W2000967104 @default.
- W4249932213 cites W2094756095 @default.
- W4249932213 cites W2096645269 @default.
- W4249932213 cites W2097117768 @default.
- W4249932213 cites W2102543317 @default.
- W4249932213 cites W2117539524 @default.
- W4249932213 cites W2117696986 @default.
- W4249932213 cites W2125203716 @default.
- W4249932213 cites W2142801765 @default.
- W4249932213 cites W3004171485 @default.
- W4249932213 cites W4256629673 @default.
- W4249932213 cites W77656806 @default.
- W4249932213 doi "https://doi.org/10.1109/micro.2016.7783725" @default.
- W4249932213 hasPublicationYear "2016" @default.
- W4249932213 type Work @default.
- W4249932213 citedByCount "295" @default.
- W4249932213 countsByYear W42499322132017 @default.
- W4249932213 countsByYear W42499322132018 @default.
- W4249932213 countsByYear W42499322132019 @default.
- W4249932213 countsByYear W42499322132020 @default.
- W4249932213 countsByYear W42499322132021 @default.
- W4249932213 countsByYear W42499322132022 @default.
- W4249932213 countsByYear W42499322132023 @default.
- W4249932213 crossrefType "proceedings-article" @default.
- W4249932213 hasAuthorship W4249932213A5004586541 @default.
- W4249932213 hasAuthorship W4249932213A5010051312 @default.
- W4249932213 hasAuthorship W4249932213A5012860782 @default.
- W4249932213 hasAuthorship W4249932213A5061612911 @default.
- W4249932213 hasConcept C108583219 @default.
- W4249932213 hasConcept C111919701 @default.
- W4249932213 hasConcept C118524514 @default.
- W4249932213 hasConcept C119599485 @default.
- W4249932213 hasConcept C127413603 @default.
- W4249932213 hasConcept C13164978 @default.
- W4249932213 hasConcept C138885662 @default.
- W4249932213 hasConcept C141353440 @default.
- W4249932213 hasConcept C154945302 @default.
- W4249932213 hasConcept C165005293 @default.
- W4249932213 hasConcept C173608175 @default.
- W4249932213 hasConcept C178790620 @default.
- W4249932213 hasConcept C185592680 @default.
- W4249932213 hasConcept C2776401178 @default.
- W4249932213 hasConcept C2777674469 @default.
- W4249932213 hasConcept C2779227376 @default.
- W4249932213 hasConcept C41008148 @default.
- W4249932213 hasConcept C41895202 @default.
- W4249932213 hasConcept C42935608 @default.
- W4249932213 hasConcept C76155785 @default.
- W4249932213 hasConcept C81363708 @default.
- W4249932213 hasConcept C9390403 @default.
- W4249932213 hasConcept C96324660 @default.
- W4249932213 hasConcept C98045186 @default.
- W4249932213 hasConceptScore W4249932213C108583219 @default.
- W4249932213 hasConceptScore W4249932213C111919701 @default.
- W4249932213 hasConceptScore W4249932213C118524514 @default.
- W4249932213 hasConceptScore W4249932213C119599485 @default.
- W4249932213 hasConceptScore W4249932213C127413603 @default.
- W4249932213 hasConceptScore W4249932213C13164978 @default.
- W4249932213 hasConceptScore W4249932213C138885662 @default.
- W4249932213 hasConceptScore W4249932213C141353440 @default.
- W4249932213 hasConceptScore W4249932213C154945302 @default.
- W4249932213 hasConceptScore W4249932213C165005293 @default.
- W4249932213 hasConceptScore W4249932213C173608175 @default.
- W4249932213 hasConceptScore W4249932213C178790620 @default.
- W4249932213 hasConceptScore W4249932213C185592680 @default.
- W4249932213 hasConceptScore W4249932213C2776401178 @default.
- W4249932213 hasConceptScore W4249932213C2777674469 @default.
- W4249932213 hasConceptScore W4249932213C2779227376 @default.
- W4249932213 hasConceptScore W4249932213C41008148 @default.
- W4249932213 hasConceptScore W4249932213C41895202 @default.
- W4249932213 hasConceptScore W4249932213C42935608 @default.
- W4249932213 hasConceptScore W4249932213C76155785 @default.
- W4249932213 hasConceptScore W4249932213C81363708 @default.
- W4249932213 hasConceptScore W4249932213C9390403 @default.
- W4249932213 hasConceptScore W4249932213C96324660 @default.
- W4249932213 hasConceptScore W4249932213C98045186 @default.
- W4249932213 hasLocation W42499322131 @default.
- W4249932213 hasOpenAccess W4249932213 @default.
- W4249932213 hasPrimaryLocation W42499322131 @default.
- W4249932213 hasRelatedWork W2062932566 @default.
- W4249932213 hasRelatedWork W2085828379 @default.
- W4249932213 hasRelatedWork W2170552397 @default.
- W4249932213 hasRelatedWork W2271847574 @default.
- W4249932213 hasRelatedWork W2348562861 @default.
- W4249932213 hasRelatedWork W2390042878 @default.
- W4249932213 hasRelatedWork W2540393334 @default.
- W4249932213 hasRelatedWork W2546284597 @default.
- W4249932213 hasRelatedWork W2999668243 @default.
- W4249932213 hasRelatedWork W2544043553 @default.
- W4249932213 isParatext "false" @default.