FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312965666> ?p ?o ?g. }

• W4312965666 endingPage "377" @default.
• W4312965666 startingPage "364" @default.
• W4312965666 abstract "Computing-in-memory (CIM) relieves the Von Neumann bottleneck by storing the weights of neural networks in memory arrays. However, two challenges still exist, hindering the efficient acceleration of convolutional neural networks (CNN) in artificial intelligence (AI) edge devices. Firstly, the activations for sliding window (SW) operations in CNN still bring high memory access pressure. This can be alleviated by increasing the SW parallelism, but simple array replication suffers from poor array utilization and large peripheral circuits overhead. Secondly, the partial sums from individual CIM arrays, which are usually accumulated to obtain the final sum, introduce large latency due to enormous shift-and-add operations. Moreover, high-resolution ADCs are also needed to reduce the quantization error of partial sums, further increasing the hardware costs. In this paper, a hardware-efficient CIM accelerator, ARBiS, is proposed with improved activation reusability and bit-scalable matrix-vector-multiplication (MVM) for CNN acceleration in AI edge applications. The cyclic-shift weight duplication exploits a third dimension of receptive field (RF) depth for SW weight mapping to reduce the memory accesses of activations, improving the array utilization. The parasitic-capacitance charge sharing is employed to realize high-precision analog MVM in order to reduce the ADC cost. Compared with conventional architectures, ARBiS with parallel processing of 9 SW operations achieves 56.6%~58.8% alleviation of memory access pressure. Meanwhile, ARBiS configured with 8-bit ADCs saves 92.53%~94.53% ADC energy consumption. An ARBiS accelerator is evaluated to realize a computational efficiency (CE) of 10.28 (10.43) TOPS/mm2, an energy efficiency (EE) of 91.19 (112.36) TOPS/W with 8-bit (4-bit) ADCs, achieving <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$11.4sim 11.7times $ </tex-math></inline-formula> ( <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$11.6sim 11.8times $ </tex-math></inline-formula> ), <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$1.1sim 3.3times $ </tex-math></inline-formula> ( <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$1.4sim 4times $ </tex-math></inline-formula> ) improvements over state-of-the-art works, respectively." @default.
• W4312965666 created "2023-01-05" @default.
• W4312965666 creator A5028407427 @default.
• W4312965666 creator A5034082523 @default.
• W4312965666 creator A5034197769 @default.
• W4312965666 creator A5040236486 @default.
• W4312965666 creator A5078773129 @default.
• W4312965666 date "2023-01-01" @default.
• W4312965666 modified "2023-10-08" @default.
• W4312965666 title "ARBiS: A Hardware-Efficient SRAM CIM CNN Accelerator With Cyclic-Shift Weight Duplication and Parasitic-Capacitance Charge Sharing for AI Edge Application" @default.
• W4312965666 cites W1996982010 @default.
• W4312965666 cites W2015604477 @default.
• W4312965666 cites W2046103068 @default.
• W4312965666 cites W2048266589 @default.
• W4312965666 cites W2100483769 @default.
• W4312965666 cites W2105175332 @default.
• W4312965666 cites W2111186324 @default.
• W4312965666 cites W2183341477 @default.
• W4312965666 cites W2220689418 @default.
• W4312965666 cites W2508602506 @default.
• W4312965666 cites W2518281301 @default.
• W4312965666 cites W2525543902 @default.
• W4312965666 cites W2613989746 @default.
• W4312965666 cites W2798554798 @default.
• W4312965666 cites W2799404913 @default.
• W4312965666 cites W2800129760 @default.
• W4312965666 cites W2889005033 @default.
• W4312965666 cites W2894696827 @default.
• W4312965666 cites W2943765373 @default.
• W4312965666 cites W2945149766 @default.
• W4312965666 cites W2987467556 @default.
• W4312965666 cites W2997869757 @default.
• W4312965666 cites W3003821665 @default.
• W4312965666 cites W3005901860 @default.
• W4312965666 cites W3013080934 @default.
• W4312965666 cites W3032986074 @default.
• W4312965666 cites W3092083602 @default.
• W4312965666 cites W3092137903 @default.
• W4312965666 cites W3092585568 @default.
• W4312965666 cites W3133754064 @default.
• W4312965666 cites W3134304371 @default.
• W4312965666 cites W3134703406 @default.
• W4312965666 cites W3138828421 @default.
• W4312965666 cites W3165929947 @default.
• W4312965666 cites W3167271488 @default.
• W4312965666 cites W3214143889 @default.
• W4312965666 cites W4200067071 @default.
• W4312965666 cites W4221038786 @default.
• W4312965666 cites W4245731639 @default.
• W4312965666 cites W4280530370 @default.
• W4312965666 cites W811768827 @default.
• W4312965666 doi "https://doi.org/10.1109/tcsi.2022.3215535" @default.
• W4312965666 hasPublicationYear "2023" @default.
• W4312965666 type Work @default.
• W4312965666 citedByCount "1" @default.
• W4312965666 countsByYear W43129656662023 @default.
• W4312965666 crossrefType "journal-article" @default.
• W4312965666 hasAuthorship W4312965666A5028407427 @default.
• W4312965666 hasAuthorship W4312965666A5034082523 @default.
• W4312965666 hasAuthorship W4312965666A5034197769 @default.
• W4312965666 hasAuthorship W4312965666A5040236486 @default.
• W4312965666 hasAuthorship W4312965666A5078773129 @default.
• W4312965666 hasConcept C111919701 @default.
• W4312965666 hasConcept C13164978 @default.
• W4312965666 hasConcept C138236772 @default.
• W4312965666 hasConcept C152890283 @default.
• W4312965666 hasConcept C154945302 @default.
• W4312965666 hasConcept C173608175 @default.
• W4312965666 hasConcept C176649486 @default.
• W4312965666 hasConcept C41008148 @default.
• W4312965666 hasConcept C42935608 @default.
• W4312965666 hasConcept C63511323 @default.
• W4312965666 hasConcept C68043766 @default.
• W4312965666 hasConcept C79974875 @default.
• W4312965666 hasConcept C81363708 @default.
• W4312965666 hasConcept C9390403 @default.
• W4312965666 hasConcept C98986596 @default.
• W4312965666 hasConceptScore W4312965666C111919701 @default.
• W4312965666 hasConceptScore W4312965666C13164978 @default.
• W4312965666 hasConceptScore W4312965666C138236772 @default.
• W4312965666 hasConceptScore W4312965666C152890283 @default.
• W4312965666 hasConceptScore W4312965666C154945302 @default.
• W4312965666 hasConceptScore W4312965666C173608175 @default.
• W4312965666 hasConceptScore W4312965666C176649486 @default.
• W4312965666 hasConceptScore W4312965666C41008148 @default.
• W4312965666 hasConceptScore W4312965666C42935608 @default.
• W4312965666 hasConceptScore W4312965666C63511323 @default.
• W4312965666 hasConceptScore W4312965666C68043766 @default.
• W4312965666 hasConceptScore W4312965666C79974875 @default.
• W4312965666 hasConceptScore W4312965666C81363708 @default.
• W4312965666 hasConceptScore W4312965666C9390403 @default.
• W4312965666 hasConceptScore W4312965666C98986596 @default.
• W4312965666 hasFunder F4320321001 @default.
• W4312965666 hasFunder F4320321885 @default.
• W4312965666 hasIssue "1" @default.
• W4312965666 hasLocation W43129656661 @default.
• W4312965666 hasOpenAccess W4312965666 @default.
• W4312965666 hasPrimaryLocation W43129656661 @default.

• next