FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385079957> ?p ?o ?g. }

• W4385079957 abstract "Analog non-volatile memory (NVM)-based accelerators for Deep Neural Networks (DNNs) can achieve high-throughput and energy-efficient multiply-accumulate (MAC) operations by taking advantage of massively parallelized analog compute, implemented with Ohm's law and Kirchhoff's current law on arrays of resistive memory devices. Competitive end-to-end DNN accuracies can be obtained, provided that weights are accurately programmed onto NVM devices and MAC operations are sufficiently linear. In this paper, we report architectural and circuit advances for such Analog NVM-based accelerators. We describe a highly heterogeneous and programmable accelerator architecture for DNN inference that combines analog NVM memory-array “Tiles” for weight-stationary, energy-efficient MAC operations, together with heterogeneous special-function Compute-Cores for auxiliary digital computation. Massively parallel vectors of neuron-activation data are exchanged over short distances using a dense and efficient circuit-switched 2D mesh, enabling a wide range of DNN workloads, including CNNs, LSTMs, and Transformers. We also show a 14-nm inference chip consisting of multiple <tex xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>$mathbf{512}times mathbf{512}$</tex> arrays of Phase Change Memory (PCM) devices which implements multiple DNN benchmarks using such a circuit-switched 2D mesh." @default.
• W4385079957 created "2023-07-23" @default.
• W4385079957 creator A5001830342 @default.
• W4385079957 creator A5004201759 @default.
• W4385079957 creator A5004393978 @default.
• W4385079957 creator A5008815965 @default.
• W4385079957 creator A5014981734 @default.
• W4385079957 creator A5015811754 @default.
• W4385079957 creator A5016427894 @default.
• W4385079957 creator A5022844115 @default.
• W4385079957 creator A5026800272 @default.
• W4385079957 creator A5026837733 @default.
• W4385079957 creator A5028455655 @default.
• W4385079957 creator A5041739657 @default.
• W4385079957 creator A5042571447 @default.
• W4385079957 creator A5051232628 @default.
• W4385079957 creator A5055829440 @default.
• W4385079957 creator A5058290381 @default.
• W4385079957 creator A5065539173 @default.
• W4385079957 creator A5066277873 @default.
• W4385079957 creator A5072695145 @default.
• W4385079957 creator A5079173979 @default.
• W4385079957 creator A5084000491 @default.
• W4385079957 creator A5086674756 @default.
• W4385079957 creator A5086723230 @default.
• W4385079957 date "2023-05-21" @default.
• W4385079957 modified "2023-10-16" @default.
• W4385079957 title "Architectures and Circuits for Analog-memory-based Hardware Accelerators for Deep Neural Networks (Invited)" @default.
• W4385079957 cites W2751084147 @default.
• W4385079957 cites W2777126322 @default.
• W4385079957 cites W2803163155 @default.
• W4385079957 cites W2806113867 @default.
• W4385079957 cites W2890826549 @default.
• W4385079957 cites W2969812992 @default.
• W4385079957 cites W3183264033 @default.
• W4385079957 cites W3205692284 @default.
• W4385079957 cites W4200192033 @default.
• W4385079957 cites W4239824580 @default.
• W4385079957 cites W4313251083 @default.
• W4385079957 doi "https://doi.org/10.1109/iscas46773.2023.10181650" @default.
• W4385079957 hasPublicationYear "2023" @default.
• W4385079957 type Work @default.
• W4385079957 citedByCount "0" @default.
• W4385079957 crossrefType "proceedings-article" @default.
• W4385079957 hasAuthorship W4385079957A5001830342 @default.
• W4385079957 hasAuthorship W4385079957A5004201759 @default.
• W4385079957 hasAuthorship W4385079957A5004393978 @default.
• W4385079957 hasAuthorship W4385079957A5008815965 @default.
• W4385079957 hasAuthorship W4385079957A5014981734 @default.
• W4385079957 hasAuthorship W4385079957A5015811754 @default.
• W4385079957 hasAuthorship W4385079957A5016427894 @default.
• W4385079957 hasAuthorship W4385079957A5022844115 @default.
• W4385079957 hasAuthorship W4385079957A5026800272 @default.
• W4385079957 hasAuthorship W4385079957A5026837733 @default.
• W4385079957 hasAuthorship W4385079957A5028455655 @default.
• W4385079957 hasAuthorship W4385079957A5041739657 @default.
• W4385079957 hasAuthorship W4385079957A5042571447 @default.
• W4385079957 hasAuthorship W4385079957A5051232628 @default.
• W4385079957 hasAuthorship W4385079957A5055829440 @default.
• W4385079957 hasAuthorship W4385079957A5058290381 @default.
• W4385079957 hasAuthorship W4385079957A5065539173 @default.
• W4385079957 hasAuthorship W4385079957A5066277873 @default.
• W4385079957 hasAuthorship W4385079957A5072695145 @default.
• W4385079957 hasAuthorship W4385079957A5079173979 @default.
• W4385079957 hasAuthorship W4385079957A5084000491 @default.
• W4385079957 hasAuthorship W4385079957A5086674756 @default.
• W4385079957 hasAuthorship W4385079957A5086723230 @default.
• W4385079957 hasConcept C11413529 @default.
• W4385079957 hasConcept C119599485 @default.
• W4385079957 hasConcept C127413603 @default.
• W4385079957 hasConcept C134146338 @default.
• W4385079957 hasConcept C154945302 @default.
• W4385079957 hasConcept C165801399 @default.
• W4385079957 hasConcept C173608175 @default.
• W4385079957 hasConcept C182019814 @default.
• W4385079957 hasConcept C190475519 @default.
• W4385079957 hasConcept C29984679 @default.
• W4385079957 hasConcept C41008148 @default.
• W4385079957 hasConcept C45374587 @default.
• W4385079957 hasConcept C50644808 @default.
• W4385079957 hasConcept C76155785 @default.
• W4385079957 hasConcept C9390403 @default.
• W4385079957 hasConceptScore W4385079957C11413529 @default.
• W4385079957 hasConceptScore W4385079957C119599485 @default.
• W4385079957 hasConceptScore W4385079957C127413603 @default.
• W4385079957 hasConceptScore W4385079957C134146338 @default.
• W4385079957 hasConceptScore W4385079957C154945302 @default.
• W4385079957 hasConceptScore W4385079957C165801399 @default.
• W4385079957 hasConceptScore W4385079957C173608175 @default.
• W4385079957 hasConceptScore W4385079957C182019814 @default.
• W4385079957 hasConceptScore W4385079957C190475519 @default.
• W4385079957 hasConceptScore W4385079957C29984679 @default.
• W4385079957 hasConceptScore W4385079957C41008148 @default.
• W4385079957 hasConceptScore W4385079957C45374587 @default.
• W4385079957 hasConceptScore W4385079957C50644808 @default.
• W4385079957 hasConceptScore W4385079957C76155785 @default.
• W4385079957 hasConceptScore W4385079957C9390403 @default.
• W4385079957 hasFunder F4320307762 @default.
• W4385079957 hasLocation W43850799571 @default.
• W4385079957 hasOpenAccess W4385079957 @default.

• next