Matches in SemOpenAlex for { <https://semopenalex.org/work/W4301904874> ?p ?o ?g. }
Showing items 1 to 81 of
81
with 100 items per page.
- W4301904874 abstract "Many recent works have designed accelerators for Convolutional Neural Networks (CNNs). While digital accelerators have relied on near data processing, analog accelerators have further reduced data movement by performing in-situ computation. Recent works take advantage of highly parallel analog in-situ computation in memristor crossbars to accelerate the many vector-matrix multiplication operations in CNNs. However, these in-situ accelerators have two significant short-comings that we address in this work. First, the ADCs account for a large fraction of chip power and area. Second, these accelerators adopt a homogeneous design where every resource is provisioned for the worst case. By addressing both problems, the new architecture, Newton, moves closer to achieving optimal energy-per-neuron for crossbar accelerators. We introduce multiple new techniques that apply at different levels of the tile hierarchy. Two of the techniques leverage heterogeneity: one adapts ADC precision based on the requirements of every sub-computation (with zero impact on accuracy), and the other designs tiles customized for convolutions or classifiers. Two other techniques rely on divide-and-conquer numeric algorithms to reduce computations and ADC pressure. Finally, we place constraints on how a workload is mapped to tiles, thus helping reduce resource provisioning in tiles. For a wide range of CNN dataflows and structures, Newton achieves a 77% decrease in power, 51% improvement in energy efficiency, and 2.2x higher throughput/area, relative to the state-of-the-art ISAAC accelerator." @default.
- W4301904874 created "2022-10-06" @default.
- W4301904874 creator A5013135203 @default.
- W4301904874 creator A5034067036 @default.
- W4301904874 creator A5053755565 @default.
- W4301904874 creator A5067008084 @default.
- W4301904874 creator A5089864904 @default.
- W4301904874 date "2018-03-10" @default.
- W4301904874 modified "2023-09-26" @default.
- W4301904874 title "Newton: Gravitating Towards the Physical Limits of Crossbar Acceleration" @default.
- W4301904874 doi "https://doi.org/10.48550/arxiv.1803.06913" @default.
- W4301904874 hasPublicationYear "2018" @default.
- W4301904874 type Work @default.
- W4301904874 citedByCount "0" @default.
- W4301904874 crossrefType "posted-content" @default.
- W4301904874 hasAuthorship W4301904874A5013135203 @default.
- W4301904874 hasAuthorship W4301904874A5034067036 @default.
- W4301904874 hasAuthorship W4301904874A5053755565 @default.
- W4301904874 hasAuthorship W4301904874A5067008084 @default.
- W4301904874 hasAuthorship W4301904874A5089864904 @default.
- W4301904874 hasBestOaLocation W43019048741 @default.
- W4301904874 hasConcept C113775141 @default.
- W4301904874 hasConcept C11413529 @default.
- W4301904874 hasConcept C119599485 @default.
- W4301904874 hasConcept C121332964 @default.
- W4301904874 hasConcept C127413603 @default.
- W4301904874 hasConcept C13164978 @default.
- W4301904874 hasConcept C154945302 @default.
- W4301904874 hasConcept C172191483 @default.
- W4301904874 hasConcept C17349429 @default.
- W4301904874 hasConcept C173608175 @default.
- W4301904874 hasConcept C2742236 @default.
- W4301904874 hasConcept C29984679 @default.
- W4301904874 hasConcept C3826847 @default.
- W4301904874 hasConcept C41008148 @default.
- W4301904874 hasConcept C42935608 @default.
- W4301904874 hasConcept C45374587 @default.
- W4301904874 hasConcept C459310 @default.
- W4301904874 hasConcept C62520636 @default.
- W4301904874 hasConcept C76155785 @default.
- W4301904874 hasConcept C81363708 @default.
- W4301904874 hasConcept C84114770 @default.
- W4301904874 hasConcept C9390403 @default.
- W4301904874 hasConceptScore W4301904874C113775141 @default.
- W4301904874 hasConceptScore W4301904874C11413529 @default.
- W4301904874 hasConceptScore W4301904874C119599485 @default.
- W4301904874 hasConceptScore W4301904874C121332964 @default.
- W4301904874 hasConceptScore W4301904874C127413603 @default.
- W4301904874 hasConceptScore W4301904874C13164978 @default.
- W4301904874 hasConceptScore W4301904874C154945302 @default.
- W4301904874 hasConceptScore W4301904874C172191483 @default.
- W4301904874 hasConceptScore W4301904874C17349429 @default.
- W4301904874 hasConceptScore W4301904874C173608175 @default.
- W4301904874 hasConceptScore W4301904874C2742236 @default.
- W4301904874 hasConceptScore W4301904874C29984679 @default.
- W4301904874 hasConceptScore W4301904874C3826847 @default.
- W4301904874 hasConceptScore W4301904874C41008148 @default.
- W4301904874 hasConceptScore W4301904874C42935608 @default.
- W4301904874 hasConceptScore W4301904874C45374587 @default.
- W4301904874 hasConceptScore W4301904874C459310 @default.
- W4301904874 hasConceptScore W4301904874C62520636 @default.
- W4301904874 hasConceptScore W4301904874C76155785 @default.
- W4301904874 hasConceptScore W4301904874C81363708 @default.
- W4301904874 hasConceptScore W4301904874C84114770 @default.
- W4301904874 hasConceptScore W4301904874C9390403 @default.
- W4301904874 hasLocation W43019048741 @default.
- W4301904874 hasOpenAccess W4301904874 @default.
- W4301904874 hasPrimaryLocation W43019048741 @default.
- W4301904874 hasRelatedWork W1429949169 @default.
- W4301904874 hasRelatedWork W1543798151 @default.
- W4301904874 hasRelatedWork W1970548269 @default.
- W4301904874 hasRelatedWork W2042026112 @default.
- W4301904874 hasRelatedWork W2380696053 @default.
- W4301904874 hasRelatedWork W2791204867 @default.
- W4301904874 hasRelatedWork W2792900782 @default.
- W4301904874 hasRelatedWork W2963027573 @default.
- W4301904874 hasRelatedWork W4224944742 @default.
- W4301904874 hasRelatedWork W4301904874 @default.
- W4301904874 isParatext "false" @default.
- W4301904874 isRetracted "false" @default.
- W4301904874 workType "article" @default.