FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4242495425> ?p ?o ?g. }

• W4242495425 endingPage "26" @default.
• W4242495425 startingPage "13" @default.
• W4242495425 abstract "Deep Neural Networks (DNNs) have demonstrated state-of-the-art performance on a broad range of tasks involving natural language, speech, image, and video processing, and are deployed in many real world applications. However, DNNs impose significant computational challenges owing to the complexity of the networks and the amount of data they process, both of which are projected to grow in the future. To improve the efficiency of DNNs, we propose ScaleDeep, a dense, scalable server architecture, whose processing, memory and interconnect subsystems are specialized to leverage the compute and communication characteristics of DNNs. While several DNN accelerator designs have been proposed in recent years, the key difference is that ScaleDeep primarily targets DNN training, as opposed to only inference or evaluation. The key architectural features from which ScaleDeep derives its efficiency are: (i) heterogeneous processing tiles and chips to match the wide diversity in computational characteristics (FLOPs and Bytes/FLOP ratio) that manifest at different levels of granularity in DNNs, (ii) a memory hierarchy and 3-tiered interconnect topology that is suited to the memory access and communication patterns in DNNs, (iii) a low-overhead synchronization mechanism based on hardware data-flow trackers, and (iv) methods to map DNNs to the proposed architecture that minimize data movement and improve core utilization through nested pipelining. We have developed a compiler to allow any DNN topology to be programmed onto ScaleDeep, and a detailed architectural simulator to estimate performance and energy. The simulator incorporates timing and power models of ScaleDeep's components based on synthesis to Intel's 14nm technology. We evaluate an embodiment of ScaleDeep with 7032 processing tiles that operates at 600 MHz and has a peak performance of 680 TFLOPs (single precision) and 1.35 PFLOPs (half-precision) at 1.4KW. Across 11 state-of-the-art DNNs containing 0.65M-14.9M neurons and 6.8M-145.9M weights, including winners from 5 years of the ImageNet competition, ScaleDeep demonstrates 6x-28x speedup at iso-power over the state-of-the-art performance on GPUs." @default.
• W4242495425 created "2022-05-12" @default.
• W4242495425 creator A5010094713 @default.
• W4242495425 creator A5013922094 @default.
• W4242495425 creator A5018287086 @default.
• W4242495425 creator A5021152315 @default.
• W4242495425 creator A5026393182 @default.
• W4242495425 creator A5032238070 @default.
• W4242495425 creator A5038226209 @default.
• W4242495425 creator A5047643363 @default.
• W4242495425 creator A5052810907 @default.
• W4242495425 creator A5065766721 @default.
• W4242495425 creator A5083700279 @default.
• W4242495425 date "2017-06-24" @default.
• W4242495425 modified "2023-09-29" @default.
• W4242495425 title "ScaleDeep" @default.
• W4242495425 cites W1884620995 @default.
• W4242495425 cites W1980446076 @default.
• W4242495425 cites W1998917233 @default.
• W4242495425 cites W2043607059 @default.
• W4242495425 cites W2044535169 @default.
• W4242495425 cites W2048266589 @default.
• W4242495425 cites W2057434193 @default.
• W4242495425 cites W2060969833 @default.
• W4242495425 cites W2097117768 @default.
• W4242495425 cites W2108598243 @default.
• W4242495425 cites W2117539524 @default.
• W4242495425 cites W2139501017 @default.
• W4242495425 cites W2152839228 @default.
• W4242495425 cites W2293746900 @default.
• W4242495425 cites W2606722458 @default.
• W4242495425 cites W2950656546 @default.
• W4242495425 cites W3004171485 @default.
• W4242495425 cites W4243519499 @default.
• W4242495425 cites W4245199738 @default.
• W4242495425 cites W4251575795 @default.
• W4242495425 doi "https://doi.org/10.1145/3140659.3080244" @default.
• W4242495425 hasPublicationYear "2017" @default.
• W4242495425 type Work @default.
• W4242495425 citedByCount "19" @default.
• W4242495425 countsByYear W42424954252019 @default.
• W4242495425 countsByYear W42424954252020 @default.
• W4242495425 countsByYear W42424954252021 @default.
• W4242495425 countsByYear W42424954252023 @default.
• W4242495425 crossrefType "journal-article" @default.
• W4242495425 hasAuthorship W4242495425A5010094713 @default.
• W4242495425 hasAuthorship W4242495425A5013922094 @default.
• W4242495425 hasAuthorship W4242495425A5018287086 @default.
• W4242495425 hasAuthorship W4242495425A5021152315 @default.
• W4242495425 hasAuthorship W4242495425A5026393182 @default.
• W4242495425 hasAuthorship W4242495425A5032238070 @default.
• W4242495425 hasAuthorship W4242495425A5038226209 @default.
• W4242495425 hasAuthorship W4242495425A5047643363 @default.
• W4242495425 hasAuthorship W4242495425A5052810907 @default.
• W4242495425 hasAuthorship W4242495425A5065766721 @default.
• W4242495425 hasAuthorship W4242495425A5083700279 @default.
• W4242495425 hasConcept C111919701 @default.
• W4242495425 hasConcept C113775141 @default.
• W4242495425 hasConcept C118524514 @default.
• W4242495425 hasConcept C120314980 @default.
• W4242495425 hasConcept C153083717 @default.
• W4242495425 hasConcept C154945302 @default.
• W4242495425 hasConcept C173608175 @default.
• W4242495425 hasConcept C26517878 @default.
• W4242495425 hasConcept C41008148 @default.
• W4242495425 hasConcept C48044578 @default.
• W4242495425 hasConceptScore W4242495425C111919701 @default.
• W4242495425 hasConceptScore W4242495425C113775141 @default.
• W4242495425 hasConceptScore W4242495425C118524514 @default.
• W4242495425 hasConceptScore W4242495425C120314980 @default.
• W4242495425 hasConceptScore W4242495425C153083717 @default.
• W4242495425 hasConceptScore W4242495425C154945302 @default.
• W4242495425 hasConceptScore W4242495425C173608175 @default.
• W4242495425 hasConceptScore W4242495425C26517878 @default.
• W4242495425 hasConceptScore W4242495425C41008148 @default.
• W4242495425 hasConceptScore W4242495425C48044578 @default.
• W4242495425 hasIssue "2" @default.
• W4242495425 hasLocation W42424954251 @default.
• W4242495425 hasOpenAccess W4242495425 @default.
• W4242495425 hasPrimaryLocation W42424954251 @default.
• W4242495425 hasRelatedWork W1569389315 @default.
• W4242495425 hasRelatedWork W1596201972 @default.
• W4242495425 hasRelatedWork W1604898313 @default.
• W4242495425 hasRelatedWork W1767718647 @default.
• W4242495425 hasRelatedWork W1788737569 @default.
• W4242495425 hasRelatedWork W1967954938 @default.
• W4242495425 hasRelatedWork W1986253068 @default.
• W4242495425 hasRelatedWork W2364921833 @default.
• W4242495425 hasRelatedWork W2385146268 @default.
• W4242495425 hasRelatedWork W2503642292 @default.
• W4242495425 hasVolume "45" @default.
• W4242495425 isParatext "false" @default.
• W4242495425 isRetracted "false" @default.
• W4242495425 workType "article" @default.