DC Field | Value | Language |
---|---|---|
dc.contributor.author | Park, Kangkyu | ko |
dc.contributor.author | Choi, Seungkyu | ko |
dc.contributor.author | Choi, Yeongjae | ko |
dc.contributor.author | Kim, Lee-Sup | ko |
dc.date.accessioned | 2022-04-13T06:50:37Z | - |
dc.date.available | 2022-04-13T06:50:37Z | - |
dc.date.created | 2021-06-15 | - |
dc.date.created | 2021-06-15 | - |
dc.date.created | 2021-06-15 | - |
dc.date.issued | 2022-04 | - |
dc.identifier.citation | IEEE TRANSACTIONS ON COMPUTERS, v.71, no.4, pp.795 - 808 | - |
dc.identifier.issn | 0018-9340 | - |
dc.identifier.uri | http://hdl.handle.net/10203/292582 | - |
dc.description.abstract | Recent research shows that 4-bit data precision is sufficient for Deep Neural Network (DNN) inference without accuracy degradation. Due to the low bit-width, a large amount of data is repeated. In this paper, we propose a hardware architecture, named Rare Computing Architecture (RCA), that skips redundant computations due to the repeated data in the networks. By exploiting redundancy, RCA is not significantly affected by data-sparsity and maintains great improvements in performance and energy efficiency, while the improvements of existing DNN accelerators are vulnerable to variations in sparsity. In the RCA, repeated data in a window for censoring repetition are detected by a Redundancy Censoring Unit (RCU) and processed at a time, achieving high effective throughput. Additionally, we present a dataflow that exploits abundant data-reusability in DNNs, which enables the high-throughput computations to be ceaselessly performed without an increase of bandwidth for data-read. The proposed architecture is evaluated in two ways of exploiting weight- and activation-repetition. In the evaluation, RCA is compared to a value-agnostic computation and UCNN that is the state-of-the-art accelerator exploiting weight-repetition. Additionally, RCA is compared to Bit-pragmatic that exploits bit-level sparsity. Both evaluations demonstrate that the RCA shows steadily high improvements in performance and energy-efficiency. | - |
dc.language | English | - |
dc.publisher | IEEE COMPUTER SOC | - |
dc.title | Rare Computing: Removing Redundant Multiplications from Sparse and Repetitive Data in Deep Neural Networks | - |
dc.type | Article | - |
dc.identifier.wosid | 000767844300005 | - |
dc.identifier.scopusid | 2-s2.0-85102253809 | - |
dc.type.rims | ART | - |
dc.citation.volume | 71 | - |
dc.citation.issue | 4 | - |
dc.citation.beginningpage | 795 | - |
dc.citation.endingpage | 808 | - |
dc.citation.publicationname | IEEE TRANSACTIONS ON COMPUTERS | - |
dc.identifier.doi | 10.1109/TC.2021.3063269 | - |
dc.contributor.localauthor | Kim, Lee-Sup | - |
dc.description.isOpenAccess | N | - |
dc.type.journalArticle | Article | - |
dc.subject.keywordAuthor | Computer architectureHardwareQuantization (signal)Neural networksComputational modelingBandwidthRedundancyDeep neural networksaccelerator architecturehardware acceleration | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.