Matches in SemOpenAlex for { <https://semopenalex.org/work/W3178289424> ?p ?o ?g. }
- W3178289424 endingPage "21" @default.
- W3178289424 startingPage "1" @default.
- W3178289424 abstract "Approximate computing has emerged as a promising design alternative for delivering power-efficient systems and circuits by exploiting the inherent error resiliency of numerous applications. The current article aims to tackle the increased hardware cost of floating-point multiplication units, which prohibits their usage in embedded computing. We introduce AFMU (Approximate Floating-point MUltiplier), an area/power-efficient family of multipliers, which apply two approximation techniques in the resource-hungry mantissa multiplication and can be seamlessly extended to support dynamic configuration of the approximation levels via gating signals. AFMU offers large accuracy configuration margins, provides negligible logic overhead for dynamic configuration, and detects unexpected results that may arise due to the approximations. Our evaluation shows that AFMU delivers energy gains in the range 3.6%–53.5% for half-precision and 37.2%–82.4% for single-precision, in exchange for mean relative error around 0.05%–3.33% and 0.01%–2.20%, respectively. In comparison with state-of-the-art multipliers, AFMU exhibits up to 4–6× smaller error on average while delivering more energy-efficient computing. The evaluation in image processing shows that AFMU provides sufficient quality of service, i.e., more than 50 db PSNR and near 1 SSIM values, and up to 57.4% power reduction. When used in floating-point CNNs, the accuracy loss is small (or zero), i.e., up to 5.4% for MNIST and CIFAR-10, in exchange for up to 63.8% power gain." @default.
- W3178289424 created "2021-07-19" @default.
- W3178289424 creator A5021555896 @default.
- W3178289424 creator A5042122147 @default.
- W3178289424 creator A5043131021 @default.
- W3178289424 creator A5048723394 @default.
- W3178289424 creator A5088085592 @default.
- W3178289424 date "2021-07-09" @default.
- W3178289424 modified "2023-10-09" @default.
- W3178289424 title "Improving Power of DSP and CNN Hardware Accelerators Using Approximate Floating-point Multipliers" @default.
- W3178289424 cites W1491307824 @default.
- W3178289424 cites W2005652364 @default.
- W3178289424 cites W2006312753 @default.
- W3178289424 cites W2020217519 @default.
- W3178289424 cites W2026005150 @default.
- W3178289424 cites W2035378788 @default.
- W3178289424 cites W2057434193 @default.
- W3178289424 cites W2065359960 @default.
- W3178289424 cites W2080592089 @default.
- W3178289424 cites W2098404312 @default.
- W3178289424 cites W2106484393 @default.
- W3178289424 cites W2124651102 @default.
- W3178289424 cites W2187230075 @default.
- W3178289424 cites W2265166184 @default.
- W3178289424 cites W2399535694 @default.
- W3178289424 cites W2461274627 @default.
- W3178289424 cites W2577531088 @default.
- W3178289424 cites W2589986623 @default.
- W3178289424 cites W2612315906 @default.
- W3178289424 cites W2625483679 @default.
- W3178289424 cites W2742536119 @default.
- W3178289424 cites W2769672439 @default.
- W3178289424 cites W2884822385 @default.
- W3178289424 cites W2885108132 @default.
- W3178289424 cites W2908518198 @default.
- W3178289424 cites W2914247084 @default.
- W3178289424 cites W2934610915 @default.
- W3178289424 cites W2945030472 @default.
- W3178289424 cites W2946778009 @default.
- W3178289424 cites W2954748826 @default.
- W3178289424 cites W3009178428 @default.
- W3178289424 cites W3116537075 @default.
- W3178289424 cites W4234461763 @default.
- W3178289424 cites W4249697915 @default.
- W3178289424 cites W4256586690 @default.
- W3178289424 cites W4362215237 @default.
- W3178289424 doi "https://doi.org/10.1145/3448980" @default.
- W3178289424 hasPublicationYear "2021" @default.
- W3178289424 type Work @default.
- W3178289424 sameAs 3178289424 @default.
- W3178289424 citedByCount "10" @default.
- W3178289424 countsByYear W31782894242021 @default.
- W3178289424 countsByYear W31782894242022 @default.
- W3178289424 countsByYear W31782894242023 @default.
- W3178289424 crossrefType "journal-article" @default.
- W3178289424 hasAuthorship W3178289424A5021555896 @default.
- W3178289424 hasAuthorship W3178289424A5042122147 @default.
- W3178289424 hasAuthorship W3178289424A5043131021 @default.
- W3178289424 hasAuthorship W3178289424A5048723394 @default.
- W3178289424 hasAuthorship W3178289424A5088085592 @default.
- W3178289424 hasConcept C108583219 @default.
- W3178289424 hasConcept C111919701 @default.
- W3178289424 hasConcept C11413529 @default.
- W3178289424 hasConcept C122383733 @default.
- W3178289424 hasConcept C124584101 @default.
- W3178289424 hasConcept C139719470 @default.
- W3178289424 hasConcept C154945302 @default.
- W3178289424 hasConcept C162324750 @default.
- W3178289424 hasConcept C190502265 @default.
- W3178289424 hasConcept C2779960059 @default.
- W3178289424 hasConcept C41008148 @default.
- W3178289424 hasConcept C84211073 @default.
- W3178289424 hasConcept C84462506 @default.
- W3178289424 hasConcept C9390403 @default.
- W3178289424 hasConceptScore W3178289424C108583219 @default.
- W3178289424 hasConceptScore W3178289424C111919701 @default.
- W3178289424 hasConceptScore W3178289424C11413529 @default.
- W3178289424 hasConceptScore W3178289424C122383733 @default.
- W3178289424 hasConceptScore W3178289424C124584101 @default.
- W3178289424 hasConceptScore W3178289424C139719470 @default.
- W3178289424 hasConceptScore W3178289424C154945302 @default.
- W3178289424 hasConceptScore W3178289424C162324750 @default.
- W3178289424 hasConceptScore W3178289424C190502265 @default.
- W3178289424 hasConceptScore W3178289424C2779960059 @default.
- W3178289424 hasConceptScore W3178289424C41008148 @default.
- W3178289424 hasConceptScore W3178289424C84211073 @default.
- W3178289424 hasConceptScore W3178289424C84462506 @default.
- W3178289424 hasConceptScore W3178289424C9390403 @default.
- W3178289424 hasIssue "5" @default.
- W3178289424 hasLocation W31782894241 @default.
- W3178289424 hasOpenAccess W3178289424 @default.
- W3178289424 hasPrimaryLocation W31782894241 @default.
- W3178289424 hasRelatedWork W1642189246 @default.
- W3178289424 hasRelatedWork W2045121105 @default.
- W3178289424 hasRelatedWork W2111937329 @default.
- W3178289424 hasRelatedWork W2182869718 @default.
- W3178289424 hasRelatedWork W2615104581 @default.
- W3178289424 hasRelatedWork W2916088568 @default.