{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,2]],"date-time":"2026-07-02T23:38:14Z","timestamp":1783035494411,"version":"3.54.6"},"reference-count":49,"publisher":"Association for Computing Machinery (ACM)","issue":"5","license":[{"start":{"date-parts":[[2021,7,9]],"date-time":"2021-07-09T00:00:00Z","timestamp":1625788800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Embed. Comput. Syst."],"published-print":{"date-parts":[[2021,9,30]]},"abstract":"<jats:p>\n            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%\u201353.5% for half-precision and 37.2%\u201382.4% for single-precision, in exchange for mean relative error around 0.05%\u20133.33% and 0.01%\u20132.20%, respectively. In comparison with state-of-the-art multipliers, AFMU exhibits up to 4\u20136\u00d7 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\n            <jats:italic>db<\/jats:italic>\n            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.\n          <\/jats:p>","DOI":"10.1145\/3448980","type":"journal-article","created":{"date-parts":[[2021,7,9]],"date-time":"2021-07-09T15:04:44Z","timestamp":1625843084000},"page":"1-21","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":28,"title":["Improving Power of DSP and CNN Hardware Accelerators Using Approximate Floating-point Multipliers"],"prefix":"10.1145","volume":"20","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0503-8246","authenticated-orcid":false,"given":"Vasileios","family":"Leon","sequence":"first","affiliation":[{"name":"National Technical University of Athens, Athens, Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Theodora","family":"Paparouni","sequence":"additional","affiliation":[{"name":"National Technical University of Athens, Athens, Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Evangelos","family":"Petrongonas","sequence":"additional","affiliation":[{"name":"National Technical University of Athens, Athens, Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dimitrios","family":"Soudris","sequence":"additional","affiliation":[{"name":"National Technical University of Athens, Athens, Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kiamal","family":"Pekmestzi","sequence":"additional","affiliation":[{"name":"National Technical University of Athens, Athens, Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2021,7,9]]},"reference":[{"key":"e_1_2_1_1_1","volume-title":"IEEE standard for floating-point arithmetic","author":"IEEE.","year":"2019","unstructured":"IEEE. 2019. IEEE standard for floating-point arithmetic . IEEE Std 754- 2019 (Revision of IEEE 754-2008) (July 2019), 1\u201384. https:\/\/doi.org\/10.1109\/IEEESTD.2019.8766229 10.1109\/IEEESTD.2019.8766229 IEEE. 2019. IEEE standard for floating-point arithmetic. IEEE Std 754-2019 (Revision of IEEE 754-2008) (July 2019), 1\u201384. https:\/\/doi.org\/10.1109\/IEEESTD.2019.8766229"},{"key":"e_1_2_1_2_1","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2016.2643003"},{"key":"e_1_2_1_3_1","volume-title":"the Design Automation Conference (DAC\u201910)","author":"Srimat","year":"1837","unstructured":"Srimat T. Chakradhar and Anand Raghunathan. 2010. Best-effort computing: Re-thinking parallel software and hardware . In the Design Automation Conference (DAC\u201910) . 865\u2013870. https:\/\/doi.org\/10.1145\/ 1837 274.1837492 10.1145\/1837274.1837492 Srimat T. Chakradhar and Anand Raghunathan. 2010. Best-effort computing: Re-thinking parallel software and hardware. In the Design Automation Conference (DAC\u201910). 865\u2013870. https:\/\/doi.org\/10.1145\/1837274.1837492"},{"key":"e_1_2_1_4_1","doi-asserted-by":"publisher","DOI":"10.1109\/IISWC.2009.5306797"},{"key":"e_1_2_1_5_1","doi-asserted-by":"publisher","DOI":"10.1145\/2463209.2488873"},{"key":"e_1_2_1_6_1","doi-asserted-by":"publisher","DOI":"10.1145\/2000064.2000108"},{"key":"e_1_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO.2012.48"},{"key":"e_1_2_1_8_1","doi-asserted-by":"publisher","DOI":"10.1109\/ETS.2013.6569370"},{"key":"e_1_2_1_9_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD.2015.7372600"},{"key":"e_1_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1145\/3218603.3218621"},{"key":"e_1_2_1_11_1","doi-asserted-by":"publisher","DOI":"10.1145\/3061639.3062210"},{"key":"e_1_2_1_12_1","doi-asserted-by":"publisher","DOI":"10.1145\/3316781.3317774"},{"key":"e_1_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.1145\/2742060.2743760"},{"key":"e_1_2_1_14_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2015.2493547"},{"key":"e_1_2_1_15_1","doi-asserted-by":"publisher","DOI":"10.1145\/3094124"},{"key":"e_1_2_1_16_1","first-page":"1","article-title":"Low-power approximate unsigned multipliers with configurable error recovery","volume":"66","author":"Jiang Honglan","year":"2019","unstructured":"Honglan Jiang , Cong Liu , Fabrizio Lombardi , and Jie Han . 2019 . Low-power approximate unsigned multipliers with configurable error recovery . IEEE Trans. Circ. Syst. I: Reg. Papers 66 , 1 (Jan. 2019), 189\u2013202. https:\/\/doi.org\/10.1109\/TCSI.2018.2856245 10.1109\/TCSI.2018.2856245 Honglan Jiang, Cong Liu, Fabrizio Lombardi, and Jie Han. 2019. Low-power approximate unsigned multipliers with configurable error recovery. IEEE Trans. Circ. Syst. I: Reg. Papers 66, 1 (Jan. 2019), 189\u2013202. https:\/\/doi.org\/10.1109\/TCSI.2018.2856245","journal-title":"IEEE Trans. Circ. Syst. I: Reg. Papers"},{"key":"e_1_2_1_17_1","volume-title":"the Conference on Design, Automation & Test in Europe (DATE\u201917)","author":"Jiao Xun","year":"2017","unstructured":"Xun Jiao , Vincent Camus , Mattia Cacciotti , Yu Jiang , Christian Enz , and Rajesh K. Gupta . 2017. Combining structural and timing errors in overclocked inexact speculative adders . In the Conference on Design, Automation & Test in Europe (DATE\u201917) . 482\u2013487. https:\/\/doi.org\/10.23919\/DATE. 2017 .7927037 10.23919\/DATE.2017.7927037 Xun Jiao, Vincent Camus, Mattia Cacciotti, Yu Jiang, Christian Enz, and Rajesh K. Gupta. 2017. Combining structural and timing errors in overclocked inexact speculative adders. In the Conference on Design, Automation & Test in Europe (DATE\u201917). 482\u2013487. https:\/\/doi.org\/10.23919\/DATE.2017.7927037"},{"key":"e_1_2_1_18_1","volume-title":"the Conference on Design, Automation & Test in Europe (DATE\u201917)","author":"Jiao Xun","year":"2017","unstructured":"Xun Jiao , Yu Jiang , Abbas Rahimi , and Rajesh K. Gupta . 2017. SLoT: A supervised learning model to predict dynamic timing errors of functional units . In the Conference on Design, Automation & Test in Europe (DATE\u201917) . 1183\u20131188. https:\/\/doi.org\/10.23919\/DATE. 2017 .7927168 10.23919\/DATE.2017.7927168 Xun Jiao, Yu Jiang, Abbas Rahimi, and Rajesh K. Gupta. 2017. SLoT: A supervised learning model to predict dynamic timing errors of functional units. In the Conference on Design, Automation & Test in Europe (DATE\u201917). 1183\u20131188. https:\/\/doi.org\/10.23919\/DATE.2017.7927168"},{"key":"e_1_2_1_19_1","volume-title":"the Design Automation Conference (DAC\u201912)","author":"Andrew","unstructured":"Andrew B. Kahng and Seokhyeong Kang. 2012. Accuracy-configurable adder for approximate arithmetic designs . In the Design Automation Conference (DAC\u201912) . 820\u2013825. https:\/\/doi.org\/10.1145\/2228360.2228509 10.1145\/2228360.2228509 Andrew B. Kahng and Seokhyeong Kang. 2012. Accuracy-configurable adder for approximate arithmetic designs. In the Design Automation Conference (DAC\u201912). 820\u2013825. https:\/\/doi.org\/10.1145\/2228360.2228509"},{"key":"e_1_2_1_20_1","volume-title":"the Conference on Design, Automation & Test in Europe (DATE\u201914)","author":"Kamal Mehdi","year":"2014","unstructured":"Mehdi Kamal , Amin Ghasemazar , Ali Afzali-Kusha , and Massoud Pedram . 2014 . Improving efficiency of extensible processors by using approximate custom instructions . In the Conference on Design, Automation & Test in Europe (DATE\u201914) . 1\u20134. https:\/\/doi.org\/10.7873\/DATE.2014.238 10.7873\/DATE.2014.238 Mehdi Kamal, Amin Ghasemazar, Ali Afzali-Kusha, and Massoud Pedram. 2014. Improving efficiency of extensible processors by using approximate custom instructions. In the Conference on Design, Automation & Test in Europe (DATE\u201914). 1\u20134. https:\/\/doi.org\/10.7873\/DATE.2014.238"},{"key":"e_1_2_1_21_1","volume-title":"the International Conference on Neural Information Processing Systems (NIPS\u201917)","author":"K\u00f6ster Urs","year":"2017","unstructured":"Urs K\u00f6ster , Tristan J. Webb , Xin Wang , Marcel Nassar , Arjun K. Bansal , William H. Constable , O\u011fuz H. Elibol , Scott Gray , Stewart Hall , Luke Hornof , Amir Khosrowshahi , Carey Kloss , Ruby J. Pai , and Naveen Rao . 2017 . Flexpoint: An adaptive numerical format for efficient training of deep neural networks . In the International Conference on Neural Information Processing Systems (NIPS\u201917) . 1742\u20131752. Urs K\u00f6ster, Tristan J. Webb, Xin Wang, Marcel Nassar, Arjun K. Bansal, William H. Constable, O\u011fuz H. Elibol, Scott Gray, Stewart Hall, Luke Hornof, Amir Khosrowshahi, Carey Kloss, Ruby J. Pai, and Naveen Rao. 2017. Flexpoint: An adaptive numerical format for efficient training of deep neural networks. In the International Conference on Neural Information Processing Systems (NIPS\u201917). 1742\u20131752."},{"key":"e_1_2_1_22_1","unstructured":"Alex Krizhevsky Vinod Nair and Geoffrey Hinton. 2020. CIFAR-10 Dataset. Retrieved from https:\/\/www.cs.toronto.edu\/kriz\/cifar.html.  Alex Krizhevsky Vinod Nair and Geoffrey Hinton. 2020. CIFAR-10 Dataset. Retrieved from https:\/\/www.cs.toronto.edu\/kriz\/cifar.html."},{"key":"e_1_2_1_23_1","doi-asserted-by":"publisher","DOI":"10.1109\/VLSID.2011.51"},{"key":"e_1_2_1_24_1","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2009.2016665"},{"key":"e_1_2_1_25_1","doi-asserted-by":"publisher","DOI":"10.1109\/ARITH.2015.18"},{"key":"e_1_2_1_26_1","unstructured":"Yann LeCun Corinna Cortes and C. J. Burges. 2020. MNIST Handwritten Digit Database. Retrieved from http:\/\/yann.lecun.com\/exdb\/mnist\/.  Yann LeCun Corinna Cortes and C. J. Burges. 2020. MNIST Handwritten Digit Database. Retrieved from http:\/\/yann.lecun.com\/exdb\/mnist\/."},{"key":"e_1_2_1_27_1","doi-asserted-by":"publisher","DOI":"10.1145\/3316781.3317793"},{"key":"e_1_2_1_28_1","doi-asserted-by":"publisher","DOI":"10.1049\/iet-cds.2018.5039"},{"key":"e_1_2_1_29_1","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2017.2767858"},{"key":"e_1_2_1_30_1","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2018.043191124"},{"key":"e_1_2_1_31_1","volume-title":"the Conference on Design, Automation & Test in Europe (DATE\u201914)","author":"Liu Cong","year":"2014","unstructured":"Cong Liu , Jie Han , and Fabrizio Lombardi . 2014 . A low-power, high-performance approximate multiplier with configurable partial error recovery . In the Conference on Design, Automation & Test in Europe (DATE\u201914) . 1\u20134. https:\/\/doi.org\/10.7873\/DATE.2014.108 10.7873\/DATE.2014.108 Cong Liu, Jie Han, and Fabrizio Lombardi. 2014. A low-power, high-performance approximate multiplier with configurable partial error recovery. In the Conference on Design, Automation & Test in Europe (DATE\u201914). 1\u20134. https:\/\/doi.org\/10.7873\/DATE.2014.108"},{"key":"e_1_2_1_32_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2018.2890222"},{"key":"e_1_2_1_33_1","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2020.2975695"},{"key":"e_1_2_1_34_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2017.2672976"},{"key":"e_1_2_1_35_1","doi-asserted-by":"publisher","DOI":"10.1145\/2893356"},{"key":"e_1_2_1_36_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2014.2308214"},{"key":"e_1_2_1_37_1","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2014.2333366"},{"key":"e_1_2_1_38_1","doi-asserted-by":"publisher","DOI":"10.23919\/DATE.2017.7927036"},{"key":"e_1_2_1_39_1","doi-asserted-by":"publisher","DOI":"10.1145\/2744769.2744778"},{"key":"e_1_2_1_40_1","doi-asserted-by":"publisher","DOI":"10.1145\/2897937.2906199"},{"key":"e_1_2_1_41_1","volume-title":"the Conference on Design, Automation & Test in Europe (DATE\u201910)","author":"Shin Doochul","year":"2010","unstructured":"Doochul Shin and Sandeep K. Gupta . 2010. Approximate logic synthesis for error tolerant applications . In the Conference on Design, Automation & Test in Europe (DATE\u201910) . 957\u2013960. https:\/\/doi.org\/10.1109\/DATE. 2010 .5456913 10.1109\/DATE.2010.5456913 Doochul Shin and Sandeep K. Gupta. 2010. Approximate logic synthesis for error tolerant applications. In the Conference on Design, Automation & Test in Europe (DATE\u201910). 957\u2013960. https:\/\/doi.org\/10.1109\/DATE.2010.5456913"},{"key":"e_1_2_1_42_1","first-page":"3","article-title":"Reducing power by optimizing the necessary precision\/range of floating-point arithmetic","volume":"8","author":"Fai Tong Jonathan Ying","year":"2000","unstructured":"Jonathan Ying Fai Tong , David Nagle , and Rob A. Rutenbar . 2000 . Reducing power by optimizing the necessary precision\/range of floating-point arithmetic . IEEE Trans. Very Large Scale Integ. Syst. 8 , 3 (June 2000), 273\u2013286. https:\/\/doi.org\/10.1109\/92.845894 10.1109\/92.845894 Jonathan Ying Fai Tong, David Nagle, and Rob A. Rutenbar. 2000. Reducing power by optimizing the necessary precision\/range of floating-point arithmetic. IEEE Trans. Very Large Scale Integ. Syst. 8, 3 (June 2000), 273\u2013286. https:\/\/doi.org\/10.1109\/92.845894","journal-title":"IEEE Trans. Very Large Scale Integ. Syst."},{"key":"e_1_2_1_43_1","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2018.2890712"},{"key":"e_1_2_1_44_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2019.2926275"},{"key":"e_1_2_1_45_1","doi-asserted-by":"publisher","DOI":"10.1145\/2540708.2540710"},{"key":"e_1_2_1_46_1","doi-asserted-by":"publisher","DOI":"10.7873\/DATE.2015.0513"},{"key":"e_1_2_1_47_1","first-page":"4","article-title":"Designs of approximate floating-point multipliers with variable accuracy for error-tolerant applications","volume":"90","author":"Yin Peipei","year":"2018","unstructured":"Peipei Yin , Chenghua Wang , Weiqiang Liu , Earl E. Swartzlander , and Fabrizio Lombardi . 2018 . Designs of approximate floating-point multipliers with variable accuracy for error-tolerant applications . J. Sig. Proc. Syst. 90 , 4 (Apr. 2018), 641\u2013654. https:\/\/doi.org\/10.1007\/s11265-017-1280-4 10.1007\/s11265-017-1280-4 Peipei Yin, Chenghua Wang, Weiqiang Liu, Earl E. Swartzlander, and Fabrizio Lombardi. 2018. Designs of approximate floating-point multipliers with variable accuracy for error-tolerant applications. J. Sig. Proc. Syst. 90, 4 (Apr. 2018), 641\u2013654. https:\/\/doi.org\/10.1007\/s11265-017-1280-4","journal-title":"J. Sig. Proc. Syst."},{"key":"e_1_2_1_48_1","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2016.2587696"},{"key":"e_1_2_1_49_1","doi-asserted-by":"publisher","DOI":"10.18130\/V3QM2S"}],"container-title":["ACM Transactions on Embedded Computing Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3448980","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3448980","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,17]],"date-time":"2025-06-17T20:47:55Z","timestamp":1750193275000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3448980"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,7,9]]},"references-count":49,"journal-issue":{"issue":"5","published-print":{"date-parts":[[2021,9,30]]}},"alternative-id":["10.1145\/3448980"],"URL":"https:\/\/doi.org\/10.1145\/3448980","relation":{},"ISSN":["1539-9087","1558-3465"],"issn-type":[{"value":"1539-9087","type":"print"},{"value":"1558-3465","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,7,9]]},"assertion":[{"value":"2020-07-01","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2021-02-01","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2021-07-09","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}