{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,5]],"date-time":"2026-05-05T12:31:50Z","timestamp":1777984310378,"version":"3.51.4"},"reference-count":48,"publisher":"Association for Computing Machinery (ACM)","issue":"4","content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["J. Emerg. Technol. Comput. Syst."],"published-print":{"date-parts":[[2025,10,31]]},"abstract":"\n On-chip training at the edge becomes a primary requisite for real-time and security-sensitive artificial neural network (ANN) applications. In-memory computation (IMC) techniques have been proposed to facilitate data-intensive computational operations in ANNs. IMC-based multiply-accumulate (MAC) accelerates ANN training but suffers from significant communication overhead between the MAC engine and the off-chip storage for the intermediate data. This article proposes an analog temporary on-chip memory (ATOM) to store this intermediate data during ANN training. The ANN training architecture with the proposed ATOM has two significant advantages. First, the energy required to store intermediate data is scaled down by\n \n \\(\\sim\\)<\/jats:tex-math>\n <\/jats:inline-formula>\n 40\n \n \\(\\times\\)<\/jats:tex-math>\n <\/jats:inline-formula>\n due to the on-chip and analog nature of the memory. Second, the proposed architecture avoids power and area-consuming analog-to-digital converters (ADCs) between neural network stages. The ATOM cell measurements are carried out from 20 fabricated chips, and the impact of ATOM characteristics on ANN system performance accuracy is analyzed. This article shows significant latency improvement of\n \n \\(\\sim\\)<\/jats:tex-math>\n <\/jats:inline-formula>\n 9\n \n \\(\\times\\)<\/jats:tex-math>\n <\/jats:inline-formula>\n and area savings of\n \n \\(\\sim\\)<\/jats:tex-math>\n <\/jats:inline-formula>\n 5\n \n \\(\\times\\)<\/jats:tex-math>\n <\/jats:inline-formula>\n for intermediate data storage compared to the on-chip SRAM during ANN training\u2019s forward and backward pass operations. An improvement in the area and latency will be beneficial to instrument the area- and energy-efficient hardware system for on-chip ANN applications.\n <\/jats:p>","DOI":"10.1145\/3765899","type":"journal-article","created":{"date-parts":[[2025,9,5]],"date-time":"2025-09-05T15:10:55Z","timestamp":1757085055000},"page":"1-18","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Analog and Temporary On-chip Memory for ANN Training and Inference"],"prefix":"10.1145","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0086-1276","authenticated-orcid":false,"given":"Shreyas","family":"Deshmukh","sequence":"first","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2104-8068","authenticated-orcid":false,"given":"Raghav","family":"Singhal","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0009-0002-2315-9906","authenticated-orcid":false,"given":"Shruti","family":"Landge","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Indore, Indore, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9191-1632","authenticated-orcid":false,"given":"Vivek","family":"Saraswat","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2363-5864","authenticated-orcid":false,"given":"Anmol","family":"Biswas","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6106-5623","authenticated-orcid":false,"given":"Abhishek","family":"Kadam","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6170-1340","authenticated-orcid":false,"given":"Ajay K.","family":"Singh","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5372-0173","authenticated-orcid":false,"given":"Sreenivas","family":"Subramoney","sequence":"additional","affiliation":[{"name":"Intel Labs, Bengaluru, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4330-3103","authenticated-orcid":false,"given":"Laxmeesha","family":"Somappa","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6568-3736","authenticated-orcid":false,"given":"Maryam Shojaei","family":"Baghini","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1498-5993","authenticated-orcid":false,"given":"Udayan","family":"Ganguly","sequence":"additional","affiliation":[{"name":"Indian Institute of Technology Bombay, Mumbai, India"}]}],"member":"320","published-online":{"date-parts":[[2025,10,9]]},"reference":[{"key":"e_1_3_1_2_2","doi-asserted-by":"publisher","DOI":"10.1145\/3269985"},{"key":"e_1_3_1_3_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2017.7870354"},{"key":"e_1_3_1_4_2","doi-asserted-by":"publisher","DOI":"10.1109\/DCIS53048.2021.9666181"},{"key":"e_1_3_1_5_2","doi-asserted-by":"publisher","DOI":"10.1147\/JRD.2019.2934050"},{"key":"e_1_3_1_6_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.vlsi.2020.03.002"},{"key":"e_1_3_1_7_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2016.2616357"},{"key":"e_1_3_1_8_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC42613.2021.9366045"},{"key":"e_1_3_1_9_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2021.3056447"},{"key":"e_1_3_1_10_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISCAS46773.2023.10181912"},{"key":"e_1_3_1_11_2","doi-asserted-by":"publisher","unstructured":"Anteneh Gebregiorgis Hoang Anh Du Nguyen Jintao Yu Rajendra Bishnoi Mottaqiallah Taouil Francky Catthoor and Said Hamdioui. 2022. A survey on memory-centric computer architectures. Journal on Emerging Technologies in Computing Systems 18 4 (Oct. 2022) Article 79 50 pages. DOI: 10.1145\/3544974","DOI":"10.1145\/3544974"},{"key":"e_1_3_1_12_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2018.8310398"},{"key":"e_1_3_1_13_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2014.6757323"},{"key":"e_1_3_1_14_2","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2019.2929245"},{"key":"e_1_3_1_15_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2020.2987714"},{"key":"e_1_3_1_16_2","doi-asserted-by":"publisher","DOI":"10.1109\/TCSI.2022.3208755"},{"key":"e_1_3_1_17_2","doi-asserted-by":"publisher","DOI":"10.1109\/MDAT.2021.3050715"},{"key":"e_1_3_1_18_2","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2020.2980533"},{"key":"e_1_3_1_19_2","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2024.3487975"},{"key":"e_1_3_1_20_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2021.3061508"},{"key":"e_1_3_1_21_2","doi-asserted-by":"publisher","DOI":"10.1145\/3528104"},{"key":"e_1_3_1_22_2","doi-asserted-by":"publisher","DOI":"10.1145\/3485823"},{"key":"e_1_3_1_23_2","unstructured":"Alex Krizhevsky and Geoffrey Hinton. 2009. Learning Multiple Layers of Features from Tiny Images. Technical Report 0. University of Toronto Toronto Ontario. Retrieved from https:\/\/www.cs.toronto.edu\/\u223ckriz\/learning-features-2009-TR.pdf"},{"key":"e_1_3_1_24_2","article-title":"ImageNet classification with deep convolutional neural networks","volume":"25","author":"Krizhevsky Alex","year":"2012","unstructured":"Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. 2012. ImageNet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems, Vol. 25. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2012\/file\/c399862d3b9d6b76c8436e924a68c45b-Paper.pdf","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_3_1_25_2","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPSW.2019.00148"},{"key":"e_1_3_1_26_2","doi-asserted-by":"publisher","DOI":"10.1109\/TCSII.2023.3329261"},{"key":"e_1_3_1_27_2","doi-asserted-by":"publisher","DOI":"10.1145\/3316781.3317874"},{"key":"e_1_3_1_28_2","unstructured":"Ilya Loshchilov and Frank Hutter. 2017. SGDR: Stochastic gradient descent with warm restarts. arXiv:1608.03983. Retrieved from https:\/\/arxiv.org\/abs\/1608.03983"},{"key":"e_1_3_1_29_2","doi-asserted-by":"publisher","DOI":"10.1145\/3451212"},{"key":"e_1_3_1_30_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.sysarc.2021.102276"},{"key":"e_1_3_1_31_2","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2019.2896880"},{"key":"e_1_3_1_32_2","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2024.3439231"},{"key":"e_1_3_1_33_2","doi-asserted-by":"publisher","DOI":"10.1145\/2996192"},{"key":"e_1_3_1_34_2","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2020.3043731"},{"key":"e_1_3_1_35_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2022.3140753"},{"key":"e_1_3_1_36_2","doi-asserted-by":"publisher","DOI":"10.1145\/3195970.3195998"},{"key":"e_1_3_1_37_2","doi-asserted-by":"publisher","DOI":"10.1109\/IMW48823.2020.9108142"},{"key":"e_1_3_1_38_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41565-020-0655-z"},{"key":"e_1_3_1_39_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA.2016.12"},{"key":"e_1_3_1_40_2","unstructured":"Karen Simonyan and Andrew Zisserman. 2015. Very deep convolutional networks for large-scale image recognition. arXiv:1409.1556. Retrieved from https:\/\/arxiv.org\/abs\/1409.1556"},{"key":"e_1_3_1_41_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2023.07.001"},{"key":"e_1_3_1_42_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2021.3108344"},{"key":"e_1_3_1_43_2","doi-asserted-by":"publisher","DOI":"10.1109\/MSSC.2020.3002140"},{"key":"e_1_3_1_44_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA.2018.00037"},{"key":"e_1_3_1_45_2","doi-asserted-by":"publisher","DOI":"10.1109\/MSSC.2019.2922889"},{"key":"e_1_3_1_46_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2022.3232601"},{"key":"e_1_3_1_47_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2022.3211290"},{"key":"e_1_3_1_48_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2016.2642198"},{"key":"e_1_3_1_49_2","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA57654.2024.00071"}],"container-title":["ACM Journal on Emerging Technologies in Computing Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3765899","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,9]],"date-time":"2025-10-09T11:16:49Z","timestamp":1760008609000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3765899"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,9]]},"references-count":48,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2025,10,31]]}},"alternative-id":["10.1145\/3765899"],"URL":"https:\/\/doi.org\/10.1145\/3765899","relation":{"has-preprint":[{"id-type":"doi","id":"10.36227\/techrxiv.174612846.67587473\/v2","asserted-by":"object"},{"id-type":"doi","id":"10.36227\/techrxiv.174612846.67587473\/v1","asserted-by":"object"}]},"ISSN":["1550-4832","1550-4840"],"issn-type":[{"value":"1550-4832","type":"print"},{"value":"1550-4840","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,10,9]]},"assertion":[{"value":"2024-05-07","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-08-22","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-10-09","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}