{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,25]],"date-time":"2026-05-25T20:06:36Z","timestamp":1779739596656,"version":"3.53.1"},"reference-count":10,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"1","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"name":"Institute for Information & communications Technology Planning","award":["RS-2024-00347394"],"award-info":[{"award-number":["RS-2024-00347394"]}]},{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Republic of Korea","award":["RS-2024-00456287"],"award-info":[{"award-number":["RS-2024-00456287"]}]},{"name":"Global Scholars Invitation Program"},{"name":"Samsung Electronics Coompany","award":["IO251223-14975-01"],"award-info":[{"award-number":["IO251223-14975-01"]}]},{"name":"ICT at Seoul National University"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Comput. Arch. Lett."],"published-print":{"date-parts":[[2026,1]]},"DOI":"10.1109\/lca.2026.3689575","type":"journal-article","created":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T19:34:47Z","timestamp":1777923287000},"page":"174-177","source":"Crossref","is-referenced-by-count":0,"title":["KiF: Accelerating Low-Batch LLM Inference Using In-Flash KV Cache"],"prefix":"10.1109","volume":"25","author":[{"ORCID":"https:\/\/orcid.org\/0009-0003-7359-2676","authenticated-orcid":false,"given":"Haeeun","family":"Jung","sequence":"first","affiliation":[{"name":"Department of Computer Science and Engineering, Seoul National University, Seoul, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2724-8888","authenticated-orcid":false,"given":"Jaeyong","family":"Lee","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, Seoul National University, Seoul, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sanggu","family":"Lee","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, Seoul National University, Seoul, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-2549-3364","authenticated-orcid":false,"given":"Huiwon","family":"Yun","sequence":"additional","affiliation":[{"name":"Graduate School of Convergence Science and Technology, Seoul National University, Seoul, South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0395-8076","authenticated-orcid":false,"given":"Dongsuk","family":"Jeon","sequence":"additional","affiliation":[{"name":"Graduate School of Convergence Science and Technology, Seoul National University, Seoul, South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0442-5634","authenticated-orcid":false,"given":"Nam Sung","family":"Kim","sequence":"additional","affiliation":[{"name":"Department of Electrical & Computer Engineering, University of Illinois, Urbana-Champaign, IL, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7977-9883","authenticated-orcid":false,"given":"Jihong","family":"Kim","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, Seoul National University, Seoul, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1145\/3695053.3731008"},{"key":"ref2","first-page":"31094","article-title":"FlexGen: High-throughput generative inference of large language models with a single GPU","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Sheng","year":"2023"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA61900.2025.00113"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/3695053.3731073"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO61859.2024.00108"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD66269.2025.11240826"},{"key":"ref7","first-page":"379","article-title":"NVMeVirt: A Versatile software-defined virtual NVMe device","volume-title":"Proc. 21st USENIX Conf. File Storage Technol.","author":"Kim","year":"2023"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1145\/3725843.3756134"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO56248.2022.00069"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA61900.2025.00128"}],"container-title":["IEEE Computer Architecture Letters"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/10208\/11321093\/11503412.pdf?arnumber=11503412","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,25]],"date-time":"2026-05-25T19:52:55Z","timestamp":1779738775000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11503412\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,1]]},"references-count":10,"journal-issue":{"issue":"1"},"URL":"https:\/\/doi.org\/10.1109\/lca.2026.3689575","relation":{},"ISSN":["1556-6056","1556-6064","2473-2575"],"issn-type":[{"value":"1556-6056","type":"print"},{"value":"1556-6064","type":"electronic"},{"value":"2473-2575","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,1]]}}}