{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,19]],"date-time":"2026-06-19T05:52:55Z","timestamp":1781848375372,"version":"3.54.5"},"reference-count":26,"publisher":"IEEE","license":[{"start":{"date-parts":[[2026,5,24]],"date-time":"2026-05-24T00:00:00Z","timestamp":1779580800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,5,24]],"date-time":"2026-05-24T00:00:00Z","timestamp":1779580800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001352","name":"National University of Singapore","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100001352","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001321","name":"National Research Foundation","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100001321","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,5,24]]},"DOI":"10.1109\/iscas66217.2026.11562728","type":"proceedings-article","created":{"date-parts":[[2026,6,18]],"date-time":"2026-06-18T20:06:41Z","timestamp":1781813201000},"page":"190-194","source":"Crossref","is-referenced-by-count":0,"title":["LOKI: An LLM Accelerator with Optimized KV Cache and In-Memory Computing Cores"],"prefix":"10.1109","author":[{"given":"Avadh","family":"Harkishanka","sequence":"first","affiliation":[{"name":"National University of Singapore,Department of Electrical Engineering,Singapore"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Abhishek","family":"Tyagi","sequence":"additional","affiliation":[{"name":"National University of Singapore,Department of Electrical Engineering,Singapore"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xuanyao","family":"Fong","sequence":"additional","affiliation":[{"name":"National University of Singapore,Department of Electrical Engineering,Singapore"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Large language models for code summarization","author":"Szalontai","year":"2024"},{"key":"ref2","first-page":"2764","article-title":"On context utilization in summarization with large language models","volume-title":"Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","author":"Ravaut"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1038\/s43588-025-00854-1"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/3649329.3657327"},{"key":"ref5","first-page":"1","article-title":"Retransformer: Reram-based processing-in-memory architecture for transformer acceleration","volume-title":"2020 IEEE\/ACM International Conference On Computer Aided Design (ICCAD)","author":"Yang"},{"key":"ref6","first-page":"897","volume-title":"POD-Attention: Unlocking Full Prefill-Decode Overlap for Faster LLM Inference","author":"Kamath","year":"2025"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD66269.2025.11240722"},{"key":"ref8","article-title":"Hpim: Heterogeneous processing-in-memory-based accelerator for large language models inference","author":"Duan","year":"2025"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/SOCC62300.2024.10737844"},{"key":"ref10","article-title":"Inference-time hyper-scaling with kv cache compression","author":"\u0141a\u0144cucki","year":"2025"},{"key":"ref11","doi-asserted-by":"crossref","DOI":"10.18653\/v1\/2024.findings-acl.195","article-title":"Pyramidinfer: Pyramid kv cache compression for high-throughput llm inference","author":"Yang","year":"2024"},{"key":"ref12","article-title":"Lazyllm: Dynamic token pruning for efficient long context llm inference","author":"Fu","year":"2024"},{"key":"ref13","article-title":"Aqua: Attention via query magnitudes for memory and compute efficient inference in llms","author":"S","year":"2025"},{"key":"ref14","article-title":"Dialogue without limits: Constant-sized kv caches for extended responses in llms","author":"Ghadia","year":"2025"},{"key":"ref15","doi-asserted-by":"crossref","DOI":"10.1109\/DAC63849.2025.11133273","article-title":"Unicaim: A unified cam\/cim architecture with static-dynamic kv cache pruning for efficient long-context llm inference","author":"Xu","year":"2025"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2024.3402174"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1145\/3716368.3735145"},{"key":"ref18","doi-asserted-by":"crossref","DOI":"10.52202\/068431-1189","article-title":"Flashattention: Fast and memory-efficient exact attention with io-awareness","author":"Dao","year":"2022"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/MSE.2007.44"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO.2007.33"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/isca.2008.16"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/ISVLSI.2014.94"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC42613.2021.9365803"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/ISCAS.2019.8702715"},{"key":"ref25","article-title":"Pointer sentinel mixture models","author":"Merity","year":"2016"},{"key":"ref26","article-title":"Llama: Open and efficient foundation language models","author":"Touvron","year":"2023"}],"event":{"name":"2026 IEEE International Symposium on Circuits and Systems (ISCAS)","location":"Shanghai, China","start":{"date-parts":[[2026,5,24]]},"end":{"date-parts":[[2026,5,28]]}},"container-title":["2026 IEEE International Symposium on Circuits and Systems (ISCAS)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11561899\/11561804\/11562728.pdf?arnumber=11562728","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,19]],"date-time":"2026-06-19T05:28:33Z","timestamp":1781846913000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11562728\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,24]]},"references-count":26,"URL":"https:\/\/doi.org\/10.1109\/iscas66217.2026.11562728","relation":{},"subject":[],"published":{"date-parts":[[2026,5,24]]}}}