{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T07:16:22Z","timestamp":1772694982912,"version":"3.50.1"},"reference-count":69,"publisher":"IEEE","license":[{"start":{"date-parts":[[2026,1,31]],"date-time":"2026-01-31T00:00:00Z","timestamp":1769817600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,1,31]],"date-time":"2026-01-31T00:00:00Z","timestamp":1769817600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62302302,62232011"],"award-info":[{"award-number":["62302302,62232011"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100007219","name":"Natural Science Foundation of Shanghai Municipality","doi-asserted-by":"publisher","award":["24ZR1430500"],"award-info":[{"award-number":["24ZR1430500"]}],"id":[{"id":"10.13039\/100007219","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,1,31]]},"DOI":"10.1109\/hpca68181.2026.11408477","type":"proceedings-article","created":{"date-parts":[[2026,3,4]],"date-time":"2026-03-04T20:47:22Z","timestamp":1772657242000},"page":"1-14","source":"Crossref","is-referenced-by-count":0,"title":["LEGO: Supporting LLM-Enhanced Games with One Gaming GPU"],"prefix":"10.1109","author":[{"given":"Han","family":"Zhao","sequence":"first","affiliation":[{"name":"Shanghai Jiao Tong University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Weihao","family":"Cui","sequence":"additional","affiliation":[{"name":"Shanghai Jiao Tong University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zeshen","family":"Zhang","sequence":"additional","affiliation":[{"name":"Tongji University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wenhao","family":"Zhang","sequence":"additional","affiliation":[{"name":"Shanghai Jiao Tong University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jiangtong","family":"Li","sequence":"additional","affiliation":[{"name":"Tongji University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Quan","family":"Chen","sequence":"additional","affiliation":[{"name":"Shanghai Jiao Tong University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pu","family":"Pang","sequence":"additional","affiliation":[{"name":"Shanghai Jiao Tong University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zijun","family":"Li","sequence":"additional","affiliation":[{"name":"Shanghai Jiao Tong University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhenhua","family":"Han","sequence":"additional","affiliation":[{"name":"Shanghai Qiji Zhifeng Co., Ltd."}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yuqing","family":"Yang","sequence":"additional","affiliation":[{"name":"Microsoft Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Minyi","family":"Guo","sequence":"additional","affiliation":[{"name":"Shanghai Jiao Tong University"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","year":"2025","journal-title":"Actions per minute define in lark"},{"key":"ref2","year":"2025","journal-title":"Actions per minute define in wikipedia"},{"key":"ref3","year":"2025","journal-title":"Evaluate 11 ms in real time with street fighter iii"},{"key":"ref4","year":"2025","journal-title":"Mistral-7b and mistral-4b"},{"key":"ref5","year":"2025","journal-title":"Nvidia nsight systems"},{"key":"ref6","author":"A","year":"2024","journal-title":"How to Run Open-Source LLMs Locally with the OpenAI Connector and Ollama in Mendix"},{"key":"ref7","article-title":"Dota 2 with large scale deep reinforcement learning","author":"Berner","year":"2019","journal-title":"arXiv preprint"},{"key":"ref8","article-title":"Fireact: Toward language agent fine-tuning","author":"Chen","year":"2023","journal-title":"arXiv preprint"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1145\/3620678.3624660"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/TCC.2024.3521657"},{"key":"ref11","article-title":"Can vlms play action role-playing games? take black myth wukong as a study case","author":"Chen","year":"2024","journal-title":"arXiv preprint"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/2954679.2872368"},{"key":"ref13","article-title":"Streamlining redundant layers to compress large language models","volume-title":"The Thirteenth International Conference on Learning Representations","author":"Chen","year":"2025"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.52202\/079017-4019"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1145\/3419111.3421284"},{"key":"ref16","article-title":"The llama 3 herd of models","author":"Dubey","year":"2024","journal-title":"arXiv preprint"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.681"},{"key":"ref18","author":"Games","journal-title":"Unreal engine."},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1145\/3437801.3441578"},{"key":"ref20","year":"2022","journal-title":"LlaMa.cpp"},{"key":"ref21","year":"2024","journal-title":"LLM Game Agent Papers"},{"key":"ref22","year":"2024","journal-title":"Google recaps how its LLMs could change in-game interactions"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1201\/9781315267845"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1145\/3579371.3589038"},{"key":"ref25","first-page":"539","article-title":"Microsecond-scale preemption for concurrent \\{GPU-accelerated\\} \\{DNN\\} inferences","volume-title":"16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22)","author":"Han","year":"2022"},{"key":"ref26","volume-title":"Creating Next-Gen Agents in KRAFTON\u2019s inZOI","year":"2025"},{"key":"ref27","author":"Lambe","year":"2025","journal-title":"The NEW Surprising Number of Steam Games that Use GenAI"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/CoG60054.2024.10645625"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1145\/3307681.3325409"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2020.3024068"},{"key":"ref31","first-page":"161","article-title":"Zico: Efficient \\{GPU\\} memory sharing for concurrent \\{DNN\\} training","volume-title":"2021 USENIX Annual Technical Conference (USENIX ATC 21)","author":"Lim","year":"2021"},{"key":"ref32","first-page":"929","article-title":"Parrot: Efficient serving of \\{LLM-based\\} applications with semantic variable","volume-title":"18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24)","author":"Lin","year":"2024"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1145\/3503222.3507752"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1515\/9781683922902"},{"key":"ref35","article-title":"Adaptive layer-skipping in pre-trained 1 lms","author":"Luo","year":"2025","journal-title":"arXiv preprint"},{"key":"ref36","article-title":"Large language models play starcraft ii: Benchmarks and a chain of summarization approach","author":"Ma","year":"2023","journal-title":"arXiv preprint"},{"key":"ref37","article-title":"Shortgpt: Layers in large language models are more redundant than you expect","author":"Men","year":"2024","journal-title":"arXiv preprint"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1262"},{"key":"ref39","year":"2025","journal-title":"NetEase to add game version of ChatGPT to Justice Online Mobile for dialogue generation and unique reactions"},{"key":"ref40","author":"News","year":"2024","journal-title":"Can AI make video games more immersive?"},{"key":"ref41","year":"2023","journal-title":"RTX 4090 whitepaper"},{"key":"ref42","year":"2025","journal-title":"GeForce NOW"},{"key":"ref43","year":"2025","journal-title":"NVIDIA ACE for Games"},{"key":"ref44","year":"2025","journal-title":"NVIDIA Announces First Digital Human Technologies OnDevice Small Language Model"},{"key":"ref45","year":"2025","journal-title":"NVIDIA Redefines Game AI With ACE Autonomous Game Characters"},{"key":"ref46","year":"2025","journal-title":"RTX 5090 whitepaper"},{"key":"ref47","article-title":"Tensorflow-serving: Flexible, highperformance ml serving","author":"Olston","year":"2017","journal-title":"arXiv preprint"},{"key":"ref48","article-title":"Generative agents: Interactive simulacra of human behavior. arxiv","author":"Park","year":"2023","journal-title":"arXiv preprint"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA59077.2024.00019"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1145\/2632216"},{"key":"ref51","author":"Sarfaraz","year":"2025","journal-title":"1 in 5 Steam games released in 2025 use Generative AI"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1269"},{"key":"ref53","article-title":"Swarmbrain: Embodied agent for real-time strategy game starcraft ii via large language models","author":"Shao","year":"2024","journal-title":"arXiv preprint"},{"key":"ref54","author":"Siebler","year":"2023","journal-title":"Optimizing latency for Azure OpenAI Service"},{"key":"ref55","year":"2025","journal-title":"AI_Roguelite"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1145\/3627703.3629578"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1145\/3640794.3665582"},{"key":"ref58","article-title":"Accelerating llama inference by enabling intermediate layer decoding via instruction tuning with lite","author":"Varshney","year":"2023","journal-title":"arXiv preprint"},{"key":"ref59","article-title":"Attention is all you need","author":"Vaswani","year":"2017","journal-title":"Advances in neural information processing systems"},{"key":"ref60","author":"Verge","year":"2025","journal-title":"Nvidia\u2019s AI NPCs are no longer chatbots - they\u2019re your new PUBG teammate"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1109\/ICCCNT51525.2021.9579618"},{"key":"ref62","first-page":"696","article-title":"Wavelet: Efficient dnn training with tick-tock scheduling","volume-title":"Proceedings of Machine Learning and Systems","volume":"3","author":"Wang","year":"2021"},{"key":"ref63","article-title":"Voyager: An open-ended embodied agent with large language models","author":"Wang","year":"2023","journal-title":"arXiv preprint"},{"key":"ref64","first-page":"533","article-title":"\\{AntMan\\}: Dynamic scaling on \\{GPU\\} clusters for deep learning","volume-title":"14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20)","author":"Xiao","year":"2020"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2013.288"},{"key":"ref66","first-page":"217","article-title":"\\{PilotFish\\}: Harvesting free cycles of cloud gaming with deep learning training","volume-title":"2022 USENIX Annual Technical Conference (USENIX ATC 22)","author":"Zhang","year":"2022"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1145\/3330345.3330351"},{"key":"ref68","first-page":"193","article-title":"\\{DistServe\\}: Disaggregating prefill and decoding for goodput-optimized large language model serving","volume-title":"18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24)","author":"Zhong","year":"2024"},{"key":"ref69","first-page":"749","article-title":"\\{NanoFlow\\}: Towards optimal large language model serving throughput","volume-title":"19th USENIX Symposium on Operating Systems Design and Implementation (OSDI 25)","author":"Zhu","year":"2025"}],"event":{"name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","location":"Sydney, Australia","start":{"date-parts":[[2026,1,31]]},"end":{"date-parts":[[2026,2,4]]}},"container-title":["2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11408404\/11408433\/11408477.pdf?arnumber=11408477","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T06:40:04Z","timestamp":1772692804000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11408477\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,1,31]]},"references-count":69,"URL":"https:\/\/doi.org\/10.1109\/hpca68181.2026.11408477","relation":{},"subject":[],"published":{"date-parts":[[2026,1,31]]}}}