"Skip to main content\narXiv is now an independent nonprofit!\nLearn more\n×\nSearch\nSubmit\nDonate\nLog in\nShowing 1–7 of 7 results for all: long context KV cache disaggregated serving\nSearch v0.5.6 released 2020-02-24 \nSearch term or terms\nAll fields\nTitle\nAuthor(s)\nAbstract\nComments\nJournal reference\nACM classification\nMSC classification\nReport number\narXiv identifier\nDOI\nORCID\nLicense (URI)\narXiv author ID\nHelp pages\nFull text\nSearch\n Show abstracts Hide abstracts\nAdvanced Search\n25\n50\n100\n200\n results per page.\nSort results by \nAnnouncement date (newest first)\nAnnouncement date (oldest first)\nSubmission date (newest first)\nSubmission date (oldest first)\nRelevance\nGo\n\narXiv:2607.01831 [pdf, ps, other] \n\n cs.DC cs.LG\n\nLynx: Progressive Speculative Quantization for accelerating KV Transfer in Long-Context Inference\n\nAuthors: Wenchen Han, Gingfung Matthew Yeung, Marco Barletta, William Toner, Amory Hoste, Adam Barker\n\nAbstract: Long-… ▽ More\n\nSubmitted 2 July, 2026; originally announced July 2026.\n\nComments: 15 pages, 12 figures. This manuscript was originally submitted to SIGCOMM '26 in February 2026\n\nACM Class: C.2.4; I.2.11\n\narXiv:2607.01617 [pdf, ps, other] \n\n cs.AR \ndoi\n10.1109/LCA.2026.3709108\n\n3DLS: A 3D Logic-Stacked Architecture for Disaggregated LLM Serving\n\nAuthors: Jaehun Lee, In-Jun Jung, Joo-Young Kim\n\nAbstract: Large language model (LLM) serving increasingly combines prefill-decode (PD)… ▽ More\n\nSubmitted 1 July, 2026; originally announced July 2026.\n\nComments: Accepted to IEEE Computer Architecture Letters (CAL), 2026\n\narXiv:2606.19746 [pdf, ps, other] \n\n cs.DC\n\nSAC: Disaggregated KV Cache System for Sparse Attention LLMs with CXL\n\nAuthors: Ruiyang Ma, Teng Ma, Junru Li, Hantian Zha, Xuchun Shang, Qingda Hu, Zheng Liu, Xinjun Yang, Tao Ma, Guojie Luo\n\nAbstract: The scaling of LLMs toward long-… ▽ More\n\nSubmitted 17 June, 2026; originally announced June 2026.\n\narXiv:2604.15039 [pdf, ps, other] \n\n cs.DC\n\nPrefill-as-a-Service: KVCache of Next-Generation Models Could Go Cross-Datacenter\n\nAuthors: Ruoyu Qin, Weiran He, Yaoyu Wang, Zheming Li, Xinran Xu, Yongwei Wu, Weimin Zheng, Mingxing Zhang\n\nAbstract: Prefill-decode (PD) disaggregation has become the standard architecture for large-scale LLM… ▽ More\n\nSubmitted 21 April, 2026; v1 submitted 16 April, 2026; originally announced April 2026.\n\nComments: 16 pages, 5 figures, 6 tables\n\narXiv:2604.09613 \n\n cs.DC cs.AI\n\nToken-Budget-Aware Pool Routing for Cost-Efficient LLM Inference\n\nAuthors: Huamin Chen, Xunzhuo Liu, Junchen Jiang, Bowei He, Xue Liu\n\nAbstract: Production vLLM fleets provision every instance for worst-case context length, wasting 4-8x concurrency on the 80-95% of requests that are short and simultaneously triggering… ▽ More\n\nSubmitted 14 April, 2026; v1 submitted 13 March, 2026; originally announced April 2026.\n\nComments: duplicate of arXiv:2604.08075\n\narXiv:2604.08075 [pdf, ps, other] \n\n cs.CL\n\nDual-Pool Token-Budget Routing for Cost-Efficient and Reliable LLM Serving\n\nAuthors: Xunzhuo Liu, Bowei He, Xue Liu, Andy Luo, Haichen Zhang, Huamin Chen\n\nAbstract: Production vLLM fleets typically provision each instance for the worst-case context length, leading to substantial… ▽ More\n\nSubmitted 9 April, 2026; originally announced April 2026.\n\narXiv:2601.22001 [pdf, ps, other] \n\n cs.AI cs.AR cs.DC\n\nHeterogeneous Computing: The Key to Powering the Future of AI Agent Inference\n\nAuthors: Yiren Zhao, Junyi Liu\n\nAbstract: …the memory capacity wall. Across agentic workflows (chat, coding, web use, computer use) and base model choices (GQA/MLA, MoE, quantization), OI/CF can shift dramatically, with long… ▽ More\n\nSubmitted 29 January, 2026; originally announced January 2026.\n\nWe gratefully acknowledge support from our major funders, member institutions, and all contributors.\nAbout\n·\nHelp\n·\nContact\n·\nSubscribe\n·\nCopyright\n·\nPrivacy\n·\nAccessibility\n·\nOperational Status\n(opens in new tab)\nMajor funding support from"