{"slug": "unlocking-non-uniform-kv-cache-for-efficient-multi-turn-llm-serving", "title": "Unlocking Non-Uniform KV Cache for Efficient Multi-Turn LLM Serving", "summary": "Researchers introduced Tangram, a serving system that enables non-uniform Key-Value cache compression for multi-turn large language model inference. The system uses deterministic budget allocation, head group page clustering, and ahead-of-time load balancing to overcome memory fragmentation and scheduling inefficiencies, achieving up to 2.6x throughput improvement over existing baselines without compromising model accuracy.", "body_md": "# Computer Science > Machine Learning\n\n[Submitted on 4 Jun 2026]\n\n# Title:Tangram: Unlocking Non-Uniform KV Cache for Efficient Multi-turn LLM Serving\n\n[View PDF](/pdf/2606.06302)\n\n[HTML (experimental)](https://arxiv.org/html/2606.06302v1)\n\nAbstract:Multi-turn Large Language Model (LLM) serving is critical for consistent user experiences, yet the linear growth of the Key-Value (KV) cache imposes significant pressure on GPU memory and bandwidth. Non-uniform KV compression effectively preserves more information by considering the individual importance of each KV cache. However, such KV cache heterogeneity introduces various systemic challenges - including memory fragmentation, scheduling complexities, and diminished kernel utilization - which collectively lead to significant inefficiencies in existing LLM serving systems.\n\nTo overcome these challenges, we present Tangram, a novel serving system designed to make Non-uniform KV caches practical. Tangram addresses systemic inefficiencies through three core techniques: (1) Deterministic Budget Allocation assigns a static memory footprint to each head based on its intrinsic pattern, entirely eliminating dynamic scheduling overhead and prefill stalls; (2) Head Group Page clusters attention heads with similar retention demands and manages them with independent, vectorized page tables, thereby maximizing physical memory reclamation; and (3) Ahead-of-Time (AOT) Load Balancing leverages static budget profiles to ensure uniform GPU utilization without runtime overhead. Experimental results show that Tangram improves throughput by up to 2.6x compared to existing baselines, while fully preserving model accuracy. Our implementation is publicly available at[this https URL].\n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer\n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers\n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps\n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations\n\n*(*[What are Smart Citations?](https://www.scite.ai/))# Code, Data and Media Associated with this Article\n\nalphaXiv\n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers\n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub\n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub\n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face\n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast\n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower\n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender\n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))\nIArxiv Recommender\n\n*(*[What is IArxiv?](https://iarxiv.org/about))# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [ Learn more about arXivLabs](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/unlocking-non-uniform-kv-cache-for-efficient-multi-turn-llm-serving", "canonical_source": "https://arxiv.org/abs/2606.06302", "published_at": "2026-06-06 01:20:21+00:00", "updated_at": "2026-06-06 01:47:04.958951+00:00", "lang": "en", "topics": ["machine-learning", "large-language-models", "ai-infrastructure", "ai-research", "neural-networks"], "entities": ["Tangram", "GPU"], "alternates": {"html": "https://wpnews.pro/news/unlocking-non-uniform-kv-cache-for-efficient-multi-turn-llm-serving", "markdown": "https://wpnews.pro/news/unlocking-non-uniform-kv-cache-for-efficient-multi-turn-llm-serving.md", "text": "https://wpnews.pro/news/unlocking-non-uniform-kv-cache-for-efficient-multi-turn-llm-serving.txt", "jsonld": "https://wpnews.pro/news/unlocking-non-uniform-kv-cache-for-efficient-multi-turn-llm-serving.jsonld"}}