paper-with-me

Papers

OrbitFlow: SLO-Aware Long-Context LLM Serving with Fine-Grained KV Cache Reconfiguration

2026-01-05 · Xinyue Ma, Heelim Hong, Taegeon Um, Jongseop Lee, Seoyeong Choy, Woo-Yeon Lee, Myeongjae Jeon arxiv

Serving long-context LLMs is challenging because request lengths and batch composition vary during token generation, causing the memory footprint to fluctuate significantly at runtime. Offloading KV caches to host memory limits effective memory usage, but existing static and predetermined offloading strategies cannot adapt to the rapidly shifting memory demands of long-context serving. This often leads to excessive CPU-to-GPU KV transfers that translate into latency spikes and frequent SLO violations. To address these challenges, we introduce OrbitFlow, a fine-grained and adaptive KV cache management system that meets latency SLOs in long-context LLM serving. OrbitFlow employs a lightweight ILP solver to decide which layers' KV caches to retain on the GPU for each request, within memory capacity constraints. It continuously refines KV placements based on runtime feedback when the active plan becomes suboptimal during token generation. Under heavy load, OrbitFlow invokes a fallback mechanism to temporarily defer in-flight requests with large memory footprints, preserving overall SLO attainment. Our experiments demonstrate that OrbitFlow improves SLO attainment for TPOT and TBT by up to 66% and 48%, respectively, while reducing the 95th percentile latency by 38% and achieving up to 3.3x higher throughput compared to existing offloading methods.

📄 PDF Abstract BibTeX arXiv:2601.10729

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Q-ROAR: Outlier-Aware Rescaling for RoPE Position Interpolation in Quantized Long-Context LLMs

2025-09-17 · Ye Qiao, Sitao Huang arxiv

Extending LLM context windows is crucial for long range tasks. RoPE-based position interpolation (PI) methods like linear and frequency-aware scaling extend input lengths without retraining, while post-training quantizat…

Benchmarking KV-Cache Optimizations across Task Quality and System Performance for Long-Context Serving

2026-05-03 · Nikita Agrawal, Ruben Mayer arxiv

Large language model serving is increasingly limited by KV-cache growth under long-context workloads, yet existing KV-cache compression techniques are difficult to compare because they were evaluated on different models,…

Few-Shot Learning

MARDoc: A Memory-Aware Refinement Agent Framework for Multimodal Long Document QA

2026-06-04 · Kaifeng Chen, Hongtao Liu, Qiyao Peng, Jian Yang 외 arxiv

Iterative retrieval-reasoning agents have recently shown promise for multimodal long-document question answering. However, most existing systems maintain a single growing context that mixes retrieval traces, observations…

Question Answering

Medha: Efficiently Serving Multi-Million Context Length LLM Inference Requests Without Approximations

2024-09-25 · Amey Agrawal, Haoran Qiu, Junda Chen, Íñigo Goiri 외

As large language models (LLMs) handle increasingly longer contexts, serving long inference requests of millions of tokens presents unique challenges. We show that existing work for long context inference is largely base…

BlockingChunkingScheduling

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention

2026-06-04 · Yang Liu, ZhaoKai Luo, HuaYi Jin, ZhiYong Wang 외 arxiv

As the input length of large language model (LLM) serving continues to grow, the KV cache has become a dominant bottleneck in AI infrastructure. It limits GPU memory capacity, serving concurrency, cache reuse, and distri…