# ConServe: Fine-Grained GPU Harvesting for LLM Online and Offline Co-Serving

arXiv: https://arxiv.org/abs/2410.01228 · HTML: https://mlsanity.com/paper/2410.01228/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 70 references that are in the map.
- First posted on arXiv 2 Oct 2024.

## Builds on (most cited first, 19 of 19)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [fairseq: A Fast, Extensible Toolkit for Sequence Modeling](https://mlsanity.com/paper/1904.01038/) (2019)
- [Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization](https://mlsanity.com/paper/1808.08745/) (2018)
- [Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference](https://mlsanity.com/paper/2403.04132/) (2024)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU](https://mlsanity.com/paper/2303.06865/) (2023)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve](https://mlsanity.com/paper/2403.02310/) (2024)
- [Multi-News: a Large-Scale Multi-Document Summarization Dataset and Abstractive Hierarchical Model](https://mlsanity.com/paper/1906.01749/) (2019)
- [DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scale](https://mlsanity.com/paper/2207.00032/) (2022)
- [Clipper: A Low-Latency Online Prediction Serving System](https://mlsanity.com/paper/1612.03079/) (2016)
- [Serving DNNs like Clockwork: Performance Predictability from the Bottom Up](https://mlsanity.com/paper/2006.02464/) (2020)
- [DuReader: a Chinese Machine Reading Comprehension Dataset from Real-world Applications](https://mlsanity.com/paper/1711.05073/) (2017)
- [A Comprehensive Survey on Process-Oriented Automatic Text Summarization with Exploration of LLM-Based Methods](https://mlsanity.com/paper/2403.02901/) (2024)
- [SpotServe: Serving Generative Large Language Models on Preemptible Instances](https://mlsanity.com/paper/2311.15566/) (2023)
- [ServerlessLLM: Low-Latency Serverless Inference for Large Language Models](https://mlsanity.com/paper/2401.14351/) (2024)
- [Fairness in Serving Large Language Models](https://mlsanity.com/paper/2401.00588/) (2023)
- [FlexLLM: Token-Level Co-Serving of LLM Inference and Finetuning with SLO Guarantees](https://mlsanity.com/paper/2402.18789/) (2024)

## Built on by (most cited first, 5 of 5)

- [Demystifying Cost-Efficiency in LLM Serving over Heterogeneous GPUs](https://mlsanity.com/paper/2502.00722/) (2025)
- [Foundry: Template-Based CUDA Graph Context Materialization for Fast LLM Serving Cold Start](https://mlsanity.com/paper/2604.06664/) (2026)
- [Blink: CPU-Free LLM Inference by Delegating the Serving Stack to GPU and SmartNIC](https://mlsanity.com/paper/2604.07609/) (2026)
- [Cascade: Exploiting SLO-Aware latency budget for fair and high goodput LLM inference serving](https://mlsanity.com/paper/2608.06557/) (2026)
- [RW-TTT: Batched Serving for Request-Owned Test-Time Training State](https://mlsanity.com/paper/2605.28053/) (2026)

## Abstract (arXiv)

Large language model (LLM) serving demands low latency and high throughput, but high load variability makes it challenging to achieve high GPU utilization. In this paper, we identify a synergetic but overlooked opportunity to co-serve latency-critical online requests alongside latency-tolerant offline tasks such as model benchmarking. While promising, existing serving systems fail to co-serve them efficiently, as their coarse-grained resource management at the request or iteration level cannot harvest millisecond-level GPU idle cycles without introducing interference that violates online latency objectives. ConServe is a new LLM co-serving system that achieves high throughput and strong online latency guarantees by managing resources at finer granularities. ConServe introduces three techniques: (1) a latency-aware token-level scheduler that precisely sizes offline batches and tokens to fit within online latency objectives; (2) sub-iteration, layer-wise preemption that allows offline tasks to yield to online load spikes; and (3) incremental KV cache management that enables preempting and resuming offline requests at near-zero cost. Evaluations with Llama-3.1 and Qwen-2.5 models on real-world workloads show that ConServe delivers an average of 2.2× higher throughput and reduces online serving tail latency by 2.9× on average compared to state-of-the-art systems.