# Helix: Serving Large Language Models over Heterogeneous GPUs and Network via Max-Flow

arXiv: https://arxiv.org/abs/2406.01566 · HTML: https://mlsanity.com/paper/2406.01566/

## Key facts

- Cited by 55 papers in the map (data through 14 Oct 2026).
- Builds on 28 of its 61 references that are in the map.
- First posted on arXiv 3 Jun 2024.

## Builds on (most cited first, 28 of 28)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Code Llama: Open Foundation Models for Code](https://mlsanity.com/paper/2308.12950/) (2023)
- [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://mlsanity.com/paper/1909.08053/) (2019)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints](https://mlsanity.com/paper/2305.13245/) (2023)
- [WizardCoder: Empowering Code Large Language Models with Evol-Instruct](https://mlsanity.com/paper/2306.08568/) (2023)
- [Fast Transformer Decoding: One Write-Head is All You Need](https://mlsanity.com/paper/1911.02150/) (2019)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Efficiently Scaling Transformer Inference](https://mlsanity.com/paper/2211.05102/) (2022)
- [SpecInfer: Accelerating Generative Large Language Model Serving with Tree-based Speculative Inference and Verification](https://mlsanity.com/paper/2305.09781/) (2023)
- [SARATHI: Efficient LLM Inference by Piggybacking Decodes with Chunked Prefills](https://mlsanity.com/paper/2308.16369/) (2023)
- [Serving DNNs like Clockwork: Performance Predictability from the Bottom Up](https://mlsanity.com/paper/2006.02464/) (2020)
- [Towards Efficient Generative Large Language Model Serving: A Survey from Algorithms to Systems](https://mlsanity.com/paper/2312.15234/) (2023)
- [Decentralized Training of Foundation Models in Heterogeneous Environments](https://mlsanity.com/paper/2206.01288/) (2022)
- [Petals: Collaborative Inference and Fine-tuning of Large Models](https://mlsanity.com/paper/2209.01188/) (2022)
- [TensorFlow-Serving: Flexible, High-Performance ML Serving](https://mlsanity.com/paper/1712.06139/) (2017)
- [Distributed Deep Learning in Open Collaborations](https://mlsanity.com/paper/2106.10207/) (2021)
- [Bamboo: Making Preemptible Instances Resilient for Affordable Training of Large DNNs](https://mlsanity.com/paper/2204.12013/) (2022)
- [HetPipe: Enabling Large DNN Training on (Whimpy) Heterogeneous GPU Clusters through Integration of Pipelined Model Parallelism and Data Parallelism](https://mlsanity.com/paper/2005.14038/) (2020)
- [Towards Crowdsourced Training of Large Neural Networks using Decentralized Mixture-of-Experts](https://mlsanity.com/paper/2002.04013/) (2020)
- [Distributed Inference and Fine-tuning of Large Language Models Over The Internet](https://mlsanity.com/paper/2312.08361/) (2023)
- [M'elange: Cost Efficient Large Language Model Serving by Exploiting GPU Heterogeneity](https://mlsanity.com/paper/2404.14527/) (2024)
- [HAP: SPMD DNN Training on Heterogeneous GPU Clusters with Automated Program Synthesis](https://mlsanity.com/paper/2401.05965/) (2024)
- [Interleaved Weighted Round-Robin: A Network Calculus Analysis](https://mlsanity.com/paper/2003.08372/) (2020)

## Built on by (most cited first, 30 of 55)

- [SLOs-Serve: Optimized Serving of Multi-SLO LLMs](https://mlsanity.com/paper/2504.08784/) (2025)
- [L-MTP: Leap Multi-Token Prediction Beyond Adjacent Context for Large Language Models](https://mlsanity.com/paper/2505.17505/) (2025)
- [RetroInfer: A Vector Storage Engine for Scalable Long-Context LLM Inference](https://mlsanity.com/paper/2505.02922/) (2025)
- [Nexus:Proactive Intra-GPU Disaggregation of Prefill and Decode in LLM Serving](https://mlsanity.com/paper/2507.06608/) (2025)
- [Parallax: Efficient LLM Inference Service over Decentralized Environment](https://mlsanity.com/paper/2509.26182/) (2025)
- [UCCL-EP: Portable Expert-Parallel Communication](https://mlsanity.com/paper/2512.19849/) (2025)
- [TPI-LLM: Serving 70B-scale LLMs Efficiently on Low-resource Edge Devices](https://mlsanity.com/paper/2410.00531/) (2024)
- [Prefill-as-a-Service: KVCache of Next-Generation Models Could Go Cross-Datacenter](https://mlsanity.com/paper/2604.15039/) (2026)
- [AReaL-Hex: Accommodating Asynchronous RL Training over Heterogeneous GPUs](https://mlsanity.com/paper/2511.00796/) (2025)
- [Janus: Disaggregating Attention and Experts for Scalable MoE Inference](https://mlsanity.com/paper/2512.13525/) (2025)
- [Agentic Services Computing](https://mlsanity.com/paper/2509.24380/) (2025)
- [OrbitFlow: SLO-Aware Long-Context LLM Serving with Fine-Grained KV Cache Reconfiguration](https://mlsanity.com/paper/2601.10729/) (2026)
- [BOute: Cost-Efficient LLM Serving with Heterogeneous LLMs and GPUs via Multi-Objective Bayesian Optimization](https://mlsanity.com/paper/2602.10729/) (2026)
- [DeServe: Towards Affordable Offline LLM Inference via Decentralization](https://mlsanity.com/paper/2501.14784/) (2025)
- [TAH-QUANT: Effective Activation Quantization in Pipeline Parallelism over Slow Network](https://mlsanity.com/paper/2506.01352/) (2025)
- [TetriServe: Efficiently Serving Mixed DiT Workloads](https://mlsanity.com/paper/2510.01565/) (2025)
- [Cornfigurator: Automated Planning for Any-to-Any Multimodal Model Serving](https://mlsanity.com/paper/2512.14098/) (2025)
- [KV Admission: Learning What to Write for Efficient Long-Context LLM Inference](https://mlsanity.com/paper/2512.17452/) (2025)
- [Unifying Sparse Attention with Hierarchical Memory for Scalable Long-Context LLM Serving](https://mlsanity.com/paper/2604.26837/) (2026)
- [ShadowServe: Interference-Free KV Cache Fetching for Distributed Prefix Caching](https://mlsanity.com/paper/2509.16857/) (2025)
- [Rearchitecting Datacenter Lifecycle for AI: A TCO-Driven Framework](https://mlsanity.com/paper/2509.26534/) (2025)
- [Rethinking Latency Denial-of-Service: Attacking the LLM Serving Framework, Not the Model](https://mlsanity.com/paper/2602.07878/) (2026)
- [Coral: Cost-Efficient Multi-LLM Serving over Heterogeneous Cloud GPUs](https://mlsanity.com/paper/2605.04357/) (2026)
- [Observation, Not Prediction: Conversation-Level Disaggregated Scheduling for Agentic Serving](https://mlsanity.com/paper/2606.01839/) (2026)
- [SMetric: Rethink LLM Scheduling for Serving Agents with Balanced Session-centric Scheduling](https://mlsanity.com/paper/2607.08565/) (2026)
- [SmartGen: Seamless Disaggregated LLM Inference with Selective KV Cache Transfer](https://mlsanity.com/paper/2607.28150/) (2026)
- [AdaServe: Accelerating Multi-SLO LLM Serving with SLO-Customized Speculative Decoding](https://mlsanity.com/paper/2501.12162/) (2025)
- [Hardwired-Neurons Language Processing Units as General-Purpose Cognitive Substrates](https://mlsanity.com/paper/2508.16151/) (2025)
- [GoodSpeed: Optimizing Fair Goodput with Adaptive Speculative Decoding in Distributed Edge Inference](https://mlsanity.com/paper/2512.09963/) (2025)
- [TIDE: Temporal Incremental Draft Engine for Self-Improving LLM Inference](https://mlsanity.com/paper/2602.05145/) (2026)

## Abstract (arXiv)

This paper introduces Helix, a distributed system for high-throughput, low-latency large language model (LLM) serving in heterogeneous GPU clusters. The key idea behind Helix is to formulate inference computation of LLMs over heterogeneous GPUs and network connections as a max-flow problem on directed, weighted graphs, whose nodes represent GPU instances and edges capture both GPU and network heterogeneity through their capacities. Helix then uses a mixed integer linear programming (MILP) algorithm to discover highly optimized strategies to serve LLMs on heterogeneous GPUs. This approach allows Helix to jointly optimize model placement and request scheduling, two highly entangled tasks in heterogeneous LLM serving. Our evaluation on several heterogeneous clusters ranging from 24 to 42 GPU nodes shows that Helix improves serving throughput by up to 3.3x and reduces prompting and decoding latency by up to 66% and 24%, respectively, compared to existing approaches. Helix is available at https://github.com/Thesys-lab/Helix-ASPLOS25.