# HeteGen: Heterogeneous Parallel Inference for Large Language Models on Resource-Constrained Devices

arXiv: https://arxiv.org/abs/2403.01164 · HTML: https://mlsanity.com/paper/2403.01164/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Mar 2024.

## Built on by (most cited first, 9 of 9)

- [NEO: Saving GPU Memory Crisis with CPU Offloading for Online LLM Inference](https://mlsanity.com/paper/2411.01142/) (2024)
- [Deploying Foundation Model Powered Agent Services: A Survey](https://mlsanity.com/paper/2412.13437/) (2024)
- [A Cost-Effective Near-Storage Processing Solution for Offline Inference of Long-Context LLMs](https://mlsanity.com/paper/2502.09921/) (2025)
- [Adaptive and Resource-efficient Agentic AI Systems for Mobile and Embedded Devices: A Survey](https://mlsanity.com/paper/2510.00078/) (2025)
- [SuperInfer: SLO-Aware Rotary Scheduling and Memory Management for LLM Inference on Superchips](https://mlsanity.com/paper/2601.20309/) (2026)
- [Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots](https://mlsanity.com/paper/2607.02501/) (2026)
- [Spatio-Temporal Parallelism for Diffusion Model Inference on Heterogeneous Multi-GPU Systems](https://mlsanity.com/paper/2509.04719/) (2025)
- [WindVE: Collaborative CPU-NPU Vector Embedding](https://mlsanity.com/paper/2504.14941/) (2025)
- [APEX: Asynchronous Parallel CPU-GPU Execution for Online LLM Inference on Constrained GPUs](https://mlsanity.com/paper/2506.03296/) (2025)

## Abstract (arXiv)

In recent times, the emergence of Large Language Models (LLMs) has resulted in increasingly larger model size, posing challenges for inference on low-resource devices. Prior approaches have explored offloading to facilitate low-memory inference but often suffer from efficiency due to I/O bottlenecks. To achieve low-latency LLMs inference on resource-constrained devices, we introduce HeteGen, a novel approach that presents a principled framework for heterogeneous parallel computing using CPUs and GPUs. Based on this framework, HeteGen further employs heterogeneous parallel computing and asynchronous overlap for LLMs to mitigate I/O bottlenecks. Our experiments demonstrate a substantial improvement in inference speed, surpassing state-of-the-art methods by over 317% at most.