# Characterizing Mobile SoC for Accelerating Heterogeneous LLM Inference

arXiv: https://arxiv.org/abs/2501.14794 · HTML: https://mlsanity.com/paper/2501.14794/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 63 references that are in the map.
- First posted on arXiv 11 Jan 2025.

## Builds on (most cited first, 27 of 27)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer](https://mlsanity.com/paper/2408.06072/) (2024)
- [GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers](https://mlsanity.com/paper/2210.17323/) (2022)
- [AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration](https://mlsanity.com/paper/2306.00978/) (2023)
- [ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools](https://mlsanity.com/paper/2406.12793/) (2024)
- [CogVideo: Large-scale Pretraining for Text-to-Video Generation via Transformers](https://mlsanity.com/paper/2205.15868/) (2022)
- [CogAgent: A Visual Language Model for GUI Agents](https://mlsanity.com/paper/2312.08914/) (2023)
- [In-Datacenter Performance Analysis of a Tensor Processing Unit](https://mlsanity.com/paper/1704.04760/) (2017)
- [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)
- [Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve](https://mlsanity.com/paper/2403.02310/) (2024)
- [Mobile-Agent: Autonomous Multi-Modal Mobile Device Agent with Visual Perception](https://mlsanity.com/paper/2401.16158/) (2024)
- [Ansor: Generating High-Performance Tensor Programs for Deep Learning](https://mlsanity.com/paper/2006.06762/) (2020)
- [InternLM-XComposer-2.5: A Versatile Large Vision Language Model Supporting Long-Contextual Input and Output](https://mlsanity.com/paper/2407.03320/) (2024)
- [Mobile-Agent-v2: Mobile Device Operation Assistant with Effective Navigation via Multi-Agent Collaboration](https://mlsanity.com/paper/2406.01014/) (2024)
- [Ferret-UI: Grounded Mobile UI Understanding with Multimodal LLMs](https://mlsanity.com/paper/2404.05719/) (2024)
- [LUT-GEMM: Quantized Matrix Multiplication based on LUTs for Efficient Inference in Large-Scale Generative Language Models](https://mlsanity.com/paper/2206.09557/) (2022)
- [PowerInfer-2: Fast Large Language Model Inference on a Smartphone](https://mlsanity.com/paper/2406.06282/) (2024)
- [ServerlessLLM: Low-Latency Serverless Inference for Large Language Models](https://mlsanity.com/paper/2401.14351/) (2024)
- [Theano: A Python framework for fast computation of mathematical expressions](https://mlsanity.com/paper/1605.02688/) (2016)
- [Large Language Model Inference Acceleration: A Comprehensive Hardware Perspective](https://mlsanity.com/paper/2410.04466/) (2024)
- [Optimizing CNN Model Inference on CPUs](https://mlsanity.com/paper/1809.02697/) (2018)
- [TEQ: Trainable Equivalent Transformation for Quantization of LLMs](https://mlsanity.com/paper/2310.10944/) (2023)
- [SCAR: Scheduling Multi-Model AI Workloads on Heterogeneous Multi-Chiplet Module Accelerators](https://mlsanity.com/paper/2405.00790/) (2024)
- [GPTPU: Accelerating Applications using Edge Tensor Processing Units](https://mlsanity.com/paper/2107.05473/) (2021)
- [Hardware-Assisted Virtualization of Neural Processing Units for Cloud Platforms](https://mlsanity.com/paper/2408.04104/) (2024)

## Built on by (most cited first, 22 of 22)

- [Agent.xpu: Efficient Scheduling of Agentic LLM Workloads on Heterogeneous SoC](https://mlsanity.com/paper/2506.24045/) (2025)
- [Scaling LLM Test-Time Compute with Mobile NPU on Smartphones](https://mlsanity.com/paper/2509.23324/) (2025)
- [Accelerating Mobile Language Model via Speculative Decoding and NPU-Coordinated Execution](https://mlsanity.com/paper/2510.15312/) (2025)
- [Parameter-Free Encoders Remain Viable for RDB Foundation Models](https://mlsanity.com/paper/2607.05476/) (2026)
- [EdgeInfinite-Instruct: Bridging SFT-Based Optimization and NPU-Level Efficiency for Edge Devices](https://mlsanity.com/paper/2508.00370/) (2025)
- [FlexServe: A Fast and Secure LLM Serving System for Mobile Devices with Flexible Resource Isolation](https://mlsanity.com/paper/2603.09046/) (2026)
- [When NPUs Are Not Always Faster: A Stage-Level Analysis of Mobile LLM Inference](https://mlsanity.com/paper/2605.27435/) (2026)
- [Efficient On-Device Diffusion LLM Inference with Mobile NPU](https://mlsanity.com/paper/2606.13740/) (2026)
- [Bridging the Digital Divide: Small Language Models as a Pathway for Physics and Photonics Education in Underdeveloped Regions](https://mlsanity.com/paper/2506.12403/) (2025)
- [TriGen: NPU Architecture for End-to-End Acceleration of Large Language Models based on SW-HW Co-Design](https://mlsanity.com/paper/2602.12962/) (2026)
- [LegoDiffusion: Micro-Serving Text-to-Image Diffusion Workflows](https://mlsanity.com/paper/2604.08123/) (2026)
- [TileFuse: A Fused Mixed-Precision Kernel Library for Efficient Quantized LLM Inference on AMD NPUs](https://mlsanity.com/paper/2606.11357/) (2026)
- [FlexServe: A Fast and Secure LLM Serving System for Mobile Devices with Flexible Resource Isolation](https://mlsanity.com/paper/2606.23370/) (2026)
- [ServerlessT2I: Efficient Text-to-Image Workflow Serving on a Serverless Platform](https://mlsanity.com/paper/2607.26566/) (2026)
- [Hardware Acceleration of Block-Diffusion LLM for Edge Devices](https://mlsanity.com/paper/2609.01084/) (2026)
- [mzCache: On-Device LLM Memory Management under Multitasking](https://mlsanity.com/paper/2609.01338/) (2026)
- [EStream: Fast and Memory-Efficient MoE Prefill through Expert Virtualization on Mobile NPUs](https://mlsanity.com/paper/2609.06551/) (2026)
- [BigMoMo: Efficient Inference of Large-Scale MoE with Speculative Decoding on Mobile Devices](https://mlsanity.com/paper/2609.14643/) (2026)
- [TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching](https://mlsanity.com/paper/2609.21172/) (2026)
- [Empowering Hybrid Attention Models on NPUs](https://mlsanity.com/paper/2609.32114/) (2026)
- [Dynamic Flow, Static Graph: KV Cache Reuse for Efficient LLM Serving on Mobile NPUs](https://mlsanity.com/paper/2609.34727/) (2026)
- [SchemaWeaver: Accelerating On-Device Mobile Agentic AI via Schema-Guided Parallel LLM Decoding](https://mlsanity.com/paper/2610.17436/) (2026)

## Abstract (arXiv)

With the rapid advancement of artificial intelligence technologies such as ChatGPT, AI agents, and video generation, contemporary mobile systems have begun integrating these AI capabilities on local devices to enhance privacy and reduce response latency. To meet the computational demands of AI tasks, current mobile SoCs are equipped with diverse AI accelerators, including GPUs and Neural Processing Units (NPUs). However, there has not been a comprehensive characterization of these heterogeneous processors, and existing designs typically only leverage a single AI accelerator for LLM inference, leading to suboptimal use of computational resources and memory bandwidth. In this paper, we first summarize key performance characteristics of heterogeneous processors, SoC memory bandwidth, etc. Drawing on these observations, we propose different heterogeneous parallel mechanisms to fully exploit both GPU and NPU computational power and memory bandwidth. We further design a fast synchronization mechanism between heterogeneous processors that leverages the unified memory architecture. By employing these techniques, we present HeteroInfer, the fastest LLM inference engine in mobile devices which supports GPU-NPU heterogeneous execution. Evaluation shows that HeteroInfer delivers a 1.34x to 6.02x end-to-end speedup over state-of-the-art GPU-only and NPU-only LLM engines, while maintaining negligible interference with other applications.