EdgeMM: Multi-Core CPU with Heterogeneous AI-Extension and Activation-aware Weight Pruning for Multimodal LLMs at Edge

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bai, Kangbo, Ye, Le, Huang, Ru, Jia, Tianyu
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916739354198016
author Bai, Kangbo
Ye, Le
Huang, Ru
Jia, Tianyu
author_facet Bai, Kangbo
Ye, Le
Huang, Ru
Jia, Tianyu
contents Emerging multimodal LLMs (MLLMs) exhibit strong cross-modality perception and reasoning capabilities and hold great potential for various applications at edge. However, MLLMs typically consist of a compute-intensive modality encoder and a memory-bound LLM decoder, leading to distinct bottlenecks for hardware designs. In this work, we present a multi-core CPU solution with heterogeneous AI extensions, which are based on either the compute-centric systolic array or memory-centric digital compute-in-memory (CIM) co-processors. In addition, dynamic activation-aware weight pruning and bandwidth management are developed to enhance bandwidth efficiency and core utilization, improving overall performance. We implemented our solution using commercial 22nm technology. For representative MLLMs, our evaluations show EdgeMM can achieve 2.84x performance speedup compared to laptop 3060 GPU.
format Preprint
id arxiv_https___arxiv_org_abs_2505_10782
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle EdgeMM: Multi-Core CPU with Heterogeneous AI-Extension and Activation-aware Weight Pruning for Multimodal LLMs at Edge
Bai, Kangbo
Ye, Le
Huang, Ru
Jia, Tianyu
Hardware Architecture
Emerging multimodal LLMs (MLLMs) exhibit strong cross-modality perception and reasoning capabilities and hold great potential for various applications at edge. However, MLLMs typically consist of a compute-intensive modality encoder and a memory-bound LLM decoder, leading to distinct bottlenecks for hardware designs. In this work, we present a multi-core CPU solution with heterogeneous AI extensions, which are based on either the compute-centric systolic array or memory-centric digital compute-in-memory (CIM) co-processors. In addition, dynamic activation-aware weight pruning and bandwidth management are developed to enhance bandwidth efficiency and core utilization, improving overall performance. We implemented our solution using commercial 22nm technology. For representative MLLMs, our evaluations show EdgeMM can achieve 2.84x performance speedup compared to laptop 3060 GPU.
title EdgeMM: Multi-Core CPU with Heterogeneous AI-Extension and Activation-aware Weight Pruning for Multimodal LLMs at Edge
topic Hardware Architecture
url https://arxiv.org/abs/2505.10782