Cover Image

P3-LLM: An Integrated NPU-PIM Accelerator for Edge LLM Inference Using Hybrid Numerical Formats

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Chen, Yuzong, Fang, Chao, Dai, Xilai, Wu, Yuheng, Tambe, Thierry, Verhelst, Marian, Abdelfattah, Mohamed S.
Format:	Preprint
Published:	2025
Subjects:	Hardware Architecture Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Kratos: An FPGA Benchmark for Unrolled DNNs with Fine-Grained Sparsity and Mixed Precision
by: Dai, Xilai, et al.
Published: (2024)

NeuPIMs: NPU-PIM Heterogeneous Acceleration for Batched LLM Inferencing
by: Heo, Guseul, et al.
Published: (2024)

BitMoD: Bit-serial Mixture-of-Datatype LLM Acceleration
by: Chen, Yuzong, et al.
Published: (2024)

The Hyperscale Lottery: How State-Space Models Have Sacrificed Edge Efficiency
by: Geens, Robin, et al.
Published: (2026)

IANUS: Integrated Accelerator based on NPU-PIM Unified Memory System
by: Seo, Minseok, et al.
Published: (2024)

Hardware Generation and Exploration of Lookup Table-Based Accelerators for 1.58-bit LLM Inference
by: Geens, Robin, et al.
Published: (2026)

CD-PIM: A High-Bandwidth and Compute-Efficient LPDDR5-Based PIM for Low-Batch LLM Acceleration on Edge-Device
by: Lin, Ye, et al.
Published: (2026)

Enabling Efficient Hardware Acceleration of Hybrid Vision Transformer (ViT) Networks at the Edge
by: Dumoulin, Joren, et al.
Published: (2025)

BBS: Bi-directional Bit-level Sparsity for Deep Learning Acceleration
by: Chen, Yuzong, et al.
Published: (2024)

Anda: Unlocking Efficient LLM Inference with a Variable-Length Grouped Activation Data Format
by: Fang, Chao, et al.
Published: (2024)

Sangam: Chiplet-Based DRAM-PIM Accelerator with CXL Integration for LLM Inferencing
by: Kiyawat, Khyati, et al.
Published: (2025)

Precision-Scalable Microscaling Datapaths with Optimized Reduction Tree for Efficient NPU Integration
by: Cuyckens, Stef, et al.
Published: (2025)

iEEG Seizure Detection with a Sparse Hyperdimensional Computing Accelerator
by: Cuyckens, Stef, et al.
Published: (2025)

Hybrid Systolic Array Accelerator with Optimized Dataflow for Edge Large Language Model Inference
by: Chen, Chun-Ting, et al.
Published: (2025)

Fine-Grained Fusion: The Missing Piece in Area-Efficient State Space Model Acceleration
by: Geens, Robin, et al.
Published: (2025)

Inclusive-PIM: Hardware-Software Co-design for Broad Acceleration on Commercial PIM Architectures
by: Alsop, Johnathan, et al.
Published: (2023)

Hardware-Centric Analysis of DeepSeek's Multi-Head Latent Attention
by: Geens, Robin, et al.
Published: (2025)

Hyft: A Reconfigurable Softmax Accelerator with Hybrid Numeric Format for both Training and Inference
by: Xia, Tianhua, et al.
Published: (2023)

PIM-LLM: A High-Throughput Hybrid PIM Architecture for 1-bit LLMs
by: Malekar, Jinendra, et al.
Published: (2025)

VitaLLM: A Versatile and Tiny Accelerator for Mixed-Precision LLM Inference on Edge Devices
by: Lin, Zi-Wei, et al.
Published: (2026)

Double Duty: FPGA Architecture to Enable Concurrent LUT and Adder Chain Usage
by: Pun, Junius, et al.
Published: (2025)

SnipSnap: A Joint Compression Format and Dataflow Co-Optimization Framework for Efficient Sparse LLM Accelerator Design
by: Wu, Junyi, et al.
Published: (2025)

L3: DIMM-PIM Integrated Architecture and Coordination for Scalable Long-Context LLM Inference
by: Liu, Qingyuan, et al.
Published: (2025)

PIM-GPT: A Hybrid Process-in-Memory Accelerator for Autoregressive Transformers
by: Wu, Yuting, et al.
Published: (2023)

LEAP: LLM Inference on Scalable PIM-NoC Architecture with Balanced Dataflow and Fine-Grained Parallelism
by: Wang, Yimin, et al.
Published: (2025)

LP-Spec: Leveraging LPDDR PIM for Efficient LLM Mobile Speculative Inference with Architecture-Dataflow Co-Optimization
by: He, Siyuan, et al.
Published: (2025)

How to keep pushing ML accelerator performance? Know your rooflines!
by: Verhelst, Marian, et al.
Published: (2025)

HH-PIM: Dynamic Optimization of Power and Performance with Heterogeneous-Hybrid PIM for Edge AI Devices
by: Jeon, Sangmin, et al.
Published: (2025)

Stream: Design Space Exploration of Layer-Fused DNNs on Heterogeneous Dataflow Accelerators
by: Symons, Arne, et al.
Published: (2022)

PIM-MMU: A Memory Management Unit for Accelerating Data Transfers in Commercial PIM Systems
by: Lee, Dongjae, et al.
Published: (2024)

Efficient Precision-Scalable Hardware for Microscaling (MX) Processing in Robotics Learning
by: Cuyckens, Stef, et al.
Published: (2025)

eIQ Neutron: Redefining Edge-AI Inference with Integrated NPU and Compiler Innovations
by: Bamberg, Lennart, et al.
Published: (2025)

DataMaestro: A Versatile and Efficient Data Streaming Engine Bringing Decoupled Memory Access To Dataflow Accelerators
by: Yi, Xiaoling, et al.
Published: (2025)

PIMphony: Overcoming Bandwidth and Capacity Inefficiency in PIM-based Long-Context LLM Inference System
by: Kwon, Hyucksung, et al.
Published: (2024)

AME-PIM: Can Memory be Your Next Tensor Accelerator?
by: Venieri, Emanuele, et al.
Published: (2026)

LLM-FSM: Scaling Large Language Models for Finite-State Reasoning in RTL Code Generation
by: Wu, Yuheng, et al.
Published: (2026)

Pack my weights and run! Minimizing overheads for in-memory computing accelerators
by: Houshmand, Pouya, et al.
Published: (2024)

EdgeLLM: A Highly Efficient CPU-FPGA Heterogeneous Edge Accelerator for Large Language Models
by: Huang, Mingqiang, et al.
Published: (2024)

Membrane: Accelerating Database Analytics with Bank-Level DRAM-PIM Filtering
by: Shekar, Akhil, et al.
Published: (2025)

Cambricon-LLM: A Chiplet-Based Hybrid Architecture for On-Device Inference of 70B LLM
by: Yu, Zhongkai, et al.
Published: (2024)