Image de couverture de livre

Hardware-Centric Analysis of DeepSeek's Multi-Head Latent Attention

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Geens, Robin, Verhelst, Marian
Format:	Preprint
Publié:	2025
Sujets:	Hardware Architecture
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Hardware Generation and Exploration of Lookup Table-Based Accelerators for 1.58-bit LLM Inference
par: Geens, Robin, et autres
Publié: (2026)

Fine-Grained Fusion: The Missing Piece in Area-Efficient State Space Model Acceleration
par: Geens, Robin, et autres
Publié: (2025)

The Hyperscale Lottery: How State-Space Models Have Sacrificed Edge Efficiency
par: Geens, Robin, et autres
Publié: (2026)

Enabling Efficient Hardware Acceleration of Hybrid Vision Transformer (ViT) Networks at the Edge
par: Dumoulin, Joren, et autres
Publié: (2025)

Anda: Unlocking Efficient LLM Inference with a Variable-Length Grouped Activation Data Format
par: Fang, Chao, et autres
Publié: (2024)

Efficient Precision-Scalable Hardware for Microscaling (MX) Processing in Robotics Learning
par: Cuyckens, Stef, et autres
Publié: (2025)

How to keep pushing ML accelerator performance? Know your rooflines!
par: Verhelst, Marian, et autres
Publié: (2025)

Precision-Scalable Microscaling Datapaths with Optimized Reduction Tree for Efficient NPU Integration
par: Cuyckens, Stef, et autres
Publié: (2025)

Optimizing Layer-Fused Scheduling of Transformer Networks on Multi-accelerator Platforms
par: Colleman, Steven, et autres
Publié: (2024)

Optimising GPGPU Execution Through Runtime Micro-Architecture Parameter Analysis
par: Sarda, Giuseppe M., et autres
Publié: (2024)

Pack my weights and run! Minimizing overheads for in-memory computing accelerators
par: Houshmand, Pouya, et autres
Publié: (2024)

Decoupled Control Flow and Data Access in RISC-V GPGPUs
par: Sarda, Giuseppe M., et autres
Publié: (2025)

Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures
par: Zhao, Chenggang, et autres
Publié: (2025)

Hardware-Algorithm Co-Optimization of Early-Exit Neural Networks for Multi-Core Edge Accelerators
par: Zniber, Alaa, et autres
Publié: (2025)

DataMaestro: A Versatile and Efficient Data Streaming Engine Bringing Decoupled Memory Access To Dataflow Accelerators
par: Yi, Xiaoling, et autres
Publié: (2025)

Stream: Design Space Exploration of Layer-Fused DNNs on Heterogeneous Dataflow Accelerators
par: Symons, Arne, et autres
Publié: (2022)

iEEG Seizure Detection with a Sparse Hyperdimensional Computing Accelerator
par: Cuyckens, Stef, et autres
Publié: (2025)

A 16 nm 1.60TOPS/W High Utilization DNN Accelerator with 3D Spatial Data Reuse and Efficient Shared Memory Access
par: Yi, Xiaoling, et autres
Publié: (2026)

MC$^2$A: Enabling Algorithm-Hardware Co-Design for Efficient Markov Chain Monte Carlo Acceleration
par: Zhao, Shirui, et autres
Publié: (2025)

FlatAttention: Dataflow and Fabric Collectives Co-Optimization for Efficient Multi-Head Attention on Tile-Based Many-PE Accelerators
par: Zhang, Chi, et autres
Publié: (2025)

SwiftKV: An Edge-Oriented Attention Algorithm and Multi-Head Accelerator for Fast, Efficient LLM Decoding
par: Zhang, Junming, et autres
Publié: (2026)

Analog or Digital In-memory Computing? Benchmarking through Quantitative Modeling
par: Sun, Jiacong, et autres
Publié: (2024)

XDMA: A Distributed, Extensible DMA Architecture for Layout-Flexible Data Movements in Heterogeneous Multi-Accelerator SoCs
par: Kong, Fanchen, et autres
Publié: (2025)

An Open-Source HW-SW Co-Development Framework Enabling Efficient Multi-Accelerator Systems
par: Antonio, Ryan Albert, et autres
Publié: (2025)

SALSA: Simulated Annealing based Loop-Ordering Scheduler for DNN Accelerators
par: Jung, Victor J. B., et autres
Publié: (2023)

FloorPlan-DeepSeek (FPDS): A multimodal approach to floorplan generation using vector-based next room prediction
par: Yin, Jun, et autres
Publié: (2025)

DeFiNES: Enabling Fast Exploration of the Depth-first Scheduling Space for DNN Accelerators through Analytical Modeling
par: Mei, Linyan, et autres
Publié: (2022)

H-FA: A Hybrid Floating-Point and Logarithmic Approach to Hardware Accelerated FlashAttention
par: Alexandridis, Kosmas, et autres
Publié: (2025)

StruM: Structured Mixed Precision for Efficient Deep Learning Hardware Codesign
par: Wu, Michael, et autres
Publié: (2025)

P3-LLM: An Integrated NPU-PIM Accelerator for Edge LLM Inference Using Hybrid Numerical Formats
par: Chen, Yuzong, et autres
Publié: (2025)

Multi-Objective Hardware-Mapping Co-Optimisation for Multi-DNN Workloads on Chiplet-based Accelerators
par: Das, Abhijit, et autres
Publié: (2022)

Design and Analysis of Approximate Hardware Accelerators for VVC Intra Angular Prediction
par: de Fraga, Lucas M. Leipnitz, et autres
Publié: (2025)

OpenGeMM: A High-Utilization GeMM Accelerator Generator with Lightweight RISC-V Control and Tight Memory Coupling
par: Yi, Xiaoling, et autres
Publié: (2024)

Torrent: A Distributed DMA for Efficient and Flexible Point-to-Multipoint Data Movement
par: Deng, Yunhao, et autres
Publié: (2025)

Sustainable Hardware Specialization
par: Dangi, Pranav, et autres
Publié: (2024)

MCMComm: Hardware-Software Co-Optimization for End-to-End Communication in Multi-Chip-Modules
par: Raj, Ritik, et autres
Publié: (2025)

Marco: Configurable Graph-Based Task Solving and Multi-AI Agents Framework for Hardware Design
par: Ho, Chia-Tung, et autres
Publié: (2025)

AssertLLM: Generating Hardware Verification Assertions from Design Specifications via Multi-LLMs
par: Yan, Zhiyuan, et autres
Publié: (2024)

AssertLLM: Generating and Evaluating Hardware Verification Assertions from Design Specifications via Multi-LLMs
par: Fang, Wenji, et autres
Publié: (2024)

Swapping-Centric Neural Recording Systems
par: Ugur, Muhammed, et autres
Publié: (2024)