Gemini: Mapping and Architecture Co-exploration for Large-scale DNN Chiplet Accelerators

2023/12/27 by Jingwei Cai, Cai, Jingwei, Zuotong Wu +13 · 10 citations
Computer Science · Engineering · #3D IC and TSV technologies #FOS: Computer and information sciences #Hardware Architecture (cs.AR) #Interconnection Networks and Systems #Low-power high-performance VLSI design

paper · pdf · doi:10.48550/arxiv.2312.16436

openalex publication_date 2023/12/27 · openalex created_date 2023/12/30 · openalex updated_date 2026/07/28

Abstract

Chiplet technology enables the integration of an increasing number of transistors on a single accelerator with higher yield in the post-Moore era, addressing the immense computational demands arising from rapid AI advancements. However, it also introduces more expensive packaging costs and costly Die-to-Die (D2D) interfaces, which require more area, consume higher power, and offer lower bandwidth than on-chip interconnects. Maximizing the benefits and minimizing the drawbacks of chiplet technology is crucial for developing large-scale DNN chiplet accelerators, which poses challenges to both architecture and mapping. Despite its importance in the post-Moore era, methods to address these challenges remain scarce.

Cited by

• TEMP: A Memory Efficient Physical-aware Tensor Partition-Mapping Framework on Wafer-scale Chips
• WATOS: Efficient LLM Training Strategies and Architecture Co-exploration for Wafer-scale Chip
• LaMoSys3.5D: Enabling 3.5D-IC-Based Large Language Model Inference Serving Systems via Hardware/Software Co-Design
• Compass: Co-Exploration of Mapping and Hardware for Heterogeneous Multi-Chiplet Accelerators Targeting LLM Inference Service Workloads
• Hemlet: A Heterogeneous Compute-in-Memory Chiplet Architecture for Vision Transformers with Group-Level Parallelism
• SLOTH: Lightweight Detection and Localization of On-Chip Fail-Slow Failures for DNN Accelerators
• Mozart: A Chiplet Ecosystem-Accelerator Codesign Framework for Composable Bespoke Application Specific Integrated Circuits
• Embodied AI: From LLMs to World Models
• Characterizing LLM Kernel Access and Memory Interaction in Multi-Partition NUMA GPUs
• FlatAttention: Dataflow and Fabric Collectives Co-Optimization for Efficient Multi-Head Attention on Tile-Based Many-PE Accelerators

Related

• Mozart: A Chiplet Ecosystem-Accelerator Codesign Framework for Composable Bespoke Application Specific Integrated Circuits
• Multi-Objective Hardware-Mapping Co-Optimisation for Multi-DNN Workloads on Chiplet-based Accelerators
• Hecaton: Training Large Language Models with Scalable Chiplet Systems
• Gemmini: Enabling Systematic Deep-Learning Architecture Evaluation via Full-Stack Integration
• HADES: Hardware/Algorithm Co-design in DNN accelerators using Energy-efficient Approximate Alphabet Set Multipliers