YoVDO

HexaMesh: Scaling to Hundreds of Chiplets with an Optimized Chiplet Arrangement

Offered By: Scalable Parallel Computing Lab, SPCL @ ETH Zurich via YouTube

Tags

Network Topologies Courses Chiplets Courses

Course Description

Overview

Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore a conference talk on HexaMesh, an innovative chiplet arrangement technique for 2.5D integration. Dive into the challenges of scaling to hundreds of chiplets and optimizing inter-chiplet interconnects. Learn about the theoretical and practical advantages of HexaMesh over traditional grid arrangements, including reduced network diameter, improved bisection bandwidth, lower latency, and enhanced throughput. Understand the background of 2.5D integration, key insights, problem statement, and evaluation methods. Gain valuable knowledge on advanced packaging techniques and their impact on large-scale chiplet designs with high-performance interconnects.

Syllabus

Introduction
Background: 2.5D Integration
Key Insights & Problem Statement
Optimizing the Chiplet Arrangement
Evaluation: Theoretical Analysis
Shortcomings of Theoretical Analysis
Evaluation: Simulation Results
Conclusion


Taught by

Scalable Parallel Computing Lab, SPCL @ ETH Zurich

Related Courses

Evaluating Chiplet-based Large-Scale Interconnection Networks via Cycle-Accurate Packet-Parallel Simulation
USENIX via YouTube
A Chiplet-Based Generative Inference Architecture with Block Floating Point Datatypes
Scalable Parallel Computing Lab, SPCL @ ETH Zurich via YouTube
Post-Moore Spatial Computing: From Chips to Clusters
Scalable Parallel Computing Lab, SPCL @ ETH Zurich via YouTube
Berkeley Lab's Breakthroughs in Exascale Supercomputing and AI Energy Efficiency
SAIConference via YouTube