Skip to Content
Volume

The Neural Highway

Architecting High-Performance Network-on-Chip for Tensor Processing Units

Data movement is the silent killer of AI performance.

Strategic Objectives

• Master the architecture of high-bandwidth Network-on-Chip (NoC) designs.

• Optimize data flow to eliminate latency in large-scale tensor operations.

• Explore advanced routing algorithms tailored for neural workloads.

• Bridge the gap between raw compute power and real-world throughput.

The Core Challenge

As NPU compute power scales, the inability to move massive tensors efficiently creates a communication bottleneck that leaves processing elements idle.

01

The Evolution of NPUs

02

The Interconnect Imperative

03

NoC Fundamentals

04

Topological Design

05

Tensor Data Movement

06

Routing Algorithms

07

Flow Control Mechanisms

08

Switching Strategies

09

Router Microarchitecture

10

Arbitration and Allocation

11

Memory Hierarchies in NPUs

12

Bandwidth Optimization

13

Latency Mitigation

14

Energy-Efficient Interconnects

15

Synchronization and Coherency

16

Quality of Service (QoS)

17

Fault Tolerance and Reliability

18

3D Integration and TSVs

19

Photonic Interconnects

20

Modeling and Simulation

21

The Road Ahead for NoCs

Available eBook Editions