Abstract:AbstractHierarchical organization is widely used in high-radix routers to enable efficient scaling to higher switch port count. A general-purpose hierarchical router must be symmetrically designed with the same input buffer depth, resulting in a large amount of unused input buffers due to the different link lengths. Sharing input buffers between different input ports can improve buffer utilization, but the implementation overhead also increases with the number of shared ports. Previous work allowed input buffers to be shared among all router ports, which maximizes the buffer utilization but also introduces higher implementation complexity. Moreover, such design can impair performance when faced with long packets, due to the head-of-line blocking in intermediate buffers.In this work, we explain that sharing unused buffers between a subset of router ports is a more efficient design. Based on this observation, we propose Centralized Input Buffer Design in Hierarchical High-radix Routers (CIB-HIER), a novel centralized input buffer design for hierarchical high-radix routers. CIB-HIER integrates multiple input ports onto a single tile and organizes all unused input buffers in the tile as a centralized input buffer. CIB-HIER only allows the centralized input buffer to be shared between ports on the same tile, without introducing additional intermediate virtual channels or global scheduling circuits. Going beyond the basic design of CIB-HIER, the centralized input buffer can be used to relieve the head-of-line blocking caused by shallow intermediate buffers, by stashing long packets in the centralized input buffer. Experimental results show that CIB-HIER is highly effective and can significantly increase the throughput of high-radix routers.

An adaptive buffer regulating scheme for on-chip routers

A multi-path routing method with shared buffer on NoC

An adaptive buffer-shared routing algorithm for NoC designs

Optimizing Buffer Usage For Networks-On-Chip Design

Buffer Planning for Application-Specific Networks-on-chip Design.

Design and Implementation of Single-Buffered Routers

Evaluation of buffer organizations for network-on-chip

Achieving balanced buffer utilization with a proper co-design of flow control and routing algorithm

DBU-PG: energy-efficient noc design using dual-buffering power gating

Improving Performance of Wdm Local Packet-Switching Router with Single Buffer

Interactive buffer sizing for core routers

CRAFT: Common Router Architecture for Throughput Optimization.

A Hybrid Router Combining Circuit Switching and Packet Switching with Virtual Channels for On-Chip Networks

Reducing Power of Traffic Manager in Routers Via Dynamic On/off-Chip Scheduling

Defect-tolerant routing algorithm for low power NoCs based on buffer-shared router architecture

Pull-off Buffer: Borrowing Cache Space to Avoid Deadlock for Fault-Tolerant NoC Routing

Structure Design of Low-Latency Extensible Router in Network on Chip

Timing-driven Global Routing with Efficient Buffer Insertion

A Tightly Coupled Network-on-Chip Router Architecture

CIB-HIER: Centralized Input Buffer Design in Hierarchical High-radix Routers

Performance Enhancement of Routers in Networks-on-Chip Using Dynamic Virtual Channels Allocation