QoS Buffer Allocation in Network on Chip Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Network on Chip (NoC) systems, optimizing endpoint bandwidth allocation among multiple agents is challenging due to scalability limitations and varying bandwidth demands, leading to potential bottlenecks and traffic congestion.
Innovation Solution
Implementing Quality of Service (QoS) policies and handshake protocols that utilize credits for buffer allocation, prioritizing messages, and dynamically managing buffer allocation based on availability and priority, ensuring efficient data transmission and congestion reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional bus or crossbar interconnects are used, then simple architecture is maintained, but scalability is limited and bandwidth allocation cannot meet varying demands
Solution Approach 1:
The interconnect is segmented into multiple virtual channels (VCs) that operate independently, allowing different traffic types to be routed through different VCs. This segmentation enables fine-grained bandwidth allocation and QoS management without requiring complete architectural redesign, thus improving adaptability while controlling complexity.
Solution Approach 2:
The system dynamically allocates buffer credits to different agents based on their bandwidth demands and priority levels. Buffer allocation is not static but adapts in real-time to traffic conditions, allowing the system to meet varying bandwidth demands while maintaining a relatively simple underlying interconnect structure.
2Productivity
If buffer credits are allocated to all agents, then bandwidth demand is met, but buffer starvation occurs and congestion increases
Solution Approach 1:
The system changes the parameter of buffer credit allocation from uniform to differentiated based on agent priority and traffic type. High-priority agents receive more credits while low-priority agents receive fewer, preventing buffer starvation for critical traffic while avoiding over-allocation that would cause congestion. This parameter change enables both high bandwidth utilization and reliable buffer management.
3Ease of operation
If manual buffer management is used, then fine control is achieved, but system complexity increases and automation is reduced
Solution Approach 1:
Agents autonomously manage their own buffer credits by requesting and receiving credits based on their bandwidth needs and priority levels. The system automatically arbitrates between competing agents and allocates credits without requiring manual intervention or complex centralized control, thus improving ease of operation while keeping the control mechanism relatively simple through decentralized decision-making.
Data Source
AI summary
The present disclosure is directed to Quality of Service (QoS) and handshake protocols to facilitate endpoint bandwidth allocation among one or more agents in a Network on Chip (NoC) for an endpoint agent. The QoS policy and handshake protocols may involve the use of credits for buffer allocation which are sent to agents in the NoC to compel the acceptance of data and the allocation of an appropriate buffer. Messages sent to the agent may also have a priority associated with the message, wherein higher priority messages have automatic bandwidth allocation and lower priority messages are processed using a handshake protocol.