NoC Router Fault-Tolerant Mode Switching for Energy and Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fault-tolerant methods for network-on-chip (NoC) systems in parallel computing architectures face challenges in balancing performance, energy efficiency, and reliability, as they often result in excessive power consumption and performance degradation due to reactive and proactive techniques.
Innovation Solution
A dynamic router assembly with power-gating capabilities and machine learning algorithms, such as reinforcement learning, to optimize error handling configurations and operation modes, enabling the independent powering ON/OFF of error handling hardware and dynamically switching between different error handling strategies to balance performance, energy efficiency, and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If error handling hardware (ARQ+ECC encoder/decoder) is constantly enabled, then reliability is improved, but power consumption increases and performance degrades
Solution Approach 1:
The patent implements dynamic power gating of error handling hardware (ARQ+ECC encoder/decoder) based on real-time channel quality assessment. The router controller monitors channel conditions and dynamically switches the error handling hardware between ON and OFF states, enabling the system to adapt to varying reliability requirements while minimizing power consumption when full error handling is not needed.
Solution Approach 2:
The system changes the operational parameters of error handling hardware by adjusting power states (ON/OFF) based on channel quality metrics. When channel quality is good, the error handling hardware is powered OFF; when channel quality degrades, the hardware is powered ON, thus changing the system parameters to match environmental conditions and optimize the trade-off between reliability and power consumption.
2Use of energy by moving object
If error handling hardware is indiscriminately disabled, then power consumption is reduced, but error rate increases and retransmission traffic increases
Solution Approach 1:
The router controller implements a feedback mechanism that continuously monitors channel quality indicators and dynamically adjusts the power state of error handling hardware accordingly. This closed-loop control ensures that error handling capabilities are maintained when channel conditions indicate potential errors, while allowing power savings when conditions are favorable, thus preventing excessive error rates and retransmissions.
Solution Approach 2:
The system performs preliminary assessment of channel quality before transmitting data, and proactively enables or disables error handling hardware based on predicted channel conditions. This preliminary action prevents errors before they occur by ensuring error handling capabilities are available when channel quality deteriorates, rather than reacting after errors have already occurred.
3Reliability
If powerful error handling codes (ARQ+ECC) are applied, then reliability is improved, but latency increases and power consumption increases
Solution Approach 1:
The patent implements dynamic switching between different error handling configurations (ARQ+ECC enabled/disabled) based on real-time channel quality assessment. When channel conditions are good, the system disables error handling codes to minimize latency; when conditions deteriorate, the system enables the codes to maintain reliability, thus dynamically balancing latency and reliability requirements.
4Reliability
If reactive error detection/correction is used, then reliability is maintained, but excessive power consumption and performance degradation occur
Solution Approach 1:
The system performs preliminary assessment of channel quality before data transmission and proactively configures error handling hardware states accordingly. This prevents the need for reactive error detection and correction during transmission, as the appropriate error handling capabilities are already in place based on predicted channel conditions, thus maintaining reliability while avoiding unnecessary power consumption and performance degradation.
5Reliability
If proactive fault mitigation techniques are used, then reliability is improved, but chip area and power consumption increase due to additional hardware and rules
Solution Approach 1:
The patent implements a dynamic configuration system that adjusts error handling hardware states (ON/OFF) based on real-time channel quality assessment. This allows the system to achieve proactive fault mitigation when needed while avoiding the continuous overhead of always-on error handling hardware, thus reducing chip area requirements and power consumption while maintaining reliability when required.
Data Source
AI summary
A proactive fault-tolerant scheme which improves performance and energy efficiency for NoCs. The fault-tolerant scheme allows routers to switch among several different fault-tolerant operations. Each operation mode has different trade-offs among fault-tolerant capability, retransmission traffic, latency, and energy efficiency. Another example provides a proactive, dynamic control policy to balance and optimize the dynamic interactions and trade-offs. The example control policy uses example machine learning algorithm called reinforcement learning (RL). The example RL-based controller independently observes a set of NoC system parameters at runtime, and over time they evolve optimal per-router control policies. By automatically and optimally switching among the four fault-tolerant modes, the trained control policy results in minimizing system level network latency and maximizing energy efficiency while detecting and correcting errors.