Safety-Rated Runnable Scheduling for Thermal-Limited AV Compute
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicle computing systems face challenges in ensuring robust safety integrity levels, particularly in handling system overloads, processing delays, and thermal issues, which are not adequately addressed by existing technologies.
Innovation Solution
The implementation of a scheduling program that associates runnables and their connections with safety ratings, allowing for adaptive degradation of the software structure. This includes reducing execution frequency, morphing or truncating the software structure, and switching between compute graphs, all managed by a functional safety program.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the system executes all runnables at full frequency to maximize productivity, then processing throughput is improved, but thermal issues and system overload worsen
Solution Approach 1:
The system dynamically adjusts the execution frequency of runnables based on real-time thermal conditions and system load. The scheduling program monitors temperature and workload metrics, then adaptively modifies execution frequencies to maintain productivity within thermal constraints, rather than using a static fixed-frequency approach
Solution Approach 2:
The system changes operational parameters (execution frequency, task priority, resource allocation) based on system state. When thermal issues or overload are detected, the scheduling program modifies parameters such as reducing frequency of non-critical runnables while maintaining critical tasks, thereby resolving the contradiction between throughput and thermal management
2Reliability
If the system degrades software structure to handle system overload, then reliability is improved, but productivity deteriorates
Solution Approach 1:
The system applies degradation selectively to specific runnables based on their safety ratings rather than uniformly degrading all tasks. Critical runnables with high safety ratings maintain full execution, while non-critical runnables are degraded or suspended, preserving reliability for essential functions while minimizing productivity loss
Solution Approach 2:
The system performs partial degradation by selectively reducing execution frequency or suspending only certain runnables rather than completely shutting down the system. This partial action maintains sufficient processing capability for critical tasks while providing enough reduction to handle overload conditions
3Temperature
If the system reduces execution frequency of runnables to manage thermal issues, then temperature is improved, but processing throughput worsens
Solution Approach 1:
The scheduling program dynamically changes execution frequency parameters based on thermal conditions. When temperature thresholds are exceeded, the system adjusts frequency parameters for specific runnables, creating a variable frequency schedule that manages thermal issues while maintaining optimal throughput within thermal constraints
Data Source
AI summary
A computing system can include a sensor data input chiplet to obtain sensor data from a sensor system, a set of workload processing chiplets, and a central chiplet comprising a shared memory including a scheduling program for scheduling a set of runnables to execute based at least partially on the sensor data, where the set of runnables are included in a software structure. the set of runnables can be executed by the set of workload processing chiplets. Each runnable and each connection between runnables in the set of runnables can be associated with a safety rating to facilitate degradation of execution of the software structure.