Selective GPU Throttling for Thermal Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing devices manage thermal load by globally reducing GPU operating frequency, which can impact performance and lead to shutdowns, as they do not differentiate between high and low priority processes, thereby affecting user interface and batch processes equally.
Innovation Solution
A method and apparatus that selectively throttle GPU operations based on thermal thresholds, prioritizing high priority processes while reducing utilization for low priority processes, using a thermal daemon to adjust GPU utilization levels and manage thermal profiles dynamically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Temperature
If global GPU frequency reduction is applied to manage thermal load, then thermal load is reduced, but device performance deteriorates and shutdown risk increases
Solution Approach 1:
The patent applies local quality by differentiating between high-priority and low-priority processes, applying selective throttling only to low-priority processes while maintaining full performance for high-priority processes. This resolves the contradiction by locally reducing thermal load from low-priority processes without globally degrading device performance.
Solution Approach 2:
The patent segments GPU processes into high-priority and low-priority categories, managing thermal load by independently controlling each segment. This allows the system to reduce thermal load from the low-priority segment while preserving performance in the high-priority segment, avoiding the shutdown risk associated with uniform global reduction.
2Temperature
If global GPU frequency reduction is applied to manage thermal load, then thermal load is reduced, but device shutdown risk increases
Solution Approach 1:
The patent applies local quality by differentiating between high-priority and low-priority processes, applying selective throttling only to low-priority processes while maintaining full performance for high-priority processes. This resolves the contradiction by locally reducing thermal load from low-priority processes without globally degrading device performance.
Solution Approach 2:
The patent implements beforehand cushioning by establishing priority-based throttling thresholds that prevent thermal load from reaching critical shutdown levels. By proactively managing thermal load through selective throttling of low-priority processes, the system cushions against the risk of thermal shutdown before it occurs.
3Productivity
If selective GPU throttling is applied to differentiate process priorities, then high priority process performance is maintained, but system complexity increases
Solution Approach 1:
The patent segments GPU processes into high-priority and low-priority categories, managing thermal load by independently controlling each segment. This allows the system to reduce thermal load from the low-priority segment while preserving performance in the high-priority segment, avoiding the shutdown risk associated with uniform global reduction.
Solution Approach 2:
The patent implements self-service by enabling the thermal management system to automatically identify and throttle low-priority processes based on predefined criteria, without requiring manual intervention. The system self-regulates by monitoring process priorities and applying appropriate throttling levels, reducing the operational complexity of managing selective throttling.
Data Source
AI summary
A method and apparatus of a device that manages a thermal profile of a device by selectively throttling graphics processing unit operations of the device is described. In an exemplary embodiment, the device monitors the thermal profile of the device, where the device executes a plurality of processes that utilizes a graphics processing unit of the device. In addition, the plurality of processes include a high priority process and a low priority process. If the thermal profile of the device exceeds a thermal threshold, the device decreases a first GPU utilization for the low priority process and maintains a second GPU utilization for the high priority process. The device further executes the low priority process using the first GPU utilization with the GPU and executes the high priority process using the second GPU utilization with the GPU.