CPU-GPU Power Control Under Acoustic and Battery Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-processor laptops, the allocation of processor power between the CPU and GPU is often arbitrary, leading to inefficient utilization of total available power and reduced overall computing performance due to mismatched power limits for the CPU and GPU, especially under varying workloads and ambient conditions.
Innovation Solution
A power control architecture with independent joint processor acoustic and power controllers dynamically adjusts power settings for the CPU and GPU to ensure efficient utilization of total available power, reallocating unused or inefficiently used power to maximize overall performance while complying with acoustic and battery drain rate limits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If processor power limits are selected based on best guess allocation between CPU and GPU, then device complexity is reduced, but productivity decreases due to inefficient power utilization
Solution Approach 1:
The system implements a feedback mechanism where the controller continuously monitors the operational status of both CPU and GPU, including their current power consumption and workload characteristics. Based on this feedback, the controller dynamically adjusts the power allocation limits for each processor to optimize overall system performance. This closed-loop control ensures that power limits are continuously adapted to match actual operational needs rather than relying on static pre-configured values.
Solution Approach 2:
The patent transforms the static power allocation approach into a dynamic system where power limits for CPU and GPU are not fixed but can be adjusted in real-time. The controller modifies the power consumption limits of the first processor based on the operational status of both processors, enabling the system to adapt to varying workload conditions, ambient temperature changes, and power availability, thereby maximizing computing performance under different operating scenarios.
2Productivity
If CPU is allocated more power than needed for GPU task provisioning, then computing performance increases, but energy efficiency decreases due to unused or wasted power
Solution Approach 1:
The controller uses feedback from the GPU's operational status and task completion rate to adjust CPU power allocation. When the GPU cannot keep up with CPU-generated tasks, the system reduces CPU power limits to prevent task accumulation and energy waste. Conversely, when the GPU has capacity, the CPU power limit is increased to maximize computational throughput. This feedback-driven adjustment eliminates the energy inefficiency of allocating excessive CPU power that goes unused.
Solution Approach 2:
The system dynamically changes the power consumption parameter of the CPU based on the matching rate between CPU task provisioning and GPU task execution. By adjusting the CPU power limit parameter in response to GPU capacity and workload conditions, the system optimizes the balance between computing performance and energy efficiency, ensuring that CPU power allocation closely matches actual computational needs without significant waste.
3Productivity
If processor power limits are dynamically adjusted based on operational status, then productivity increases through optimized power utilization, but device complexity increases due to additional control mechanisms
Solution Approach 1:
The controller is designed as a multi-functional component that simultaneously performs multiple tasks: monitoring operational status of both processors, determining optimal power allocation, adjusting power limits in real-time, and adapting to various environmental conditions. This universal controller consolidates what could be multiple separate control systems into a single integrated unit, reducing overall system complexity while maintaining the benefits of dynamic power adjustment.
Solution Approach 2:
The patent merges the power management functions for both CPU and GPU into a unified control system. Instead of having independent power management mechanisms for each processor, the system combines their control under a single controller that coordinates power allocation based on the joint operational status of both processors. This merging reduces the complexity of having separate control loops while optimizing overall system performance through coordinated power management.
Data Source
AI summary
A computer-implemented method of controlling power consumption in a multi-processor computing device comprises determining a first value for a first power setting associated with a first processor based on a sound level generated by the multi-processor computing device; determining a second value for the first power setting based on a power consumption level of the multi-processor computing device; comparing the first value to the second value; and causing the first processor to perform one or more operations based on the lesser of the first value and the second value.