Shared Cooling for Thermally Linked CPU and GPU
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As microprocessors become smaller, faster, and more powerful, they generate more heat in a smaller space, leading to thermal management challenges, with existing systems being over-designed and inefficient in terms of cost and space due to separate thermal management solutions for each component, which can result in reduced performance and potential device failure.
Innovation Solution
A shared thermal management system where heat-generating components like CPUs and GPUs are thermally linked using phase change devices such as heat pipes, allowing one thermal management solution to cool both components, optimizing cooling capacity and reducing the overall thermal design power and system size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate thermal management solutions are used for each component (CPU, GPU), then each component can be cooled independently, but the system becomes over-designed with increased cost and space requirements
Solution Approach 1:
The patent merges separate thermal management solutions for CPU and GPU into a shared cooling system. The first and second thermal management devices are thermally coupled through a thermal coupling device, allowing them to share cooling capacity. This reduces overall system complexity and space requirements while maintaining reliable thermal management for both components.
Solution Approach 2:
The thermal management devices are designed to serve multiple functions: they can independently cool their respective components when needed, share cooling capacity through thermal coupling, and dynamically adjust power distribution based on thermal conditions. This multi-functionality eliminates the need for completely separate cooling systems.
2Reliability
If separate thermal management solutions are used for each component, then each component receives dedicated cooling, but the overall system size and cost increase
Solution Approach 1:
By thermally coupling the first and second thermal management devices, the patent creates a compact shared cooling system that occupies less space than separate independent cooling systems while maintaining reliable cooling for both CPU and GPU through dynamic heat redistribution.
3Reliability
If separate thermal management solutions are designed for worst-case scenarios, then each component is protected under maximum load, but the system is over-designed for infrequent peak conditions
Solution Approach 1:
The patent implements dynamic thermal management where the system continuously monitors temperatures and adjusts power distribution between CPU and GPU in real-time. The processor dynamically shifts workloads and adjusts clock speeds based on thermal conditions, ensuring reliable worst-case protection while avoiding over-design for infrequent peak scenarios. This dynamic approach optimizes cooling capacity utilization efficiency.
Solution Approach 2:
The system changes operational parameters (power distribution, clock speeds, workload allocation) based on thermal conditions. When thermal thresholds are approached, the processor dynamically adjusts these parameters to maintain reliable cooling while preventing over-design for rare peak conditions.
4Reliability
If power is limited to maintain thermal balance, then thermal management is maintained, but component performance and operating frequency are reduced
Solution Approach 1:
The patent implements dynamic power management where the processor continuously monitors thermal conditions and adjusts power distribution between CPU and GPU in real-time. This dynamic approach maintains thermal balance while maximizing overall system performance by shifting workloads rather than uniformly limiting power, thereby maintaining higher operating frequencies when thermally permissible.
Solution Approach 2:
The system dynamically changes power allocation parameters based on thermal feedback. When one component approaches thermal limits, the processor adjusts power distribution parameters to maintain thermal balance while preserving overall productivity through intelligent workload management rather than brute-force power limiting.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables efficient thermal management, minimizing cost and real estate while maintaining optimal performance by dynamically adjusting power distribution between components, allowing for higher frequency operation during intense workloads and extended device operation.
Implementation Method 1
At least one phase change device thermally links the components that generate the most heat
Implementation Method 2
For example, a heat pipe physically connects and thermally links the CPU and the GPU of the computing device
Implementation Method 3
The first thermal management solution cools not only the CPU, but also cools the GPU, and the thermal management solution for the GPU cools not only the GPU, but also cools the CPU
Data Source
AI summary
Cooling of at least two heat generating electronic components of a computing device is described herein. The computing device includes the at least two heat generating electronic components. The at least two heat generating electronic components includes a first heat generating electronic component and a second heat generating electronic component. The first heat generating electronic component and the second heat generating electronic component are in thermal communication. The computing device also includes a first sensor operable to measure a first temperature. The first temperature is associated with the first heat generating electronic component. The first heat generating electronic component, the second heat generating electronic component, another heat generating electronic component of the at least two heating generating electronic components, or any combination thereof is configured to compare the first measured temperature to a predetermined temperature, and decrease a power for the second heat generating electronic component based on the comparison.