SoC Power Budget Partitioning for Thermal Workload Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional thermal solutions struggle to manage heat generated by computing devices under heavy workloads, leading to potential thermal damage and constrained computational resources.
Innovation Solution
A method and system for dynamically distributing a power budget among compute agents within a computing device, using performance metrics to partition and allocate power budgets based on the energy consumption of consumer and producer compute agents, with a producer compute agent adjusting power coefficients to optimize heat management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Temperature
If the thermal solution uses conventional cooling systems to manage heat, then the computing device can operate under normal workloads, but the thermal solution cannot effectively manage heat when the computing device operates under heavy workloads
Solution Approach 1:
The patent implements dynamic power budget partitioning that adjusts power allocation to compute agents in real-time based on thermal conditions and workload requirements. The system transitions from static thermal management to dynamic control, where the power budget is continuously re partitioned among compute agents according to current thermal state and performance needs, enabling effective heat management during heavy workloads
Solution Approach 2:
The system changes the power consumption parameter of compute agents dynamically by adjusting power budgets based on thermal conditions. When thermal thresholds are approached, the system modifies power allocation parameters to reduce heat generation while maintaining computational productivity, resolving the contradiction between temperature control and performance
2Reliability
If the computing device constrains computational resources to manage heat, then thermal damage is prevented, but workload performance is constrained
Solution Approach 1:
The patent segments the total power budget into separate power budgets for individual compute agents. This segmentation allows selective power allocation to specific compute agents based on their thermal contribution and workload importance, preventing thermal damage to critical components while maintaining high performance in non-critical areas
Solution Approach 2:
The system applies different power budget allocations to different compute agents based on their local thermal characteristics and workload requirements. Each compute agent receives a customized power budget that reflects its specific thermal impact and performance needs, rather than applying uniform constraints across all computational resources
3Temperature
If the thermal solution adds more cooling components to handle heavy workload heat, then heat management capability improves, but device complexity increases
Solution Approach 1:
The patent replaces mechanical/physical cooling systems with a software-based power management system. Instead of adding more physical cooling components, the system uses intelligent power budget partitioning and control algorithms to manage thermal conditions, substituting mechanical complexity with computational control
Data Source
AI summary
Systems, methods, and apparatuses disclosed herein can advantageously leverage a power budget to manage heat. These systems, methods, and apparatuses can dynamically distribute the power budget to manage the heat. As part of this dynamic distribution, these systems, methods, and apparatuses can monitor their operation. In some embodiments, when this monitoring indicates that workloads are being created faster than being performed, these systems, methods, and apparatuses can be distributed more of the power budget to optimize the efficiency in performing these workloads. On the other hand, these systems, methods, and apparatuses can be distributed less of the power budget to optimize the efficiency in creating these workloads when this monitoring indicates workloads are being performed faster than being created.