Masking CPU Load via Micro-Sleeps to Reduce Power
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Processor performance management systems consume excessive energy when handling computationally-intensive tasks that are not urgent, as they select high-performance states due to high CPU utilization, leading to increased power consumption and thermal stress, even when tasks can tolerate longer execution times.
Innovation Solution
The approach involves masking the computational load of non-urgent tasks from the processor performance management algorithm by executing them in short active periods with regular micro-sleeps, reducing the average CPU load and preventing the selection of high-performance states, allowing for lower energy consumption and reduced thermal stress.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the processor performance management system selects high-performance states based on high CPU utilization, then the task execution speed is improved, but the power consumption increases
Solution Approach 1:
The patent applies periodic action by introducing micro-sleep intervals between active computation periods. The process periodically wakes up to execute a small amount of computational work, then enters a sleep state. This periodic execution pattern allows the processor to remain in low-power states while still making progress on computationally-intensive tasks, resolving the contradiction between execution speed and power consumption.
Solution Approach 2:
The patent implements dynamics by making the process execution pattern adaptive and flexible. The micro-sleep technique dynamically adjusts the balance between active computation and idle periods based on system conditions. The process can modify its execution behavior to prioritize either speed or power savings depending on circumstances, allowing the system to adaptively resolve the contradiction between these two parameters.
2Loss of time
If the processor performance management system maintains high-performance states due to high CPU utilization, then the task completion time is reduced, but the thermal stress increases
Solution Approach 1:
The micro-sleep technique uses periodic action to interrupt continuous computation with regular sleep intervals. During these sleep periods, the processor reduces its activity and generates less heat. This periodic on-off pattern allows thermal dissipation during idle periods while still completing the computational task over time, resolving the contradiction between completion time and thermal stress.
Solution Approach 2:
The patent maintains continuity of useful action by ensuring that computation progresses steadily through repeated short active periods. Although the execution is fragmented into micro-sleep cycles, the cumulative effect ensures the task completes without unnecessary delays. This continuous progress approach minimizes total completion time while the intermittent nature of execution reduces thermal stress.
3Productivity
If the process executes continuously to complete the task quickly, then the productivity is improved, but the power consumption increases
Solution Approach 1:
The micro-sleep technique implements periodic action by alternating between short computation bursts and sleep periods. This creates a rhythmic execution pattern where the process works intensively for brief periods then rests. The periodic nature allows the system to maintain productivity through cumulative progress while dramatically reducing average power consumption during sleep intervals.
Solution Approach 2:
The patent applies parameter changes by modifying the execution parameters of the process. Instead of running continuously at full speed, the process changes its operational parameters to include intermittent sleep states. This parameter modification transforms the execution profile from continuous high-power operation to periodic low-power operation, resolving the contradiction between productivity and power consumption.
Data Source
Figure 1(a)~1(b)
Figure 2~3
Figure 4(a)~4(b)
AI summary
Energy savings can be obtained by masking a computationally-intensive task from a processor performance management system which selects the processor performance state based on the load on the processor (CPU). By preventing the PPM system from reacting to the computational load the application places on the processor, the time to complete execution of the application increases but the energy used by the application may be greatly reduced and thermal stress on the CPU is also reduced (preventing noisy fans from operating). This approach makes it convenient to run a computationally intensive task as a background task. The masking can be achieved by running the task in tiny bursts, with micro-sleeps in between them, so that the average CPU load is low over a time period that the PPM system uses for measuring CPU activity/load.