Demand-Based Probe Filter Initialization for SoC Power States
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Implementing chip-wide low-power states in systems-on-chip (SOCs) is challenging due to the diverse and isochronous nature of peripheral controllers requiring consistent bandwidth, which conflicts with the idle periods of CPUs and GPUs, leading to inefficient power management.
Innovation Solution
A data fabric in the SOC delays the initialization of a probe filter into a low-power state until all requestors are operational, allowing efficient power management by maintaining the probe filter's initialization state only when necessary, thereby optimizing power consumption during idle periods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If the data fabric enters a low-power state to save energy during CPU/GPU idle periods, then power consumption is reduced, but the probe filter cannot properly service requests from peripheral controllers when they need to transfer data
Solution Approach 1:
The probe filter is designed to dynamically transition between operational and low-power states based on the activity status of requestors. When all requestors are idle, the probe filter enters a low-power state; when any requestor becomes active, the probe filter transitions back to operational state to service data transfer requests. This dynamic state adjustment resolves the contradiction by adapting power consumption to actual operational needs.
Solution Approach 2:
The system performs preliminary detection of requestor activity status before transitioning the probe filter to low-power state. The data fabric monitors whether all requestors are in an operational state, and only when confirmed that all are idle does it allow the probe filter to enter low-power mode. This preliminary check ensures that data transfer capability is preserved when needed while enabling power savings when appropriate.
2Reliability
If the probe filter remains in initialization state continuously to ensure data integrity, then data coherence is maintained, but power consumption increases during periods when no data transfer is needed
Solution Approach 1:
The probe filter operates in periodic cycles, alternating between operational state (when data transfer is needed) and low-power state (when no transfer is needed). The state transitions are triggered by periodic monitoring of requestor activity, creating a rhythm of active and dormant periods that maintains data integrity when necessary while conserving power during idle intervals.
3Speed
If the data fabric maintains full operational state to quickly service peripheral controller requests, then data transfer speed is improved, but power consumption increases during CPU/GPU idle periods
Solution Approach 1:
The data fabric implements dynamic power management where the probe filter's operational state is adjusted in real-time based on requestor activity. When peripheral controllers need to transfer data, the probe filter is in operational state providing fast service; when all requestors are idle, the probe filter transitions to low-power state. This dynamic adaptation resolves the speed-power contradiction by providing high performance only when needed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A data fabric routes requests between the plurality of requestors. A probe filter tracks the state of cached lines of memory at a probe filter coupled to the data fabric. Responsive to the data fabric leaving a non-operational power state while all requestors that are probe filter clients are in a non-operational power state, the power management controller delays a probe filter initialization state in which data regarding cached lines is initialized following the non-operational power state.