Dynamic Power Management for Memory Macro Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In embedded systems, unused memory blocks continue to consume power in active and deep sleep modes, reducing battery life due to inefficient memory allocation and freeing processes.
Innovation Solution
Dynamically power on or off memory macro blocks based on memory requirements using memory allocation and freeing functions, and update heap management structures to reflect the new memory layout.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If all RAM macro blocks are powered on at system boot, then memory allocation can be performed immediately without delay, but unused memory blocks continue to consume power in active and deep sleep modes, reducing battery life
Solution Approach 1:
The patent implements dynamic power management for RAM macro blocks by introducing a power state control mechanism that transitions blocks between awake and sleep states based on allocation status. The memory controller dynamically adjusts power supply to individual macro blocks, powering them on only when allocation is needed and powering them off when unused, thereby resolving the contradiction between immediate availability and power consumption.
Solution Approach 2:
The patent applies local quality by treating different macro blocks with different power states based on their usage status. Instead of uniformly powering all blocks, the system selectively maintains power only to blocks that are currently allocated or about to be allocated, while placing unused blocks in a low-power state. This localized differentiation resolves the contradiction by optimizing power consumption at the block level rather than system level.
2Use of energy by moving object
If memory blocks are dynamically powered on and off based on allocation status, then power consumption is reduced, but system complexity increases due to additional power management mechanisms
Solution Approach 1:
The patent merges the power management function with the existing memory controller and heap management structures. The power state transitions are integrated into the memory allocation and freeing operations, so that the same control logic that manages memory also manages power states. This consolidation reduces overall system complexity by eliminating separate power management hardware while achieving dynamic power savings.
Solution Approach 2:
The patent implements self-service by making the power management system automatically respond to allocation events without external intervention. When the heap manager allocates or frees memory, it automatically triggers corresponding power state transitions in the macro blocks. This automated feedback loop eliminates the need for complex external power management control while achieving efficient power consumption.
3Speed
If memory blocks remain in retention state during deep sleep, then quick wake-up is possible, but power consumption increases compared to complete power off
Solution Approach 1:
The patent dynamically selects between retention and complete power-off states based on the allocation status of macro blocks. Allocated blocks remain in retention state to enable fast wake-up, while unallocated blocks are completely powered off to minimize consumption. This dynamic state selection resolves the contradiction by optimizing the trade-off between wake-up speed and power consumption on a per-block basis.
Data Source
AI summary
A method of memory allocation is disclosed. An allocation size of memory to allocate is determined. A first memory block is searched for based on the allocation size, where the first memory block is a free memory block. When the first memory block is not found, a first macro block in a memory device is searched for, where the first macro block is an available macro block. When the first macro block is found, the first macro block is powered on, and the first macro block is marked as occupied.


