Non-Volatile Memory Swap Segmentation for Burst Mode
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile computing devices face challenges in quickly creating free volatile memory space due to limited capacity, leading to application termination and delays when handling unusually large data influxes, and returning to normal mode after burst mode operations incurs penalties on memory resources.
Innovation Solution
Implementing a computing device with a non-volatile memory having both multi-level cell (MLC) and single-level cell (SLC) swap areas, where data is moved to the SLC area during burst mode for faster swapping and then transferred to the MLC area upon returning to normal mode, reducing the need for application termination and minimizing memory stress.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is written directly into non-volatile memory at a high rate during burst mode with minimal use of volatile memory as a cache, then the device can handle unusually large data influxes without terminating applications, but penalties are incurred when returning to normal mode such as needing to move data from SLC to MLC and perform garbage collection operations
Solution Approach 1:
The non-volatile memory is segmented into two distinct swap areas: a first swap area with MLC memory for normal operations and a second swap area with SLC memory for burst mode operations. This segmentation allows the system to optimize for speed during burst mode by using SLC while avoiding the penalties of frequent SLC-MLC transitions and garbage collection by maintaining separate dedicated areas for each mode.
2Reliability
If the process of swapping out applications is used to create free space in volatile memory, then applications can be kept in memory to avoid re-launch delays, but the swap process takes too long when free space is needed very quickly
Solution Approach 1:
The system applies different quality characteristics to different parts of the non-volatile memory by using MLC for normal swap operations and SLC for burst mode swap operations. The SLC area provides locally optimized high-speed swap capability for critical burst scenarios, while the MLC area handles常规 operations with better capacity and efficiency.
3Loss of time
If volatile memory capacity is increased to store more applications, then application re-launch delays are reduced, but the device cost and power consumption increase
Solution Approach 1:
The system dynamically adapts its memory usage strategy based on operational mode. During burst mode, it utilizes the SLC swap area to quickly create volatile memory space without terminating applications. During normal mode, it uses the MLC swap area for efficient cache management. This dynamic adaptation allows the system to maintain low volatile memory capacity while achieving fast application availability when needed.
Data Source
AI summary
A method and system for improving swap performance are provided. In one embodiment, a computing device is provided with a volatile memory and a non-volatile memory, wherein the non-volatile memory has a first swap area with multi-level cell (MLC) memory and a second swap area with single-level cell (SLC) memory. One of the characteristics of SLC memory is that data is written more quickly in the SLC memory than the MLC memory. A determination is made whether the computing device is operating in normal mode or burst mode. If it is determined that the computing device is operating in normal mode, data is moved from the volatile memory to the first swap area during a swap operation. If it is determined that the computing device is operating in burst mode, data is moved from the volatile memory to the second swap area during a swap operation.