Concurrent Search for Last Written Page in Non-Volatile Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Non-volatile semiconductor memory systems face challenges in determining the last written block during power failures, as the write log may not be saved, leading to inefficiencies in locating the last written page due to extensive searches across multiple memory devices.
Innovation Solution
Implementing a method where read commands are concurrently issued to multiple memory devices with addresses separated by a determined distance and search range, iteratively updating these parameters to expedite the search for the last written page, ensuring unique addresses modulo N are sent to each device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a sequential search is performed across multiple memory devices to locate the last written page, then the search can be exhaustive and reliable, but the search time and processing duration increase significantly
Solution Approach 1:
The search space across multiple memory devices is divided into segments, with each device being searched independently through parallel read operations. The address space is segmented and distributed across N devices, allowing simultaneous search operations on different segments without sequential dependency.
Solution Approach 2:
Multiple sequential search operations across different memory devices are merged into a single parallel search operation. By issuing read commands to multiple devices simultaneously and evaluating results concurrently, the search process combines multiple operations into one unified parallel execution, reducing total search time while maintaining comprehensive coverage.
2Measurement precision
If read commands are issued to all memory devices sequentially, then each device can be searched completely, but the total number of operations and system complexity increase
Solution Approach 1:
The control circuitry pre-calculates and determines optimal read addresses for multiple memory devices before executing the search. By determining addresses in advance based on search range and device count, the system prepares parallel read commands that maximize search efficiency without requiring complex real-time decision-making during execution.
Solution Approach 2:
The search parameters including read addresses, search range, and distance between addresses are dynamically adjusted based on feedback from previous search iterations. The control circuitry modifies these parameters adaptively to optimize the search process, reducing complexity by automating parameter adjustment rather than requiring manual intervention.
3Reliability
If the search range is extended to cover all possible addresses, then the likelihood of finding the last written page increases, but the search duration and energy consumption increase
Solution Approach 1:
The search process uses periodic read operations at strategically determined intervals across the address space. By performing reads at periodic intervals rather than continuously scanning every address, the system maintains reliable detection while reducing total operation count and energy consumption through efficient sampling of the address space.
Data Source
AI summary
A non-volatile semiconductor memory is disclosed comprising N memory devices each comprising a plurality of blocks, wherein each block comprises a plurality of memory segments accessed through an address. A searched is performed by issuing a read command for each of the N memory devices, wherein an address of each read command is separated by a distance determined in response to the search range of addresses and N, and the search range of addresses is greater than N. Data read from at least one of the memory devices is evaluated to determine whether the search has finished.