SSD Controller Parallel Read Synchronization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The performance of read operations in data storage devices like solid state drives (SSDs) is impacted due to differing speeds of data retrieval from parallel planes, leading to increased processing time as the data stored on pages of different planes may not be similar, resulting in inefficient parallel read operations.
Innovation Solution
A controller in the data storage device determines the decode times for pages across different planes and performs parallel reads on pages with similar decode times, optimizing the read operation by synchronizing the data transfer based on the closest match in decode times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If parallel reads are performed on pages from different planes simultaneously, then read throughput is improved, but read performance deteriorates because pages with different decode times cannot be processed at the same speed
Solution Approach 1:
The patent segments the parallel read operation into multiple phases based on decode time groups. Pages are divided into different decode time groups (first group, second group, etc.), and reads are performed in sequential phases where each phase processes one group. This segmentation allows the system to maintain parallelism within each phase while avoiding the performance penalty of mixing pages with different decode times, thus resolving the contradiction between throughput and time loss.
Solution Approach 2:
The controller performs preliminary classification of pages into decode time groups before executing parallel reads. By determining decode times in advance and organizing pages into groups with similar decode characteristics, the system prepares the read operation to proceed efficiently without encountering speed mismatches during execution. This preliminary action enables optimized parallel processing while maintaining performance.
2Device complexity
If all pages from multiple planes are read in parallel without considering decode times, then device complexity is reduced, but read performance deteriorates due to speed mismatches between planes
Solution Approach 1:
The patent changes the parameter of page organization from arbitrary grouping to decode-time-based grouping. By using decode time as the organizing parameter, the controller can systematically categorize pages into groups with similar processing characteristics. This parameter change enables more efficient parallel read operations without requiring complex device logic, as the grouping criterion is straightforward and can be implemented with simple comparison operations.
Data Source
AI summary
A data storage device includes a memory device and a controller coupled to the memory device. The controller is configured to receive read requests from a host device. When a read request is received corresponding to one or more pages from a first plane and one or more pages from a second plane, the controller is configured to determine a decode time for the pages of the first plane and the pages of the second plane. Based on the decode times for pages of the first plane and pages of the second page, pages of the first plane that have a similar decode time to pages of the second plane are read in parallel.