Multi-Plane Memory Write Segmentation for Read Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In memory sub-systems, multi-plane write operations often lead to increased latency and decreased performance due to the need to suspend all planes during large data writes, causing collisions with read requests and impacting quality of service.
Innovation Solution
Converting multi-plane write operations into multiple single-plane write operations performed in parallel, allowing each plane to be accessed separately, thus enabling concurrent read and write operations without suspending the entire write process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multi-plane write operations are used to write large data, then write throughput is improved, but request latency increases and quality of service decreases due to plane suspension
Solution Approach 1:
The patent segments the multi-plane write operation into multiple independent single-plane write operations. Each plane is written separately using individual write commands, allowing the memory subsystem to interleave read and write operations on different planes without suspending entire write processes. This segmentation resolves the contradiction by maintaining write throughput through parallel plane operations while reducing request latency by allowing concurrent read access to other planes.
2Productivity
If multi-plane write operations are used, then data write capability is improved, but collision with read requests increases
Solution Approach 1:
By dividing the write operation into plane-specific segments, the patent allows read requests to be serviced on planes not currently being written to. This reduces collisions between read and write operations while maintaining overall data write capability through parallel processing across multiple planes.
Solution Approach 2:
The patent dynamically manages plane access by transitioning between single-plane and multi-plane write modes based on operational requirements. The memory subsystem can selectively suspend individual planes for writing while maintaining read access to other planes, creating a dynamic system that adapts to reduce collisions rather than statically suspending all planes during writes.
3Reliability
If all planes are suspended during large data writes, then write operation completeness is ensured, but quality of service deteriorates
Solution Approach 1:
The patent ensures write operation completeness by implementing acknowledgment mechanisms for each individual plane write operation. Each plane write is tracked and confirmed independently, ensuring that all data is successfully written across all planes while allowing other planes to remain accessible for read operations, thus maintaining quality of service.
Solution Approach 2:
The patent maintains continuity of useful action by allowing read operations to continue on planes not currently being written. Instead of completely suspending all planes during writes, the system maintains useful read actions on available planes while performing write operations on specific planes, ensuring both write completeness and ongoing service quality.
Data Source
AI summary
A processing device in a memory system receives a request to read data stored on a first plane of a plurality of planes of a memory device while a plurality of write operations are ongoing, wherein each of the plurality of write operations are performed concurrently to write each of a plurality of single-plane segments of data to a corresponding plane of the plurality of planes of the memory device, and wherein a multi-plane segment of data received with a write request is divided into the plurality of single-plane segments of data. The processing device further suspends a first write operation of the plurality of write operations, the first write operation corresponding to the first plane, and performs a read operation to read the data stored on the first plane while continuing to perform at least one other write operation of the plurality of write operations corresponding to another plane of the plurality planes.