In-Storage Processing Controller for Repeated ISP Operations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory systems lack efficient mechanisms for in-storage processing (ISP) that can detect frequently performed operations and simplify them by storing and retrieving resulting data, leading to increased operation times for repeated tasks.
Innovation Solution
A memory system with a controller that checks operation information, performs first and second ISP operations, reads and processes data, and stores resulting data, allowing for the detection of frequently performed operations and reducing operation time by storing and retrieving processed data within the non-volatile memory device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the memory system performs full ISP operations (reading data, processing, generating resulting data, and storing) for every requested operation, then the operation can be completed accurately, but the operation time increases for frequently repeated tasks
Solution Approach 1:
The controller performs preliminary processing by executing the ISP operation once, storing the resulting data in the non-volatile memory device, and recording operation information. When the same operation is requested again, the controller retrieves the pre-stored resulting data directly, avoiding repeated processing steps. This preliminary action resolves the contradiction by ensuring accurate operation completion while significantly reducing operation time for repeated tasks.
Solution Approach 2:
The controller creates a copy of the resulting data from the first ISP operation and stores it in the non-volatile memory device. For subsequent identical operations, the controller retrieves this copied data instead of regenerating it. This copying mechanism maintains data accuracy while eliminating redundant processing, thus resolving the time versus reliability contradiction.
2Productivity
If the controller stores cumulative operation information for all requested ISP operations, then frequently repeated operations can be detected, but the storage space requirement increases
Solution Approach 1:
The controller extracts only the essential operation information (process address information and requested operation information) needed to identify and detect repeated operations, storing only this extracted data in the cumulative operation information storage space. This selective extraction resolves the contradiction by enabling effective detection of repeated operations while minimizing the storage space required.
Solution Approach 2:
The controller stores cumulative operation information with specific local qualities - only the necessary fields (process address information and requested operation information) are stored with sufficient detail to detect repetitions, rather than storing complete operation logs. This localized quality approach enables productivity improvement through repetition detection while constraining storage space usage.
3Loss of time
If the controller performs data processing and stores resulting data in the non-volatile memory device, then operation results are available for quick retrieval, but the processing complexity and time for first-time operations increases
Solution Approach 1:
The controller performs the data processing and resulting data storage as a preliminary action during the first execution of the ISP operation. Although this increases processing complexity and time for the first operation, it enables rapid retrieval for all subsequent identical operations. This preliminary action resolves the contradiction by trading initial processing overhead for significantly reduced retrieval time in repeated operations.
Solution Approach 2:
The controller prepares and stores the resulting data in advance during the first operation, creating a cushion of pre-processed data that can be quickly retrieved later. This beforehand cushioning resolves the contradiction by accepting increased initial processing complexity to enable fast retrieval operations subsequently.
Data Source
AI summary
A memory system includes: a non-volatile memory device; and a controller for checking operation information of a requested ISP operation, performing a first ISP operation when the requested ISP operation is not requested in the past, and performing a second ISP operation when the requested ISP operation is also requested in the past. During the first ISP operation, the controller may read out two or more requested data from the non-volatile memory device in response to the operation information of the requested ISP operation, generate resultant data by performing a predetermined operation to the read-out requested data, output the resultant data to a host, and store the resultant data in the non-volatile memory device. During the second ISP operation, the controller may read out the resultant data corresponding to the requested ISP operation from the non-volatile memory device, and output the read-out resultant data to the host.


