Memory Array Query Acceleration with Integrated Processing Logic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory systems face inefficiencies in data query processing due to the need for external circuitry to process data, leading to delays and bandwidth bottlenecks when transferring data for query operations.
Innovation Solution
Incorporating processing circuitry directly on the memory device to perform data query operations, allowing parallel processing of data within the memory array and transferring only matching data to the host.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If data is transferred to external circuitry for processing queries, then processing can be performed using dedicated processing resources, but data transfer time increases and bandwidth bottlenecks occur
Solution Approach 1:
The patent merges the processing circuitry directly with the memory array by integrating logic units into the memory device itself. This allows query processing to occur in-place within the memory array, eliminating the need to transfer data to external processing resources. The logic units are coupled to the memory cells and can perform operations directly on stored data, resolving the contradiction between processing capability and data transfer time.
2Reliability
If all data is transferred to external circuitry for query processing, then comprehensive processing can be performed, but bandwidth limitations are exceeded and transfer delays increase
Solution Approach 1:
The patent segments the processing function by distributing logic units throughout the memory array rather than using a single external processor. Each logic unit can independently process queries on local data segments in parallel. This segmentation allows comprehensive processing to be distributed across multiple units, maintaining processing completeness while significantly improving query processing speed through parallel operation.
3Loss of time
If processing circuitry is integrated on the memory device, then data transfer time is reduced and query speed increases, but device complexity increases
Solution Approach 1:
The patent implements universal logic units that can perform multiple types of operations on data within the memory array. These logic units are designed to handle various query types and data processing tasks, making the integrated processing circuitry multi-functional. This universality reduces the overall complexity compared to having separate specialized processing units for each function, while still achieving fast query processing through in-place operations.
Data Source
AI summary
The present disclosure includes apparatuses and methods for acceleration of data queries in memory. A number of embodiments include an array of memory cells, and processing circuitry configured to receive, from a host, a query for particular data stored in the array of memory cells, wherein the particular data corresponds to a search key generated by the host, search portions of the array of memory cells for the particular data corresponding to the search key, determine data stored in the portions of the array of memory cells that matches the search key, and transfer the data that matches the search key to the host.