Content Addressable DRAM Parallel Search Architecture
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory architectures face challenges in reducing search latency and power consumption when dealing with large data sets, as they require significant power to shuttle data between processors and separate memory during search operations.
Innovation Solution
A content addressable dynamic random-access memory (DRAM) with parallel search functionality is introduced, which allows for simultaneous connection of sense amplifiers to a shared line and activation of select lines based on search queries, enabling efficient in-memory search operations without the need for large data transfers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If data is stored in separate memory and shuttled to processor for search operations, then search functionality is provided, but power consumption increases and search latency increases
Solution Approach 1:
The patent merges the processor and memory into a single integrated system where search operations are performed directly within the memory device. The memory device includes both storage cells and processing circuitry (search engine, comparators, decoders) that work together to execute search queries without external data transfer, thereby reducing power consumption while managing complexity through integration.
Solution Approach 2:
The memory device is designed with multi-functionality, serving both as storage medium and as a processing unit capable of executing search operations. The same memory structure stores data and simultaneously performs comparison, decoding, and search logic operations, eliminating the need for separate processor-memory architecture and reducing overall system power consumption.
2Loss of time
If data is shuttled from memory to processor for search, then search operations can be performed, but search latency increases
Solution Approach 1:
By combining storage and processing functions within the same memory device, the patent eliminates data transfer time between separate memory and processor components. Search operations are executed in-place within the memory array, dramatically reducing search latency while the integrated design ensures power efficiency is maintained.
3Productivity
If large datasets are searched using traditional memory-processor architecture, then search functionality is provided, but power consumption increases due to data transfer
Solution Approach 1:
The memory device is segmented into multiple independently controllable blocks or regions that can be searched in parallel. Each segment can be activated independently based on the search query, allowing large datasets to be divided into manageable portions that are processed simultaneously, improving search performance while minimizing power consumption by activating only necessary segments.
Solution Approach 2:
The patent enables partial search operations where only specific portions or blocks of the memory array are activated based on the search query requirements. This partial action approach improves performance by focusing computational resources on relevant data segments while reducing power consumption by leaving inactive segments in a low-power state.
Data Source
AI summary
An extremely dense, high speed, and low power content addressable DRAM is presented. To enable a parallel searching, a data word to be searched may be driven onto column select lines (CSLs) of a DRAM array. Although two or more primary sense amplifiers typically are not connected at the same time to the same local data line during operation of a DRAM, in various embodiments presented herein, some or all sense amplifiers in a DRAM can be activated simultaneously to enable maximum parallelism with local data line sharing being explicitly allowed. Using this architecture, a data word can be simultaneously searched in all banks and with multiple wordlines. Since no input/output transactions are required and no data needs to be driven from the bank during execution of a search, overall current, and thus power usage, can be reduced.