Multi-Stage Memory Structure for Dynamic Client Data Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data processing methods in communication systems face inefficiencies due to varying memory sizes for different client data, leading to suboptimal memory area usage and increased silicon area per bit, as well as difficulties in reallocating memory according to changing data demands across operational scenarios.
Innovation Solution
A multi-stage memory structure is employed, comprising a first stage for storing client data in FIFO memories, a second stage for storing data in burst form across shared memory banks, and a third stage for processing data in a destination memory, allowing for efficient data processing and transmission by optimizing memory allocation based on traffic characteristics and operational scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If client-specific memories are used for each piece of client data, then data processing for each client can be performed independently, but memory area efficiency deteriorates due to small sized memories and silicon area per bit increases
Solution Approach 1:
The patent merges multiple client-specific memories into a single shared memory structure. Instead of allocating separate memory blocks for each client (which causes area inefficiency), the invention combines them into one unified memory that serves all clients, thereby improving memory area efficiency while maintaining independent data processing through logical separation.
Solution Approach 2:
The shared memory structure is designed to serve multiple clients simultaneously, making a single memory resource universal. This multi-functional memory can handle data from different clients with varying data amounts, eliminating the need for dedicated memory for each client and reducing overall silicon area.
2Reliability
If client-specific memories are allocated for each client, then each client's data can be stored separately, but memory size increases in proportion to the maximum data amount and cannot be redistributed
Solution Approach 1:
The patent introduces dynamic memory allocation where memory resources can be flexibly assigned and reassigned based on actual client data needs. Instead of static client-specific memory allocation, the shared memory structure allows the system to dynamically adjust which memory regions are allocated to which clients, enabling adaptation to changing operational scenarios.
Solution Approach 2:
The system performs preliminary allocation of memory regions to clients based on expected data amounts, but retains the capability to reassign these allocations as needed. This preliminary yet flexible approach ensures data separation is maintained while allowing future redistribution when clients' data requirements change.
3Ease of manufacture
If uniform memory size is allocated to each client, then memory allocation is simple, but memory area efficiency deteriorates when some clients have small data amounts
Solution Approach 1:
The patent applies local quality by allowing different memory regions within the shared memory to have different allocation characteristics based on local client needs. Instead of enforcing uniform allocation across all clients, the system can allocate larger regions to clients with large data amounts and smaller regions to clients with small data amounts, optimizing overall memory area efficiency while maintaining allocation simplicity through centralized management.
Data Source
AI summary
A storage device, method for processing a plurality of pieces of client data, and a chipset are provided. The storage device includes a first stage storage unit configured to receive the plurality of pieces of client data generated in bursts from a plurality of clients and store the received plurality of pieces of client data; a second stage storage unit configured to receive the plurality of pieces of client data from the first stage storage unit and store the received plurality of pieces of client data in a plurality of memory banks shared by the plurality of clients, respectively, in bursts; and a third stage storage unit configured to receive each of the plurality of pieces of client data from the second stage storage unit and store data of a transaction unit corresponding to a transmission unit for data processing.