Memory Bank Data Layout for Rearrangement-Free Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing data transmission methods in matrix operations, such as from a DDR to a local memory and back, involve time-consuming data rearrangement steps, leading to reduced reading efficiency.
Innovation Solution
A method and apparatus for data transmission that involves sequentially reading data from a first memory into a buffer unit and writing it into a second memory based on a specified sequence, where each first memory unit is divided into sub-data units, allowing parallel reading and writing without additional rearrangement time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is read sequentially from first memory units and written to second memory units without rearrangement, then data reading efficiency is improved, but data sequence correctness may be compromised
Solution Approach 1:
The patent applies preliminary action by pre-dividing each first memory unit into K/m sub-data units according to bit positions before the actual data transmission process. This pre-organization of data into sub-units enables the subsequent parallel reading and writing operations to maintain correct data sequences without requiring time-consuming rearrangement steps, thus resolving the contradiction between reading efficiency and sequence correctness.
Solution Approach 2:
The patent segments each first memory unit into K/m sub-data units based on bit positions, creating discrete readable/writable portions. This segmentation allows the data transmission system to access and transfer specific sub-data units in parallel while maintaining the overall data sequence integrity, thereby improving reading efficiency without compromising correctness.
2Manufacturing precision
If data rearrangement steps are added to ensure correct sequence, then data sequence correctness is improved, but transmission time increases
Solution Approach 1:
The patent performs the data organization action in advance by dividing first memory units into sub-data units according to bit positions before transmission begins. This preliminary structuring eliminates the need for rearrangement operations during the actual transmission process, ensuring data sequence correctness while avoiding time loss.
Solution Approach 2:
The patent enables continuous data transmission by organizing data beforehand so that parallel reading and writing operations can proceed without interruption for rearrangement. The useful action of data transmission continues uninterrupted from first memory units through buffer units to second memory units, maintaining both correctness and efficiency.
3Speed
If parallel reading and writing operations are implemented, then data transmission speed is improved, but data organization complexity increases
Solution Approach 1:
The patent divides first memory units into K/m sub-data units based on bit positions, creating standardized segments that can be read and written in parallel. This segmentation provides a clear organizational structure that simplifies the control logic for parallel operations, enabling speed improvement without excessive complexity.
Solution Approach 2:
The patent applies different organizational rules to different portions of data based on their bit positions within the K-bit width. Each sub-data unit is organized according to its specific bit position requirements, allowing parallel operations to proceed with simple, localized organization rules rather than complex global reorganization.
Data Source
Figure 1A~1B
Figure 1C~1D
Figure 2A~3A
AI summary
A data transmission method, a data transmission apparatus and an electronic device. The method is used for transmitting data from a first memory to a second memory, a first storage unit at each depth in a bank has a width of K bits, and the second memory comprises a plurality of second storage units that are continuously arranged. The method comprises: sequentially reading data stored in a first storage unit at each depth among N depths, and sequentially storing the data of the N depths in a buffer unit, wherein the data has m bits, and each first storage unit is divided into K/m data sub-units according to the m bits from a low bit to a high bit (S10); and according to a specified sequence, writing the data in a bank from the buffer unit to a second memory, wherein the specified sequence comprises storing K/m pieces of data, which are located in adjacent K/m first storage units in the bank and are located at the same order, in the same second storage unit of the second memory (S20). By means of the method, the step of rearranging data does not occupy extra time, thereby improving the efficiency of data reading.