DRAM Memory Bank Grouping for Reduced Row-to-Row Delay
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional DRAMs face inefficiencies due to fixed row-to-row delay times (tRRD) that exceed CAS-to-CAS delay times, leading to data bubbles and reduced reading efficiency when switching between memory banks, limiting system resource utilization.
Innovation Solution
The memory is divided into distinct memory groups with independent driving powers, allowing for different row-to-row delay times (tRRD) to minimize the time interval between selecting rows in different groups, thereby reducing data bubbles and enhancing reading speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single driving power is used for all memory banks, then device complexity is reduced, but the row-to-row delay time increases causing data bubbles and reduced reading efficiency
Solution Approach 1:
The memory system is segmented into multiple memory groups, each with its own independent driving power. This segmentation allows parallel operation of different memory groups, eliminating the need to wait for one group to complete its row-to-row delay before accessing another group, thereby eliminating data bubbles and improving reading efficiency while maintaining manageable device complexity through modular architecture
2Device complexity
If a fixed row-to-row delay time is used for all memory banks, then control logic is simplified, but data bubbles are generated when switching between memory banks reducing system resource utilization
Solution Approach 1:
The patent implements dynamic row-to-row delay timing where different memory groups can operate with different tRRD values optimized for their specific characteristics. The control logic dynamically selects which memory group to access next based on current operation status and timing requirements, allowing overlapping operations and eliminating fixed delay constraints that cause data bubbles
3Reliability
If the row-to-row delay time is increased to ensure stable operation, then reliability is improved, but the time interval for selecting rows increases causing data bubbles
Solution Approach 1:
By dividing memory banks into separate memory groups with independent driving powers, the system can maintain longer tRRD values within each group for stable operation while simultaneously accessing different groups with shorter effective delays, achieving both reliability and high productivity through parallel operations
Data Source
AI summary
A memory and an operation method thereof are provided. The present invention divides memory banks of the memory into a plurality of memory groups, wherein each memory group has an independent driving power for providing an operating voltage to the corresponding memory bank in the memory group. The present invention specifies two tRRD times which are an inter-group interval and an intra-group interval. The intra-group interval is the minimum time interval between selecting one row of memory banks in a memory group to selecting another row in the memory banks of the same memory group and the inter-group interval is the minimum time interval between selecting one row of the memory banks in one memory group to selecting another row in a different memory group. Further, the inter-group interval is shorter than or equal to the intra-group interval.