Memory Device Latches for Address Mapping Table Read Performance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Memory systems face increased latency due to the time-consuming process of retrieving address mapping information from memory cells, which can slow down address translation and affect the performance of electronic devices, especially in high-processing applications like AI, AR, and gaming.

Innovation Solution

Storing portions of the address mapping table in memory device latches, which have faster access times than memory cells, to reduce latency during retrieval procedures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If address mapping information is stored in memory cells, then storage capacity is maintained, but retrieval latency increases

Engineering Contradiction:
Improveretrieval latencyVSAvoidmemory structure complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The patent segments the memory system into two parts: memory cells for bulk storage and latches for frequently accessed address mapping information. This segmentation allows the system to maintain large storage capacity while providing fast access paths for critical data through the latch structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements preliminary action by pre-loading address mapping information into latches before it is needed for translation operations. The controller proactively transfers frequently accessed mapping data from memory cells to latches, so that when address translation is required, the data is already in the fast access path.

Inventive Principle:
Principle #10Preliminary action

2Speed

If address mapping information is cached in controller cache, then access speed improves, but cache memory size requirements increase

Engineering Contradiction:
Improveaddress translation speedVSAvoidcache memory size
Core Design Contradiction:
SpeedVSVolume of stationary object

Solution Approach 1:

The patent adds a new dimension to the memory hierarchy by introducing latches as an intermediate storage layer between memory cells and the controller cache. This latch layer provides fast access without requiring large cache memory, effectively adding a temporal dimension to data access with multiple access paths of different speeds.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Productivity

If latches are used for storing address mapping information, then read performance improves, but power consumption increases

Engineering Contradiction:
Improveread performanceVSAvoidpower consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent applies partial action by using latches only for storing a portion of the address mapping information that is frequently accessed, rather than using latches for all stored data. This selective approach allows the system to gain read performance benefits for critical paths while avoiding the power consumption penalty of maintaining latch structures for entire memory容量.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS20240264933A1Read performance improvement using memory device latches
Publication Date: 2024.08.08 MICRON TECHNOLOGY INC
  • US20240264933A1 patent drawing
  • US20240264933A1 patent drawing
  • US20240264933A1 patent drawing

AI summary

Methods, systems, and devices for read performance improvement using memory device latches are described. A memory system may determine whether a portion of an address mapping table is absent from a cache of one or more controllers and may read the portion of the address mapping table from memory cells of the memory device based on the portion of the address mapping table being absent from the cache. The memory system may communicate, to a memory device, a write command to store the portion of the address mapping table in a set of latches of the memory device based on the address mapping table being associated with an access command that is part of a non-sequential read procedure. The memory system may update a tracking table in the cache to indicate that the portion of the address mapping table is stored in the set of latches.