Multiple Address Registers for SSD Read Write Latency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current solid state device (SSD) interface circuits, such as PCIe, often use a single base address register (BAR) for both read and write operations, which can lead to biased latency for one type of request over others and restrict caching capabilities, resulting in reduced processing speed due to uncacheable data and lack of cache coherence support.

Innovation Solution

Implementing multiple address registers (BARs), specifically a read BAR (RD-BAR) and a write BAR (WR-BAR), each with selectable cache attributes, allowing cacheable read requests and uncacheable write requests, enabling differentiated caching for improved performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a single base address register (BAR) is used for both read and write operations, then device complexity is reduced, but latency is increased due to biased request handling and lack of caching capabilities

Engineering Contradiction:
Improveaddress register structureVSAvoidrequest latency
Core Design Contradiction:
Device complexityVSLoss of time

Solution Approach 1:

The patent divides the single BAR into multiple separate BARs, specifically a read BAR and a write BAR. This segmentation allows independent configuration of cache attributes for each operation type, enabling read operations to utilize caching while write operations can be handled differently, thus reducing read latency without increasing overall device complexity significantly

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces dynamic cache attribute configuration for each BAR, allowing the system to adaptively select appropriate cache policies (cacheable or uncacheable) based on the specific operation type. This dynamic approach optimizes performance by enabling caching for read operations while maintaining flexibility for write operations

Inventive Principle:
Principle #15Dynamics

2Ease of operation

If a single BAR is used for both read and write operations, then ease of operation is improved, but processing speed is reduced due to uncacheable data and lack of cache coherence support

Engineering Contradiction:
Improveinterface simplicityVSAvoiddata processing speed
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

By segmenting the single BAR into separate read and write BARs, the patent enables differentiated cache attribute configuration. Read BAR can be set to cacheable with cache coherence support, while write BAR can be configured separately, thereby improving processing speed for data-intensive operations without significantly complicating the interface

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the cache attribute parameter for different BARs, allowing read operations to access cacheable memory regions with cache coherence support while write operations can use uncacheable regions or different cache policies. This parameter differentiation directly improves processing speed by enabling efficient caching for read-heavy workloads

Inventive Principle:
Principle #35Parameter changes

3Productivity

If multiple address registers (BARs) with selectable cache attributes are implemented, then processing speed is improved through differentiated caching, but device complexity increases

Engineering Contradiction:
Improvedata access efficiencyVSAvoidaddress register configuration
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the address register into multiple specialized BARs (read BAR, write BAR) with distinct cache attribute configurations. This segmentation improves data access efficiency by allowing read operations to utilize caching while maintaining separate control for write operations, with the complexity increase being manageable through clear functional separation

Inventive Principle:
Principle #1Segmentation

4Manufacturing precision

If a single BAR is used for both read and write operations, then manufacturing precision requirements are reduced, but latency increases due to biased request handling

Engineering Contradiction:
Improveregister configuration precisionVSAvoidrequest latency
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

By segmenting the single BAR into separate read and write BARs, the patent reduces the precision requirements for each individual BAR configuration while improving latency performance. Each BAR can be configured with simpler, more specialized parameters tailored to its specific function, reducing manufacturing complexity while enabling optimized latency handling through dedicated cache attributes

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3417377B1Apparatuses and methods for multiple address registers for a solid state device
Publication Date: 2024.04.03 MICRON TECHNOLOGY INC
  • EP3417377B1 patent drawingFigure 1
  • EP3417377B1 patent drawingFigure 2
  • EP3417377B1 patent drawingFigure 3

AI summary

The present disclosure includes apparatuses, systems, and methods related to multiple address registers for a solid state device (SSD). An example apparatus includes a controller including a plurality of base address registers (BARs) each including same addresses for data storage in a same memory resource and an SSD that includes the same memory resource.