Parallel NVDIMM Architecture with Interface Bridge Chips

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing memory module designs face challenges with high latency in accessing storage devices, signal interference, and limited performance due to the use of a single FPGA chip for data transfer between DRAM and NAND Flash chips, which results in wire clogging and increased costs.

Innovation Solution

A parallel NVDIMM architecture utilizing multiple interface bridge chips as processors to facilitate parallel data transfer between DRAM and non-volatile memory devices, allowing for a dispersed and efficient layout design without signal integrity issues and reducing the need for extensive I/O pins on the FPGA.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If a single FPGA chip is used to facilitate data transfer between DRAM and NAND Flash chips, then the layout design becomes constrained and wire clogging occurs, but using multiple interface bridge chips enables dispersed layout and reduces wire congestion

Engineering Contradiction:
Improvelayout design easeVSAvoidnumber of interface bridge chips
Core Design Contradiction:
Ease of manufactureVSDevice complexity

Solution Approach 1:

The patent divides the single FPGA chip into multiple interface bridge chips, each handling a portion of the data transfer task. This segmentation allows the components to be dispersed across the PCB, reducing wire congestion and simplifying layout design while maintaining the overall data transfer functionality.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from a single-point concentration of data transfer operations to a distributed spatial arrangement across multiple chips. This dimensional shift from centralized to distributed architecture resolves the wire clogging issue by spreading connections across multiple locations on the PCB.

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

2Device complexity

If a single FPGA chip is used for data transfer, then the I/O pin requirement increases significantly, but using multiple interface bridge chips distributes the I/O requirements

Engineering Contradiction:
ImproveI/O pin countVSAvoidcost
Core Design Contradiction:
Device complexityVSEase of manufacture

Solution Approach 1:

The patent segments the I/O pin requirements across multiple interface bridge chips instead of concentrating them on a single FPGA chip. Each chip handles a subset of data transfer operations, reducing the I/O pin count requirement for each individual chip and thereby reducing overall manufacturing cost.

Inventive Principle:
Principle #1Segmentation

3Productivity

If a single processor is used to handle data transfer between DRAM and NAND Flash, then the performance is limited by sequential processing, but using multiple processors enables parallel data transfer and improves performance

Engineering Contradiction:
Improvedata transfer performanceVSAvoidnumber of processors
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the data transfer function across multiple interface bridge chips, each capable of independent data transfer operations. This segmentation enables parallel processing of data transfer tasks, significantly improving productivity while the modular nature of the chips keeps the complexity increase manageable.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent enables continuous data transfer operations through multiple parallel processors that can simultaneously handle different data transfer tasks. This eliminates the sequential bottlenecks of a single processor, maintaining continuous productive action throughout the data transfer process.

Inventive Principle:
Principle #20Continuity of useful action

4Reliability

If all data bits are routed to a single FPGA chip location, then the data transfer function is achieved, but severe wire clogging and signal interference occur

Engineering Contradiction:
Improvesignal integrityVSAvoidwire routing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the data routing function across multiple interface bridge chips distributed on the PCB. This segmentation eliminates the need for all data bits to converge at a single location, thereby reducing wire routing complexity and improving signal integrity by shortening individual trace lengths and reducing congestion.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS12014078B2Apparatus and architecture of non-volatile memory module in parallel configuration
Publication Date: 2024.06.18 ENTRANTECH INC
  • US12014078B2 patent drawing
  • US12014078B2 patent drawing
  • US12014078B2 patent drawing

AI summary

A non-volatile memory module in parallel architecture is described. It includes memory function and data storage function in a single module. It enables host system to use memory bus to access storage devices and to use the same memory command protocol for storage device access. The parallel architecture enables contents in memory devices and storage devices to be exchanged freely on module under the control of host memory controller to boost performance of computer and to retain data even if power to computer is shut off. The configuration of non-volatile memory module can be partitioned or expanded into multiple independent channels on module seamlessly with or without ECC supports.