Parallel Memory Connection Interface for Low-Latency Data Matching

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In computer clusters, the efficient data transmission between units is hindered by high latency and bandwidth overhead due to extensive comparison processes for assigning data packets, which require significant system performance and energy consumption, especially when dealing with large lists of messages and multiple threads.

Innovation Solution

A connection interface with a shared repository of unassigned messages using parallel memories and dual-port or self-compacting memory technology, allowing for highly parallel access and reducing energy consumption by storing only message headers, enabling quick matching and efficient data transmission without storing complete data streams.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If complete data streams are stored for incoming messages, then message matching can be performed, but memory usage and energy consumption increase significantly

Engineering Contradiction:
Improvemessage matching accuracyVSAvoidenergy consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent extracts only the essential identifying information (message headers) from complete data streams for storage and comparison purposes. This allows message matching to be performed using minimal data, significantly reducing memory usage and energy consumption while maintaining matching accuracy. The complete data streams are processed only when needed, not stored continuously.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different quality levels to different parts of the data: message headers are stored with high detail for accurate matching, while complete data streams are either not stored or stored with lower priority. This local differentiation optimizes resource usage by focusing storage capacity on the most critical matching information.

Inventive Principle:
Principle #3Local quality

2Device complexity

If sequential comparison methods are used for message matching, then system complexity is reduced, but latency increases due to extensive comparison processes

Engineering Contradiction:
Improvesystem complexityVSAvoidlatency
Core Design Contradiction:
Device complexityVSLoss of time

Solution Approach 1:

The patent segments the message matching process into distinct stages: header extraction, header comparison, and full data processing. By segmenting the workflow, the system can perform quick header-level comparisons to filter messages, reducing overall latency without requiring complex parallel processing of complete data streams.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary actions by extracting and comparing message headers before processing complete data streams. This preliminary header comparison quickly identifies matching messages, reducing the amount of full data processing needed and thereby reducing latency while maintaining manageable system complexity.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If dedicated hardware is implemented for parallel message comparison, then matching speed increases, but device complexity and cost increase

Engineering Contradiction:
Improvematching speedVSAvoidhardware complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary layer (header extraction and comparison mechanism) that operates between incoming messages and the full message processing system. This intermediary performs quick filtering using simple header comparisons, enabling faster matching without requiring dedicated hardware for complete message parallel processing.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent creates simplified copies (headers) of complete messages for comparison purposes. These header copies are much smaller and faster to process than complete messages, enabling rapid parallel comparisons using standard hardware while achieving productivity improvements without proportionally increasing hardware complexity.

Inventive Principle:
Principle #26Copying

4Reliability

If message buffering is implemented until identification, then message delivery accuracy improves, but processing time and CPU load increase

Engineering Contradiction:
Improvemessage delivery accuracyVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent extracts message headers for immediate comparison and identification purposes, allowing messages to be quickly identified and routed without requiring complete buffering. Only messages that cannot be immediately identified through header comparison are buffered for later processing, minimizing processing time while maintaining delivery accuracy.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent performs preliminary identification using message headers before full message buffering is required. This preliminary action using extracted header information enables quick routing decisions for most messages, reducing the need for extended buffering and thereby reducing processing time while maintaining accurate message delivery.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2668763B1Circuit arrangement for connection interface
Publication Date: 2017.12.06 EXTOLL
  • EP2668763B1 patent drawingFigure 1
  • EP2668763B1 patent drawingFigure 2
  • EP2668763B1 patent drawingFigure 3

AI summary

The invention relates to a connection interface comprising a circuit arrangement, for associating incoming data messages with available information, comprising a means for comparing incoming messages with available information and for storing available information. To this end, the means comprises a plurality of comparison elements with which a respective memory is associated, said respective memories together forming a storage element for storing information relating to non-associated messages, and the comparison elements are designed to compare, in parallel, incoming messages with the information stored in the respective memories thereof.