Duplicator Engine for Multi-Processor Data Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multi-processor data stream processing systems incur latency and bandwidth inefficiencies due to the need to copy network data streams between processors via interconnects, such as UPIs, which affects processing efficiency and resource allocation.
Innovation Solution
An Information Handling System (IHS) with a duplicator engine that receives a data payload and duplicates it to provide separate copies to multiple processing subsystems via endpoint subsystems, eliminating the need for interprocessor copying and optimizing data distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data payloads are copied between processors via interconnects (UPIs) to enable parallel processing, then parallel processing capability is achieved, but latency increases and bandwidth efficiency deteriorates
Solution Approach 1:
The system segments the data distribution function by assigning different data payloads to different endpoint subsystems directly, rather than copying the same data through all processors. Each processing subsystem receives only the data payloads intended for it, eliminating redundant copying operations and reducing latency while maintaining parallel processing capability.
2Productivity
If data payloads are copied between processors via interconnects (UPIs) to enable parallel processing, then parallel processing capability is achieved, but bandwidth costs increase
Solution Approach 1:
The data distribution architecture is segmented so that each processing subsystem receives only its designated data payloads directly from the duplicator engine through endpoint subsystems. This eliminates redundant data transmissions across interconnects, significantly reducing bandwidth consumption while preserving parallel processing throughput.
Solution Approach 2:
The duplicator engine acts as an intermediary that intelligently distributes data payloads to appropriate endpoint subsystems based on processing requirements. This mediator function optimizes data flow paths, preventing unnecessary bandwidth consumption on interconnects while enabling efficient parallel processing.
3Device complexity
If conventional interprocessor copying is used, then system simplicity is maintained, but processing efficiency deteriorates
Solution Approach 1:
The endpoint subsystems are designed with multi-functionality, serving both as interfaces for traditional interprocessor communication and as direct recipients of duplicator engine data distributions. This universal design enables the system to adopt the efficient selective distribution model without requiring completely new hardware components, thus maintaining relative system simplicity while dramatically improving processing efficiency.
Data Source
AI summary
A multi-endpoint adapter device includes a plurality a duplicator device that is coupled to the network port and the plurality of endpoint subsystems that are each configured to couple with a respective processing subsystem. The duplicator device receives, via the network port, a data payload and determines that the data payload is to be provided to each of a first processing subsystem via a first endpoint subsystem that is included in the plurality of endpoint subsystems, and a second processing subsystem via a second endpoint subsystem that is included in the plurality of endpoint subsystems. The duplicator device then duplicates the data payload to provide a first duplicated data payload and a second duplicated data payload. The duplicator device then provides the first duplicated data payload to the first endpoint subsystem and provides the second duplicated data payload to the second endpoint subsystem.