Collective Communication Latency Reduction via Dependency Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional collective communication methods suffer from high communication latency and resource consumption due to the management of processes with and without communication dependencies using queues, leading to performance deterioration and inefficient use of resources.
Innovation Solution
A method that identifies and manages only processes with communication dependencies using queues, while allowing processes without dependencies to be directly communicated, thereby reducing latency and resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If queue management is used for all processes to ensure proper data transmission control, then communication reliability is improved, but communication latency increases and resource consumption rises
Solution Approach 1:
The patent segments processes into two categories: those with communication dependencies and those without. Processes without dependencies are handled through direct communication without queue management, while processes with dependencies use queue management. This segmentation allows the system to maintain reliability where needed while minimizing latency for independent processes.
Solution Approach 2:
Instead of applying queue management to all processes (excessive action), the patent applies it only to processes that actually require it (partial action). By identifying and excluding processes without communication dependencies from queue management, the system achieves the minimum necessary control while reducing overall latency and resource consumption.
2Reliability
If queue management is implemented for all processes, then data transmission control is improved, but resource consumption increases
Solution Approach 1:
The patent divides processes into segments based on their communication dependency characteristics. By segmenting the process set and applying queue management only to the necessary subset, the system reduces the resource consumption associated with queue operations while maintaining adequate control for processes that require it.
Solution Approach 2:
The patent extracts processes without communication dependencies from the queue management system. These extracted processes are handled through direct communication paths, removing them from the resource-consuming queue management overhead while maintaining proper control for processes that remain in the queue system.
3Reliability
If interrupt processing is triggered for each data reception to ensure proper handling, then communication reliability is improved, but system performance deteriorates due to context switching
Solution Approach 1:
The patent segments communication processes into those requiring interrupt handling and those that can use direct notification. By identifying processes without communication dependencies, the system can bypass interrupt processing for these cases, reducing context switching overhead while maintaining reliable handling for processes that genuinely require interrupt-driven communication.
4Stability of the object's composition
If all processes are managed through queue control to ensure proper communication sequencing, then communication order is maintained, but communication latency increases
Solution Approach 1:
The patent segments the communication system into queue-managed paths and direct communication paths based on process dependency relationships. This segmentation allows the system to maintain strict sequencing only where communication dependencies exist, while enabling faster direct communication for independent processes, thereby reducing overall latency while preserving necessary order.
Data Source
Figure 1~2
Figure 3A
Figure 3B
AI summary
This application provides a method for implementing collective communication, a computer device, and a communication system, to resolve problems of high communication latency and resource consumption in the conventional technology. The method provided in this application includes: identifying a work request having no communication dependency, and when forwarding work requests, directly forwarding the work request identified as having no communication dependency, and forwarding, after queue management, a work request that is not identified as having no communication dependency. In this way, communication latency caused when the work request having no communication dependency is managed by using a queue can be avoided, and resource consumption caused by performing related management can be reduced, to improve overall communication performance of collective communication.