Parallel Computer Node Role Switching for Monitoring Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-layer master-worker type parallel programs, the inefficiency in computing node utilization arises due to a large number of nodes not performing actual computation, leading to increased monitoring load and reduced efficiency compared to single-layer master-worker systems.
Innovation Solution
A parallel computer system with a hierarchical structure where each computing node operates as both a computing and monitoring process, dynamically adjusting the monitoring hierarchy based on target values for the total number of subordinate computing processes and monitoring processes to optimize node utilization and reduce monitoring load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a multi-layer master-worker type parallel program is used to reduce monitoring load, then the monitoring load on each process is reduced, but the computing node utilization efficiency decreases due to more nodes not performing actual computation
Solution Approach 1:
The patent implements dynamic role assignment where computing nodes can switch between computing process and monitoring process roles based on system state and workload. This dynamic flexibility allows the system to optimize both monitoring distribution and computing utilization, resolving the contradiction between reduced monitoring load and maintained computing efficiency.
Solution Approach 2:
Each computing node is designed to be multi-functional, capable of operating as either a computing process or a monitoring process. This universality allows nodes to be dynamically allocated to different roles, enabling the system to reduce monitoring load on individual nodes while maintaining overall computing node utilization efficiency through flexible role assignment.
2Productivity
If a single-layer master-worker type parallel program is used, then computing node utilization is higher, but the monitoring load on the master process increases in proportion to the number of worker processes
Solution Approach 1:
The patent segments the monitoring function by allowing multiple computing nodes to serve as monitoring processes, distributing the monitoring load across multiple nodes rather than concentrating it on a single master process. This segmentation reduces the monitoring burden on any single process while maintaining efficient computing node utilization.
Solution Approach 2:
The patent introduces intermediary monitoring processes that act between the master process and worker processes. These intermediary nodes help distribute and manage monitoring tasks, reducing the direct monitoring burden on the master process while maintaining system-wide monitoring effectiveness and computing efficiency.
3Device complexity
If computing nodes are allocated as monitoring processes in a multi-layer hierarchy, then the monitoring load is distributed, but the number of nodes performing actual computation decreases
Solution Approach 1:
The system dynamically adjusts the allocation of computing nodes between monitoring and computing roles based on real-time system state, workload demands, and performance metrics. This dynamic allocation ensures that monitoring load is distributed effectively while maximizing the number of nodes available for actual computation at any given time.
Solution Approach 2:
The patent employs parameter changes by adjusting the number and distribution of monitoring processes based on system conditions. When computing demand is high, fewer nodes are allocated to monitoring; when monitoring complexity increases, more nodes are dedicated to monitoring. This parameter adjustment resolves the contradiction between monitoring load distribution and computing process availability.
Data Source
AI summary
A parallel computer system includes computing nodes to execute a parallel program for generating computing processes that perform computations and monitoring processes and forming a monitoring hierarchical structure, each monitoring process monitors a monitoring process and a computing process arranged immediately lower than the monitoring process, each of the computing nodes operating as the computing process when the computing process is allocated to the computing node and operating as the monitoring process when the monitoring process is allocated to the computing node. Each of the computing nodes allocated to the monitoring process performs processing that changes the hierarchical structure based on a first target value serving as a target value for the total number of subordinate computing processes of the allocated monitoring process and a second target value serving as a target value for the number of monitoring processes and computing processes arranged immediately lower than the allocated monitoring process.