Executable Graph Dynamic Reconfiguration for Parallel Workloads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for executable graphs in parallel computing are limited in their ability to perform multiple workloads, as they are typically designed for a single workload and require significant reconfiguration to adapt to different tasks.
Innovation Solution
The development of techniques to modify executable graphs allows for their reuse across various workloads by incorporating dynamic reconfiguration and resource management within the data center infrastructure, enabling efficient task scheduling and resource allocation across different computing resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If an executable graph is designed for a single workload, then it achieves optimal performance for that specific workload, but it cannot be reused for different workloads without significant reconfiguration
Solution Approach 1:
The executable graph incorporates dynamic reconfiguration capabilities that allow its structure and parameters to be modified at runtime based on the specific workload requirements. This enables the same executable graph to adapt its configuration for different workloads while maintaining optimal performance characteristics for each.
Solution Approach 2:
The patent creates a universal executable graph framework that can perform multiple workloads through parameter modification rather than requiring separate executable graphs for each workload. The system achieves multi-functionality by allowing the same graph structure to be reused with different parameter sets for different computational tasks.
2Adaptability or versatility
If an executable graph is modified to perform different workloads, then versatility improves, but reconfiguration complexity increases
Solution Approach 1:
The system performs preliminary actions by pre-defining parameter sets and validation rules that guide the reconfiguration process. Before modifying an executable graph for a new workload, the system prepares the necessary parameter configurations and validates their compatibility, thereby reducing the actual reconfiguration complexity during runtime.
Solution Approach 2:
The patent incorporates feedback mechanisms that monitor the reconfiguration process and provide information about the current state of the executable graph. This feedback enables automated adjustment of parameters and validation of modifications, reducing the complexity of manual reconfiguration and ensuring correctness during workload transitions.
Data Source
AI summary
Techniques to modify executable graphs to perform different workloads. In at least one embodiment, an executable graph created from a task graph for a first workload is modified to perform a second workload that differs from first workload.