NIC Adaptive Triggered Operation Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In high-performance computing (HPC) environments, the efficient distribution of triggered operation data structure (TODS) entries among processes is challenging, leading to resource exhaustion and adverse effects on underutilized processes.
Innovation Solution
The solution involves distributing TODS entries among processes generating triggered operations, maintaining a window to indicate available entries, and decrementing and incrementing the window size in response to enqueuing and executing triggered operations, respectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If TODS entries are shared among multiple processes without distribution control, then resource utilization increases, but some processes may exhaust the TODS while others remain underutilized
Solution Approach 1:
The patent divides the TODS into separate segments or queues, with each process assigned to a specific segment. This segmentation prevents any single process from exhausting the entire TODS while allowing multiple processes to operate concurrently with guaranteed resource availability in their respective segments.
2Reliability
If TODS entries are distributed among processes using traditional synchronization methods, then resource exhaustion is prevented, but system performance decreases due to blocking and locking
Solution Approach 1:
Each process maintains its own window size counter that automatically tracks available TODS entries in its assigned segment. The process can independently check its window size and enqueue descriptors without requiring locks or synchronization with other processes, enabling self-service operation that prevents resource exhaustion while maintaining high performance.
3Stability of the object's composition
If window size checking and descriptor enqueuing are performed atomically, then data consistency is maintained, but operation latency increases
Solution Approach 1:
The system pre-allocates TODS segments to each process and initializes window sizes before operations begin. This preliminary setup establishes clear boundaries and available resources, allowing processes to perform non-atomic check-and-enqueue operations within their assigned segments without risking data consistency issues that would require atomic operations.
Data Source
AI summary
A system for managing triggered operations in a computing system is provided. The computing system can include a storage medium to store descriptors identifying triggered operations to be performed based on respective trigger conditions. A network interface controller of the computing system can store a data structure. During operation, the system can determine, for a process, a window size indicating a number of available entries in the data structure. If the window size indicates an available entry, the system can insert a descriptor of a triggered operation generated by the process into a corresponding work queue. The system, at the NIC, can determine a location of the descriptor in the work queue. The system can then transfer the descriptor from the location to the data structure. Subsequently, the system can decrement the window size indicating an updated number of available entries for the first process in the data structure.