graph LR
Core_Benchmarking_Engine["Core Benchmarking Engine"]
Visualization_Engine["Visualization Engine"]
Benchmarking_Orchestrator_Plot_Display["Benchmarking Orchestrator & Plot Display"]
Live_Benchmarking_Manager["Live Benchmarking Manager"]
Result_Persistence_Handler["Result Persistence Handler"]
Benchmarking_Orchestrator_Plot_Display -- "initiates" --> Core_Benchmarking_Engine
Benchmarking_Orchestrator_Plot_Display -- "delegates to" --> Visualization_Engine
Live_Benchmarking_Manager -- "interacts with" --> Core_Benchmarking_Engine
Live_Benchmarking_Manager -- "potentially interacts with" --> Visualization_Engine
Result_Persistence_Handler -- "ensures execution by" --> Core_Benchmarking_Engine
Result_Persistence_Handler -- "requests visualization for persistence from" --> Visualization_Engine
click Visualization_Engine href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/perfplot/Visualization_Engine.md" "Details"
The perfplot benchmarking subsystem is structured around a clear separation of concerns, with a central Core Benchmarking Engine responsible for executing performance measurements. User-facing components like the Benchmarking Orchestrator & Plot Display, Live Benchmarking Manager, and Result Persistence Handler serve as entry points, each orchestrating specific benchmarking workflows. These components interact with the Core Benchmarking Engine to perform the actual benchmarks and then delegate to the Visualization Engine for presenting or persisting the results. The Visualization Engine handles all aspects of plot generation and display, whether for live, interactive sessions or for static output and saving. This architecture ensures modularity, allowing for independent development and maintenance of benchmarking logic, visualization capabilities, and various user interaction modes.
This component is responsible for executing the actual performance benchmarks. It takes kernels (functions to be benchmarked) and a range of input sizes, then measures their execution time. It's the central computational unit for performance measurement.
Related Classes/Methods:
Visualization Engine [Expand]
This component handles the generation and display of plots from benchmarking results. It provides functionalities to visualize performance data, including live updates and static plot generation, often leveraging matplotlib for rendering.
Related Classes/Methods:
This component acts as the primary facade for initiating a complete benchmarking workflow. It orchestrates the execution of benchmarks via the Core Benchmarking Engine and then delegates to the Visualization Engine for displaying the generated plots. It represents the typical "run and show" user experience.
Related Classes/Methods:
This component handles interactive and real-time benchmarking sessions. It continuously executes benchmarks and updates results, leveraging the Core Benchmarking Engine for measurements and the Visualization Engine for dynamic plot updates.
Related Classes/Methods:
This component is responsible for managing the saving of benchmarking results, including generated plots and raw data. It interacts with the Core Benchmarking Engine to get results and requests visualization from the Visualization Engine for persistence.
Related Classes/Methods: