PaScal Viewer

The Parallel Scalability Viewer (PaScal Viewer) is a web-based application designed to visualize performance metrics from parallel programs running on shared-memory systems. It takes as input the output file generated by the PaScal Analyzer, offering a simplified way to analyze and optimize code for parallel scalability.

PaScal Viewer introduces an innovative approach to visual analysis through color diagrams resembling heatmaps, which provide visual support for identifying efficiency trends across the entire program or within specific regions of parallel code. Its interface supports hierarchical parallel zone analysis, enabling detailed profiling of complex parallel applications.

The PaScal Viewer is publicly available at: https://pascalsuite.imd.ufrn.br/pascal-viewer

Key Features

• Provide a clear way to understand the outputs generated by PaScal Analyzer.
• Generate heatmap diagrams illustrating:
  • Parallel efficiency
  • Overall scalability
  • Weak scalability
  • Strong scalability
• Support the visualization of hierarchical regions defined during instrumentation.
• Allow side-by-side comparisons between parent and sibling regions, as well as the whole program execution.

PaScal Viewer transforms raw performance data into intuitive, interactive visuals that help users recognize performance bottlenecks, compare execution regions, and make informed optimization decisions for parallel applications.