nvdv/vprof
Vprof
vprof ist ein visuelles Profiling-Tool für Python, das entwickelt wurde, um Engpässe bei der Ausführung zu identifizieren und den Speicherverbrauch zu überwachen. Es fungiert als CPU- und Speicher-Profiler, der Leistungsdaten in interaktive Visualisierungen umwandelt, um Prozessorzeit und Call-Stacks zu analysieren.
Das Projekt zeichnet sich durch eine Reihe visueller Diagnosen aus, darunter Flame Graphs zur Stack-Visualisierung und Heatmaps, die Ausführungshäufigkeit und -dauer direkt auf den Quellcode abbilden. Es enthält zudem einen Remote-Performance-Monitor, der funktionsspezifische Metriken von einem laufenden Server erfassen und diese Daten an ein separates Visualisierungstool streamen kann.
Das Tool deckt breite Funktionsbereiche ab, darunter sampling-basiertes CPU-Profiling, zeilenweises Speicher-Monitoring durch Garbage-Collector-Tracking sowie die Persistenz von Profildaten für die Offline-Analyse. Diese Dienstprogramme ermöglichen die Prüfung der Effizienz des Quellcodes und die Identifizierung von Speicherlecks.
Features
- Python Profilers - Provides a comprehensive suite for measuring execution time and identifying CPU bottlenecks in Python programs.
- Code Execution Profilers - Analyzes runtime performance of code using heatmaps that show execution counts and run times for every line.
- Source Code Heatmaps - Produces heatmaps of source code showing execution counts and run times for every line.
- CPU Profilers - Captures periodic snapshots of the call stack to estimate execution time and identify performance bottlenecks.
- Visual Profiling Toolkits - Provides an interactive visual interface for analyzing Python performance via flame graphs, heatmaps, and statistics.
- Execution Heatmaps - Overlays execution frequency and duration data directly onto source code for visual performance analysis.
- Execution Tracking - Instruments source code to record execution counts and timing statistics for every individual line of code.
- Memory Leak Detectors - Interfaces with the Python memory manager to track object allocations and identify memory leaks.
- Memory Consumption Trackers - Tracks objects in the garbage collector and monitors process memory usage to identify leaks and heavy allocations.
- Python Execution Visualizations - Generates interactive charts and heatmaps to analyze how a Python program performs across its call stack.
- Line-Level Resource Monitors - Provides granular profiling that pinpoints specific lines of code responsible for memory consumption and performance overhead.
- Flame Graphs - Generates interactive flame graphs of call stacks to identify resource-intensive functions and processing bottlenecks.
- Performance Code Auditing - Visualizes line-by-line execution counts and run times through heatmaps to optimize specific blocks of code.
- Remote Profiling Streams - Transports profiling metrics from a target Python process to a separate visualization server via network sockets.
- Remote Profiling Interfaces - Implements interfaces to capture and render performance statistics from functions running on a remote server.
- Remote Function Profiling - Deno profiling utility that captures performance metrics for individual functions by routing data to a remote visualization server.
- Remote Monitoring - Captures and renders performance statistics from functions running on a remote server.
- Program Execution Visualization - Transforms raw profiler data into interactive charts and visual maps of program resource consumption.
