.png){kind=small}
TornadoInsight
Type: IntelliJ IDEA Plugin
​
TornadoInsight is an open-source IntelliJ IDEA plugin that improves the developer experience when building TornadoVM applications. It provides real-time static analysis to detect unsupported Java features in TornadoVM code and includes a dynamic testing framework that allows developers to run and validate individual TornadoVM tasks directly from the IDE.
The plugin automatically generates the required execution code, runs the task using the local TornadoVM runtime, and shows the generated kernel or execution diagnostics, helping developers quickly test compatibility and debug accelerated Java code.
What it offers
-
Real-time static analysis for TornadoVM compatibility
-
Dynamic testing of TornadoVM tasks inside IntelliJ IDEA
-
Automatic generation of execution scaffolding
-
Runtime diagnostics and kernel inspection
TornadoPulse
Type: Profiling / Analysis Tool
​
TornadoVMPulse is a lightweight analysis tool that helps developers inspect and understand TornadoVM profiling logs.
It provides a quick way to process profiling output generated by TornadoVM and extract useful performance insights, making it easier to analyze execution behavior, identify bottlenecks, and evaluate the performance of accelerated Java applications.​
What it offers
-
Quick analysis of TornadoVM profiling logs
-
Performance insights for accelerated tasks
-
Simplified inspection of execution behavior
-
Support for performance debugging and optimization
TornadoViz
Type: Visualization / Analysis Tool
TornadoViz is a visualization and analysis tool that helps developers understand how their applications execute on TornadoVM.
It analyzes TornadoVM Bytecode execution logs and provides interactive visual insights into heterogeneous execution, allowing users to inspect task behavior, identify performance issues, and better understand how Java code is mapped to accelerated devices.
This helps developers debug, analyze, and optimize TornadoVM applications more effectively.
What it offers
-
Visualization of TornadoVM bytecode execution logs
-
Insights into heterogeneous task execution
-
Support for performance analysis and debugging
-
Better understanding of how Java code is accelerated on devices
TornadoVM Tools
Integration with External Tools
NVIDIA Nsight Systems
Type: Profiling / Performance Analysis
​​
NVIDIA Nsight tools can be used to profile TornadoVM applications that run on NVIDIA GPUs through the PTX backend. Nsight provides detailed insights into runtime behavior, including CPU–GPU interaction, kernel launches, and low-level performance metrics. This allows developers to analyze and optimize TornadoVM-generated GPU kernels.
Nsight includes two complementary tools to help developers identify bottlenecks and improve GPU kernel performance.​
Nsight Systems
-
Timeline-based profiling
-
CPU–GPU overlap analysis
-
Kernel launch visualization
Nsight Compute
-
Kernel-level profiling
-
Hardware performance metrics
-
Optimization suggestions
What it offers
-
Timeline-based profiling of TornadoVM applications
-
Visualization of CPU–GPU interaction
-
CUDA kernel launch inspection
-
Kernel-level performance metrics and bottleneck analysis
-
Guidance on potential performance improvements
Intel VTune Profiler
Type: Profiling / Performance Analysis
​
Profile TornadoVM applications using Intel VTune to analyze CPU and GPU hotspots, inspect kernel execution, and obtain low-level performance metrics for OpenCL and SPIR-V kernels generated by TornadoVM.​
​
Intel VTune enables developers to analyze low-level execution details such as instruction cycles, hardware utilization, and the assembly generated from GPU kernels compiled from TornadoVM code.
What it offers
-
Hotspot analysis for TornadoVM applications
-
CPU and GPU performance profiling
-
Detailed kernel-level metrics
-
Insights into OpenCL and SPIR-V kernel execution
-
Identification of performance bottlenecks