Dinamite Toolkit Quick Start User Guide Publications Technical Articles

DINAMITE toolkit

DINAMITE is a unique set of tools that deliver program comprehension and performance insight.

  • Have you ever felt frustrated when debugging a performance issue in a large multithreaded application?
  • Have you ever fumed over limitations of performance tools like perf?
  • Have you ever wanted to get more insight into a large codebase written by other people and understand not only where it spends CPU cycles, but how it actually works?

If you answered “yes” to any of these questions, DINAMITE is for you.

  • DINAMITE automatically instruments C/C++ programs to log names and timestamps of all or selected functions and/or variables into a log file, as the program executes.
  • As you run the program, it produces the execution log.
  • You can analyze the logs using sophisticated and visually rich tools. Or just use grep ;).

DINAMITE LLVM compiler is available on GitHub

The following interactive visualization tools are parts of DINAMITE’s performance debugging ecosystem. Be sure to check them out!

  • TSViz generates interactive graphs for visualizing communication between threads
  • Timesquared visualizes an annotated time line of per-thread call stacks

If you want to try out DINAMITE, click on Quick Start and follow the instructions. If you want to use it with a large complex code base, take a look at Technical Articles. If you need help using DINAMITE please get in touch

A typical DINAMITE workflow goes as follows:

  1. Build the target program with clang/clang++ and link it with libinstrumentation.so
  2. Choose a logging library flavour and build it. We currently provide three example logging libraries: binary filesystem output, text file system output and binary TCP output. The logging library emits logs of instrumented events. For more details see our page on Logging libraries
  3. Execute the instrumented binary. The log files will be created in your file system or streamed over the network.
  4. Analyze traces using our tools, 3rd party tools or Unix data analytics tools.

To get started with a simple example, read the Quick Start guide. For examples of using DINAMITE with complex systems, from integration into the build system to sophisticated trace analysis, check out Technical Articles.