Jonathan R. Madsen
593b3b69b8
* silence SFINAE disabled for fork_gotcha
* Python updates
- Options for --{module,function}-include
- libpyomnitrace is_initialized and is_finalized
- source instrumentation auto init
- atexit finalization
- improved python testing
* Documentation Update
* Fix to 'cmake -E cat' not available < cmake v3.18
* Fix for inverse tests
* Update cancelling.yml
1.6 KiB
About
.. toctree::
:glob:
:maxdepth: 4
Omnitrace is an AMD research project and should not be treated as an offical part of the ROCm software stack.
Browse Omnitrace source code on Github
Omnitrace is designed for both high-level and comprehensive application tracing and profiling on both the CPU and GPU. Omnitrace supports both binary instrumentation and sampling as a means of collecting various metrics.
Visualization of the comprehensive omnitrace results can be viewed in any modern web browser by visiting ui.perfetto.dev
and loading the perfetto output (.proto files) produced by omnitrace.
Aggregated high-level results are available in text files for human consumption and JSON files for programmatic analysis. The JSON output files are compatible with the python package hatchet which converts the performance data into pandas dataframes and facilitate multi-run comparisons, filtering, visualization in Jupyter notebooks, and much more.
Omnitrace has two distinct configuration steps:
- Configuring which functions and modules are instrumented in the target binaries (i.e. executable and/or libraries)
- Configuring what the instrumentation does happens when the instrumented binaries are executed