Add Nsight profiling support (nsys/ncu) #244
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Introduced `host_to_comm_path()` in `CommunicationLayer` and implemented Docker-specific logic in `DockerCommLayer` to resolve host/container path mapping (e.g., for file outputs inside mounted volumes). - Added new example: `campaign_nsys_ncu.py` showcasing how to run GPU benchmarks with `nsys` and `ncu` wrappers, including post-run hooks for metrics extraction. - Created `NsysWrap` and `NcuWrap`. `NsysWrap` runs `nsys profile`, then extracts memory usage stats from `report_cuda_gpu_mem_size_sum.csv`. `NcuWrap` runs `ncu` in CSV mode and parses per-kernel metrics from the log. - Refactored `AddVecBench` into its own reusable file under `examples/gpus/kit/addvec.py`. - Updated `gpus.py` to install Nsight Systems and libsmctrl by default in the Docker image. - Added a realistic CUDA benchmark `simplesleep.cu` to simulate kernel workloads with artificial delay and multiple phases. These changes improve support for automated GPU profiling and set the foundation for deeper performance analysis using NVIDIA's Nsight tooling inside Docker-based benchkit platforms. Signed-off-by: Antonio Paolillo <[email protected]>
Signed-off-by: Antonio Paolillo <[email protected]>
Signed-off-by: Antonio Paolillo <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Introduced
host_to_comm_path()inCommunicationLayerand implemented Docker-specific logic inDockerCommLayerto resolve host/container path mapping (e.g., for file outputs inside mounted volumes).Added new example:
campaign_nsys_ncu.pyshowcasing how to run GPU benchmarks withnsysandncuwrappers, including post-run hooks for metrics extraction.Created
NsysWrapandNcuWrap.NsysWraprunsnsys profile, then extracts memory usage stats fromreport_cuda_gpu_mem_size_sum.csv.NcuWraprunsncuin CSV mode and parses per-kernel metrics from the log.Refactored
AddVecBenchinto its own reusable file underexamples/gpus/kit/addvec.py.Updated
gpus.pyto install Nsight Systems and libsmctrl by default in the Docker image.Added a realistic CUDA benchmark
simplesleep.cuto simulate kernel workloads with artificial delay and multiple phases.These changes improve support for automated GPU profiling and set the foundation for deeper performance analysis using NVIDIA's Nsight tooling inside Docker-based benchkit platforms.