Profiler class and native code to support self-profiling #2066
Conversation
This file contains 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
Signed-off-by: Jason Lowe <[email protected]>
|
build |
revans2
previously approved these changes
May 24, 2024
There was a problem hiding this comment.
Really just nits, but be aware I do not know CMake so if there are scary thing in there I probably missed them. I also am not an expert on the libraries or formats that you are using, so I am not sure 100% if they are all correct, but things appeared to line up properly.
|
build |
revans2
reviewed
May 24, 2024
revans2
previously approved these changes
May 24, 2024
|
build |
|
build |
|
build |
revans2
approved these changes
May 28, 2024
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Outdated suggestions cannot be applied.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.
Contributes to NVIDIA/spark-rapids#10632. Adds a new Profiler class and native code that can be used to collect CUDA and NVTX profiling data for the current process via CUPTI. It supports
The profiler is intended to be started before the process calls any CUDA events so it can capture everything. The profiler is shipped as a separate shared library in the spark-rapids-jni jar that can be loaded before the spark-rapids-jni shared native library that contains the cudf and spark-rapids-jni CUDA kernels.
The profiling data is serialized as a series of size-prefixed flatbuffers. See profiler.fbs for the schema. This also includes a converter tool, built in target/cmake-build/profiler/spark_rapids_profile_converter which can be used to convert the flatbuffer profiling data into various formats. It currently supports converting to JSON (useful for performing custom analysis on the raw data) or the Nsight NVTXT format. NVTXT can in turn be converted into an nsys-rep file that the Nsight Systems viewer can load via the ImportNvtxt tool supplied with Nsight Systems.