[AIU profiler] Introduce AIU profiler by following Kineto XPU plugin design #1126
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
…torch#961) Summary: In this PR, the AIU profiler is introduced by following kineto plugin design under libkineto/src/plugin/xpupti. The AIU profiler is build on the foundation of IBM AIU Profiling Toolkit. The plugin is disabled by default and is only activated if the system that kineto is build has the AIUPti installed and LIBKINETO_NOAIUPTI is not set to OFF. Signed-off-by: Marcelo Amaral <[email protected]>
Signed-off-by: Marcelo Amaral <[email protected]>
|
@zejun-chen since you created the PR to add support for XPU, could you please review this PR? |
|
@aaronenyeshi since you reviewed the PR to add support for XPU support, could you please also review this PR? |
mcalman
approved these changes
Aug 26, 2025
| { | ||
| std::lock_guard<std::mutex> guard(mutex_); | ||
|
|
||
| // Differently other backends, aiuptiFlushAllActivities flushes all pending |
I can review. |
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Outdated suggestions cannot be applied.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.
Summary
This PR introduces a new IBM AIU (Artificial Intelligence Unit) plugin for Kineto.
The AIU is a dedicated accelerator designed to efficiently run large-scale AI workloads. This integration extends Kineto’s profiling capabilities to support AIU backend, enabling developers to gain detailed insights into AIU execution behavior.
Motivation
Kineto currently supports multiple backends (e.g., CUDA, ROCm, XPU) for profiling. With the growing use of IBM’s AIU in research and production AI systems, it is essential to provide native support for profiling this hardware. This allows PyTorch developers and researchers to:
Features
Changes
Tested
This PR is inspired in the plugin design introduced by the PR #961