Weekly GitHub Report for Xla: June 16, 2025 - June 23, 2025 (12:01:14)
Weekly GitHub Report for Xla
Thank you for subscribing to our weekly newsletter! Each week, we deliver a comprehensive summary of your GitHub project's latest activity right to your inbox, including an overview of your project's issues, pull requests, contributors, and commit activity.
Table of Contents
I. News
1.1 Recent Version Releases:
No recent version releases were found.
1.2 Version Information:
To provide a summary, I would need the specific version release information, including the description and creation date. Please provide those details so I can assist you effectively.
II. Issues
2.1 Top 5 Active Issues:
We consider active issues to be issues that that have been commented on most frequently within the last week. Bot comments are omitted.
-
upgrade protobuf dependency: This issue is about upgrading the protobuf dependency in the project, as the current version is outdated and causes compatibility issues when using XLA with other libraries in the same Bazel workspace. The user who opened the issue highlights the need for this upgrade to ensure seamless integration and functionality.
- A comment mentions that an update is in progress, but it has been challenging due to incompatibility between the Bazel build and newer versions of protobuf, with a related pull request expected to be merged soon.
- Number of comments this week: 1
Since there were fewer than 5 open issues, all of the open issues have been listed above.
2.2 Top 5 Stale Issues:
We consider stale issues to be issues that has had no activity within the last 30 days. The team should work together to get these issues resolved and closed as soon as possible.
As of our latest update, there are no stale issues for the project this week.
2.3 Open Issues
This section lists, groups, and then summarizes issues that were created within the last week in the repository.
Issues Opened This Week: 4
Summarized Issues:
- Torch Model Export Errors: Users face issues when exporting a Torch model to MLIR using the torch-xla torchax export API. The problems include an unregistered operation error in the 'vhlo' dialect and a failure to deserialize a portable artifact using StableHLO_v1.9.5, despite using the correct versions of torch and torchxla.
- Protobuf Dependency Upgrade: The project is hindered by an outdated protobuf dependency, which affects the use of XLA with other libraries in the same Bazel WORKSPACE. Although an update is underway, it faces challenges due to incompatibility with newer protobuf versions.
- Bazel Modules Adoption: There is a request for the XLA project to adopt Bazel modules, as it is currently the only package in the user's Bazel build lacking support for them. This is despite the widespread adoption of Bazel modules in other projects.
- GPU Collective Performance Model Bug: A bug exists in the GPU collective performance model due to a recent update that adjusted the lowLatencyBandwidth for AMD links. However, it failed to update the CUDA section, causing failures when the model is used with H100 settings.
2.4 Closed Issues
This section lists, groups, and then summarizes issues that were closed within the last week in the repository. This section also links the associated pull requests if applicable.
Issues Closed This Week: 1
Summarized Issues:
- Compilation Error in FFI Binding: This issue involves a compilation error encountered when using the
Attr<char>()option in a Foreign Function Interface (FFI) binding. It is suspected to be due to a missingXLA_FFI_REGISTER_SCALAR_ATTR_DECODINGforchar, leading to a misleading error message rather than an actual bug in the FFI interface.
2.5 Issue Discussion Insights
This section will analyze the tone and sentiment of discussions within this project's open and closed issues that occurred within the past week. It aims to identify potentially heated exchanges and to maintain a constructive project environment.
Based on our analysis, there are no instances of toxic discussions in the project's open or closed issues from the past week.
III. Pull Requests
3.1 Open Pull Requests
This section provides a summary of pull requests that were opened in the repository over the past week. The top three pull requests with the highest number of commits are highlighted as 'key' pull requests. Other pull requests are grouped based on similar characteristics for easier analysis. Up to 25 pull requests are displayed in this section, while any remaining pull requests beyond this limit are omitted for brevity.
Pull Requests Opened This Week: 10
Key Open Pull Requests
1. [XLA:GPU][oneAPI] Enable Level_zero support: This pull request aims to enable Level Zero support for the XLA project by configuring the necessary header and library paths for Level Zero and MKL libraries, ensuring that required headers and libraries are downloaded for hermetic builds, and making various updates and fixes across multiple commits to support the stream_executor.
- URL: pull/28073
- Merged: No
- Associated Commits: 55eef, cd41a, b3214, 315aa, 0c60d, f42dc, e71a4, 05795, 45e02, 088ce, d3dc8, dccda, 2e2d3, 4f230, eeffe, 68be9, 96e68
2. [XLA:GPU][oneAPI] Enable Clang compiler as the host compiler: This pull request aims to configure the Clang compiler as the host compiler for the XLA project when using oneAPI, allowing developers to switch from the default GCC host compiler to Clang by enabling the --config=icpx_clang flag, and includes several commits that update build scripts and configuration files to support this change.
- URL: pull/27904
- Merged: No
3. Multihost HLO Runner: Add CLI Option for Output Mode: This pull request introduces a new command-line interface option --output_mode to the multihost_hlo_runner tool, allowing users to configure the output mode to either return outputs from all devices, return outputs only from logical device 0, or not return any outputs at all, thereby optimizing device-to-host data transfers and memory usage.
- URL: pull/27994
- Merged: No
Other Open Pull Requests
- Swish/Silu Fusion Migration: This topic involves migrating the swish/silu fusion process to the upstream by implementing a fusion that integrates with the hipblaslt kernel when compiling with ROCm version 7. The pull request includes implementing the swish fusion, setting a threshold for the ROCm version, and renaming swish to silu.
- ROCm Backend Adjustments: This topic covers adjustments in the ROCm backend, including addressing an issue with the reduction emitter by adjusting the maximum shuffling distance during warp reduction to 16. It also involves ensuring that the AMDGPU_TARGETS are correctly passed to the crosstool wrapper, rectifying an earlier fix that was not properly merged.
- ROCm Component Test Corrections: This topic involves correcting and adapting test names within the ROCm component of the openxla/xla project. The pull request is currently not merged and focuses on ensuring test accuracy and consistency.
- Intel GPU Backend Support: This topic aims to enhance the XLA:GPU project by updating the
build_defs.bzlfile to incorporate support for the Intel GPU backend specifically for testing purposes. The pull request is dependent on another pull request, #26217, authored by @agramesh1.
- oneDNN Convolution Weight Prepacking: This topic introduces weight prepacking for oneDNN convolutions, achieving full functional feature parity with oneDNN matmuls. The pull request details the addition of convolution weight prepacking support.
- ROCm Version 7 Build Fixes: This topic addresses a build issue encountered when compiling with ROCm version 7 by implementing necessary fixes. The pull request is currently open for review and aims to resolve the compilation issues.
3.2 Closed Pull Requests
This section provides a summary of pull requests that were closed in the repository over the past week. The top three pull requests with the highest number of commits are highlighted as 'key' pull requests. Other pull requests are grouped based on similar characteristics for easier analysis. Up to 25 pull requests are displayed in this section, while any remaining pull requests beyond this limit are omitted for brevity.
Pull Requests Closed This Week: 5
Key Closed Pull Requests
1. Ignore failing tests: This pull request, titled "Ignore failing tests," involves multiple commits addressing various issues such as fixing header paths in the Triton compiler, enabling FP8 for AMD architectures, adding support for gfx1101, using MIOPEN_BETA_API, and making several other improvements and fixes related to ROCm and JAX, but ultimately was not merged.
- URL: pull/27921
- Merged: No
- Associated Commits: 13d28, a7bb0, 08c9b, 9b74a, b03cd, c62e4, 53126, 4667e, ae2d3, d3f94, d0c29, 5be95, 497cf, 25316, 5b460, 109e1, 2e04d, 76eb7, 28f10, 6484d, ea4cd, 5c042, 55a8c, f4ead
2. [ROCm] Support rocm7 compilation: This pull request aims to support compilation with ROCm 7 by introducing additional specializations necessary for compatibility with ROCm 7 libraries, as evidenced by the commits focusing on enabling ROCm 7 compilation and updating specific source files.
- URL: pull/27872
- Merged: No
3. [XLA:GPU] instantiate cuda graph executor with node priority enabled: This pull request aims to instantiate the CUDA graph executor with node priority enabled in the XLA:GPU project, as indicated by the commit message and the title, although it was not merged.
- URL: pull/27868
- Merged: No
- Associated Commits: c242c
Other Closed Pull Requests
- ROCm CI build compilation errors: This topic addresses the resolution of ROCm CI build compilation errors caused by changes in the hermetic cc toolchain. The pull request provides a fix to ensure successful builds by adjusting the toolchain settings.
- New command-line interface option: A new command-line interface option
--output_modehas been introduced to themultihost_hlo_runner. This addition allows users to select from three output modes and includes error handling for invalid input values.
3.3 Pull Request Discussion Insights
This section will analyze the tone and sentiment of discussions within this project's open and closed pull requests that occurred within the past week. It aims to identify potentially heated exchanges and to maintain a constructive project environment.
Based on our analysis, there are no instances of toxic discussions in the project's open or closed pull requests from the past week.
IV. Contributors
4.1 Contributors
Active Contributors:
We consider an active contributor in this project to be any contributor who has made at least 1 commit, opened at least 1 issue, created at least 1 pull request, or made more than 2 comments in the last month.
If there are more than 10 active contributors, the list is truncated to the top 10 based on contribution metrics for better clarity.
| Contributor | Commits | Pull Requests | Issues | Comments |
|---|---|---|---|---|
| mraunak | 24 | 2 | 0 | 0 |
| alekstheod | 10 | 5 | 0 | 2 |
| amd-songpiao | 10 | 3 | 0 | 0 |
| yliu120 | 6 | 1 | 0 | 0 |
| Zoey-Cheng | 5 | 2 | 0 | 0 |
| shawnwang18 | 4 | 1 | 0 | 0 |
| sfvaroglu | 2 | 1 | 0 | 0 |
| reedwm | 0 | 0 | 0 | 3 |
| dimvar | 1 | 1 | 0 | 0 |
| akhilgoe | 1 | 1 | 0 | 0 |