[pull] master from tensorflow:master - #8787
Merged
Merged
Conversation
Differentiating through tf.nn.softsign twice failed with "LookupError: gradient registry has no entry for: SoftsignGrad", because every other activation's backward op has a Python gradient registration while SoftsignGrad did not. The new registration differentiates the kernel expression gradients / (1 + |features|)^2, giving -2 * gradients * sign(features) / (1 + |features|)^3 for the backprop input and reusing the kernel for the gradients input. Test Plan: Ran softsign_op_test.py against a pip tf-nightly build with the patched nn_grad.py overlaid: "Ran 5 tests in 0.461s / OK (skipped=1)" including the new testGradGrad; with the pristine nn_grad.py that test fails with the LookupError quoted above. Second derivatives verified against -2*sign(x)/(1+|x|)^3 at seven points from -0.9 to 0.9. pylint with tensorflow/tools/ci_build/pylintrc rates both changed Python files 10.00/10.
Replace the pow call with three multiplications, which is cheaper and avoids a pow round trip for a small integer exponent, and cover both float32 and float64 in the second derivative test instead of float64 only. Caught in code review on pull request 126092. Test Plan: softsign_op_test.py against the nightly overlay: "OK (skipped=1)". pylint --rcfile=tensorflow/tools/ci_build/pylintrc on both files: 10.00/10.
The new SoftsignGrad registration changes the gradient exclusion tables that pywrap_gradient_exclusions.cc holds: the registration reads only the grad argument and none of op.inputs, so the op gains a full unused-inputs entry. Regenerated with the documented generator entry point; the output for unmodified master reproduces the committed file byte for byte, and the only delta here is the single new SoftsignGrad line.
The regenerated table for the SoftsignGrad registration has 489 entries but the declared std::array size stayed at 488, which fails compilation with excess elements in array initializer. Bump the size to 489.
Teach CpuFloatSupport and OneDnnFloatSupport that BF16 and F16 are supported for kSort, kCompare, and kSelect. Previously, kSort operands and comparators were unconditionally upcast to F32 during FloatNormalization, preventing low-precision sorting from utilizing optimized inlined sort logic. Also add kSort to FloatSupport::SupportsMixedPrecisions so that key-value sorts with mixed element types (e.g. BF16 keys and S32 values) are preserved. PiperOrigin-RevId: 972429241
Migrate GpuAotCompilationTest from HloTestBaseLegacy to HloPjRtGpuTestBase. Configure AOT compilation options using gpu_target_config() instead of an explicit StreamExecutor, and execute the Triton AOT compiled module via test_runner() PjRt interfaces. PiperOrigin-RevId: 972429647
Imported from GitHub PR openxla/xla#47908 Bumps [ml-dtypes](https://github.com/jax-ml/ml_dtypes) from 0.5.4 to 0.6.0. <details> <summary>Release notes</summary> <p><em>Sourced from <a href="https://github.com/jax-ml/ml_dtypes/releases">ml-dtypes's releases</a>.</em></p> <blockquote> <h2>v0.6.0</h2> <ul> <li>Added new 16-bit complex types: <code>ml_dtypes.complex32</code> (based on float16) and <code>ml_dtypes.bcomplex32</code> (based on bfloat16) (<a href="https://redirect.github.com/jax-ml/ml_dtypes/pull/351">#351</a>).</li> <li>Added new 1-bit integer types: <code>ml_dtypes.int1</code> and <code>ml_dtypes.uint1</code>.</li> <li>Added <code>__format__</code> method to custom float, complex, and integer types (<a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/341">#341</a>). Previously, formatting custom scalars (e.g. in f-strings) fell back to string formatting, which could truncate exponents or fail on numeric format specifiers.</li> <li><code>ml_dtypes.finfo</code> and <code>ml_dtypes.iinfo</code> now allow passing an array object to their constructor (<a href="https://redirect.github.com/jax-ml/ml_dtypes/pull/350">#350</a>).</li> <li>Fixed equality comparison (<code>==</code>, <code>!=</code>) when comparing custom dtypes against incompatible types like strings or <code>None</code>.</li> <li>Dropped support for Python 3.9, which reached end-of-life in October 2025.</li> <li>Dropped support for Python 3.13 free-threading, because cibuildwheel dropped support.</li> <li>Dropped support for NumPy < 2.0.</li> <li>Added support for Python 3.15.</li> <li>Switched build system to <code>scikit-build-core</code> and CMake (<a href="https://redirect.github.com/jax-ml/ml_dtypes/pull/361">#361</a>).</li> <li>Updated Eigen dependency to v5.0.1.</li> </ul> </blockquote> </details> <details> <summary>Changelog</summary> <p><em>Sourced from <a href="https://github.com/jax-ml/ml_dtypes/blob/main/CHANGELOG.md">ml-dtypes's changelog</a>.</em></p> <blockquote> <h2>[0.6.0] - 2026-08-13</h2> <ul> <li>Added new 16-bit complex types: <code>ml_dtypes.complex32</code> (based on float16) and <code>ml_dtypes.bcomplex32</code> (based on bfloat16) (<a href="https://redirect.github.com/jax-ml/ml_dtypes/pull/351">#351</a>).</li> <li>Added new 1-bit integer types: <code>ml_dtypes.int1</code> and <code>ml_dtypes.uint1</code>.</li> <li>Added <code>__format__</code> method to custom float, complex, and integer types (<a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/341">#341</a>). Previously, formatting custom scalars (e.g. in f-strings) fell back to string formatting, which could truncate exponents or fail on numeric format specifiers.</li> <li><code>ml_dtypes.finfo</code> and <code>ml_dtypes.iinfo</code> now allow passing an array object to their constructor (<a href="https://redirect.github.com/jax-ml/ml_dtypes/pull/350">#350</a>).</li> <li>Fixed equality comparison (<code>==</code>, <code>!=</code>) when comparing custom dtypes against incompatible types like strings or <code>None</code>.</li> <li>Dropped support for Python 3.9, which reached end-of-life in October 2025.</li> <li>Dropped support for Python 3.13 free-threading, because cibuildwheel dropped support.</li> <li>Dropped support for NumPy < 2.0.</li> <li>Added support for Python 3.15.</li> <li>Switched build system to <code>scikit-build-core</code> and CMake (<a href="https://redirect.github.com/jax-ml/ml_dtypes/pull/361">#361</a>).</li> <li>Updated Eigen dependency to v5.0.1.</li> </ul> </blockquote> </details> <details> <summary>Commits</summary> <ul> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/6bc762dd106292e1aa0d5de98d3867d6b642f209"><code>6bc762d</code></a> Merge pull request <a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/389">#389</a> from hawkinsp:release</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/ce2c7bf3fee50cf27e51040501278130b2ffec9e"><code>ce2c7bf</code></a> Prepare v0.6.0 release.</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/66a79825099898d56e7c21abf62ada22f1210d9f"><code>66a7982</code></a> Merge pull request <a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/390">#390</a> from hawkinsp:npy2</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/2c29114e2f531f296c1725feb655af8887e04ef1"><code>2c29114</code></a> Drop support for NumPy 1.x.</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/3a8d5da4d08e693852b1c9fb7f0965762acc4bab"><code>3a8d5da</code></a> Merge pull request <a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/388">#388</a> from hawkinsp:build</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/dec45f7a89cdc5be813cca25d264107640aba859"><code>dec45f7</code></a> Merge pull request <a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/387">#387</a> from hawkinsp:format</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/2e0e13ec6826963dd1def83ff0c585135d7f7332"><code>2e0e13e</code></a> Clean up scikit-build-core configuration in pyproject.toml.</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/90bebd5eeef5972decb18b5ce9fbeefe1bdd1eff"><code>90bebd5</code></a> Implement <strong>format</strong> for custom float, complex, and int dtypes.</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/f9c639221c3d6275adfe0deddf1721e018082027"><code>f9c6392</code></a> Merge pull request <a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/368">#368</a> from jax-ml:dependabot/github_actions/actions/downloa...</li> <li><a href="https://github.com/jax-ml/ml_dtypes/commit/14396b0a40e6ba2fcf5ffd28e091194712cea118"><code>14396b0</code></a> Merge pull request <a href="https://redirect.github.com/jax-ml/ml_dtypes/issues/384">#384</a> from jax-ml:dependabot/github_actions/actions/setup-p...</li> <li>Additional commits viewable in <a href="https://github.com/jax-ml/ml_dtypes/compare/v0.5.4...v0.6.0">compare view</a></li> </ul> </details> <br /> [](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores) Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- <details> <summary>Dependabot commands and options</summary> <br /> You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot show <dependency name> ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself) </details> Copybara import of the project: -- 2e9317f2988a2dfb8d4cc32f7bd98e8e620e0c88 by dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>: Bump ml-dtypes from 0.5.4 to 0.6.0 Bumps [ml-dtypes](https://github.com/jax-ml/ml_dtypes) from 0.5.4 to 0.6.0. - [Release notes](https://github.com/jax-ml/ml_dtypes/releases) - [Changelog](https://github.com/jax-ml/ml_dtypes/blob/main/CHANGELOG.md) - [Commits](jax-ml/ml_dtypes@v0.5.4...v0.6.0) --- updated-dependencies: - dependency-name: ml-dtypes dependency-version: 0.6.0 dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com> Merging this change closes #47908 PiperOrigin-RevId: 972433287
Imported from GitHub PR openxla/xla#47910 Bumps [astral-sh/setup-uv](https://github.com/astral-sh/setup-uv) from 10.0.0 to 10.0.1. <details> <summary>Release notes</summary> <p><em>Sourced from <a href="https://github.com/astral-sh/setup-uv/releases">astral-sh/setup-uv's releases</a>.</em></p> <blockquote> <h2>v10.0.1 🌈 Tolerate transient manifest timeouts</h2> <h2>Changes</h2> <p>Thank you <a href="https://github.com/arguile"><code>@arguile</code></a>- for making this action more resilient.</p> <h2>🐛 Bug fixes</h2> <ul> <li>Tolerate transient manifest timeouts <a href="https://github.com/arguile"><code>@arguile</code></a>- (<a href="https://redirect.github.com/astral-sh/setup-uv/issues/1016">#1016</a>)</li> </ul> <h2>🧰 Maintenance</h2> <ul> <li>chore: update known checksums for 0.12.4 @<a href="https://github.com/apps/github-actions">github-actions[bot]</a> (<a href="https://redirect.github.com/astral-sh/setup-uv/issues/1017">#1017</a>)</li> </ul> <h2>📚 Documentation</h2> <ul> <li>docs: update version references to v10.0.0 @<a href="https://github.com/apps/github-actions">github-actions[bot]</a> (<a href="https://redirect.github.com/astral-sh/setup-uv/issues/1014">#1014</a>)</li> </ul> </blockquote> </details> <details> <summary>Commits</summary> <ul> <li><a href="https://github.com/astral-sh/setup-uv/commit/20cfd1bf945f4377ade1205e4dbc17946fc9a30d"><code>20cfd1b</code></a> chore: update known checksums for 0.12.4 (<a href="https://redirect.github.com/astral-sh/setup-uv/issues/1017">#1017</a>)</li> <li><a href="https://github.com/astral-sh/setup-uv/commit/d73a0cab66a532d7afa440d9df4a67ea9fe65a30"><code>d73a0ca</code></a> Tolerate transient manifest timeouts (<a href="https://redirect.github.com/astral-sh/setup-uv/issues/1016">#1016</a>)</li> <li><a href="https://github.com/astral-sh/setup-uv/commit/ae3b92d1bdb308a10adfe7b8f408e5cc8c30f3f6"><code>ae3b92d</code></a> docs: update version references to v10.0.0 (<a href="https://redirect.github.com/astral-sh/setup-uv/issues/1014">#1014</a>)</li> <li>See full diff in <a href="https://github.com/astral-sh/setup-uv/compare/ae62891fec2bb8e7d6c99fc78c9fec3a63790f8d...20cfd1bf945f4377ade1205e4dbc17946fc9a30d">compare view</a></li> </ul> </details> <br /> [](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores) Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- <details> <summary>Dependabot commands and options</summary> <br /> You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot show <dependency name> ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself) </details> Copybara import of the project: -- 466d0bbcf6e70095833a34245cc0a2d78e34aaf8 by dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>: Bump astral-sh/setup-uv from 10.0.0 to 10.0.1 Bumps [astral-sh/setup-uv](https://github.com/astral-sh/setup-uv) from 10.0.0 to 10.0.1. - [Release notes](https://github.com/astral-sh/setup-uv/releases) - [Commits](astral-sh/setup-uv@ae62891...20cfd1b) --- updated-dependencies: - dependency-name: astral-sh/setup-uv dependency-version: 10.0.1 dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com> Merging this change closes #47910 PiperOrigin-RevId: 972433551
Imported from GitHub PR openxla/xla#47951 📝 Summary of Changes - Resolve the active GPU platform instead of hardcoding CUDA/H100. - Build `GpuTargetConfig` from the active StreamExecutor. - Use `xla_test(backends = ["gpu"])` so CUDA and ROCm receive the correct generated targets and CI tags. - Preserve existing CUDA behavior. 🎯 Justification Add more unit test coverage on ROCm platform. 🚀 Kind of Contribution 🧪 Tests Copybara import of the project: -- b20e115b198d8cadec00583b3546a78fa8877bd0 by Lin Chen1 <lin.chen1@amd.com>: Enable command buffer profiling test on ROCm Merging this change closes #47951 PiperOrigin-RevId: 972434151
Imported from GitHub PR openxla/xla#48007 oneAPI 2026.x tears down libsycl/Level Zero earlier at process exit than pre-2026.x. If SYCL runtime statics such as `device_pool_`, `stream_pool_map_`, and the cached `::sycl::context` run their destructors as usual, they touch already torn-down libsycl/Level Zero state, corrupting the heap (observed as `double free or corruption (!prev)` crashes immediately after tests report PASSED). This PR wraps these statics in `absl::NoDestructor` so they're intentionally leaked instead of destructed at exit. The OS reclaims this memory when the process exits, which is acceptable since this is a one-time allocation per program run. Copybara import of the project: -- 06e491e4cda93ea84b27d48956b1e41d20a839f6 by Bhavani Subramanian <bhavani1.subramanian@intel.com>: Leak SYCL runtime statics to fix double-free at process exit. oneAPI 2026.x tears down libsycl/Level Zero earlier at process exit than pre-2026.x. If device_pool_, stream_pool_map_, and the cached ::sycl::context run their destructors as usual, they touch already torn-down libsycl/Level Zero state, corrupting the heap ("double free or corruption (!prev)"). Wrap these statics in absl::NoDestructor so they're intentionally leaked instead of destructed at exit. The OS reclaims this memory when the process exits, which is acceptable since this is a one-time allocation per program run. Merging this change closes #48007 PiperOrigin-RevId: 972434558
Imported from GitHub PR openxla/xla#47950 📝 Summary of Changes - Remove the stale `cuda-only` tag from `dynamic_slice_fusion_rewriter_v2_test`. - Enable this backend-agnostic test in ROCm CI. - Preserve existing CUDA behavior. 🎯 Justification Add more unit test coverage on ROCm platform 🚀 Kind of Contribution 🧪 Tests Copybara import of the project: -- fbac1d2bdc2d6021168e5a484fcf49f6b3a6f63f by Lin Chen1 <lin.chen1@amd.com>: Enable dynamic slice fusion rewriter v2 test on ROCm Merging this change closes #47950 PiperOrigin-RevId: 972435961
Imported from GitHub PR openxla/xla#47909 Bumps [github/codeql-action/upload-sarif](https://github.com/github/codeql-action) from 4.37.6 to 4.37.7. <details> <summary>Release notes</summary> <p><em>Sourced from <a href="https://github.com/github/codeql-action/releases">github/codeql-action/upload-sarif's releases</a>.</em></p> <blockquote> <h2>v4.37.7</h2> <ul> <li>Update default CodeQL bundle version to <a href="https://github.com/github/codeql-action/releases/tag/codeql-bundle-v2.26.3">2.26.3</a>. <a href="https://redirect.github.com/github/codeql-action/pull/4085">#4085</a></li> </ul> </blockquote> </details> <details> <summary>Changelog</summary> <p><em>Sourced from <a href="https://github.com/github/codeql-action/blob/main/CHANGELOG.md">github/codeql-action/upload-sarif's changelog</a>.</em></p> <blockquote> <h1>CodeQL Action Changelog</h1> <p>See the <a href="https://github.com/github/codeql-action/releases">releases page</a> for the relevant changes to the CodeQL CLI and language packs.</p> <h2>[UNRELEASED]</h2> <ul> <li>Update default CodeQL bundle version to <a href="https://github.com/github/codeql-action/releases/tag/codeql-bundle-v2.26.4">2.26.4</a>. <a href="https://redirect.github.com/github/codeql-action/pull/4106">#4106</a></li> </ul> <h2>4.37.8 - 21 Aug 2026</h2> <p>No user facing changes.</p> <h2>4.37.7 - 13 Aug 2026</h2> <ul> <li>Update default CodeQL bundle version to <a href="https://github.com/github/codeql-action/releases/tag/codeql-bundle-v2.26.3">2.26.3</a>. <a href="https://redirect.github.com/github/codeql-action/pull/4085">#4085</a></li> </ul> <h2>4.37.6 - 04 Aug 2026</h2> <ul> <li>Changed the default filepath for the new remote file address format that was introduced in CodeQL Action 4.37.0 / 3.37.0 to <code>.github/codeql-config.yml</code> to align it with the suggested path that is used elsewhere. <a href="https://redirect.github.com/github/codeql-action/pull/4070">#4070</a></li> </ul> <h2>4.37.5 - 03 Aug 2026</h2> <ul> <li>Fixed a bug where a network error while streaming the download of the CodeQL bundle could terminate the <code>init</code> Action instead of falling back to downloading the bundle before extracting it. <a href="https://redirect.github.com/github/codeql-action/pull/4061">#4061</a></li> </ul> <h2>4.37.4 - 29 Jul 2026</h2> <ul> <li>This version of the CodeQL Action adds support for the <code>tools</code> input for the <code>codeql-action/init</code> step to be specified using a <code>github-codeql-tools</code> <a href="https://docs.github.com/en/organizations/managing-organization-settings/managing-custom-properties-for-repositories-in-your-organization">repository property</a>. This feature will gradually be rolled out following the release of this version. Once rolled out, this allows for the CodeQL CLI version that is used in GitHub-managed workflows, such as Default Setup, to be set to a custom value. For example, customers who run into issues with rate limits when a new CodeQL CLI version is released can set the value to <code>toolcache</code> to always use the CodeQL CLI version that is available in the runner toolcache. For Advanced Setup workflows, the value provided for <code>tools</code> in the workflow definition always takes precedence unless the value of the repository property starts with <code>!</code>. <a href="https://redirect.github.com/github/codeql-action/pull/4037">#4037</a></li> <li>Update default CodeQL bundle version to <a href="https://github.com/github/codeql-action/releases/tag/codeql-bundle-v2.26.2">2.26.2</a>. <a href="https://redirect.github.com/github/codeql-action/pull/4051">#4051</a></li> </ul> <h2>4.37.3 - 22 Jul 2026</h2> <p>No user facing changes.</p> <h2>4.37.2 - 21 Jul 2026</h2> <ul> <li>The new address format for the <code>config-file</code> input that was introduced in CodeQL Action 4.37.0 is now enabled by default. In addition to the format described there, the <code>remote=</code> prefix can now be used to explicitly indicate that the input refers to a remote file. All previous input formats continue to be accepted as well. <a href="https://redirect.github.com/github/codeql-action/pull/4023">#4023</a></li> <li>The CodeQL Action can now make use of <a href="https://docs.github.com/en/code-security/how-tos/secure-at-scale/configure-organization-security/manage-usage-and-access/giving-org-access-private-registries">configured private registries</a> in Default Setup to retrieve CodeQL configuration files from remote repositories that require authentication. This will allow customers to store their CodeQL configuration in a single repository that can then be referenced by Default Setup workflows in other repositories. We expect to roll this and other, related changes out to everyone in July. <a href="https://redirect.github.com/github/codeql-action/pull/4007">#4007</a></li> </ul> <h2>4.37.1 - 16 Jul 2026</h2> <ul> <li><em>Upcoming breaking change</em>: Add a deprecation warning for customers using CodeQL version 2.20.6 and earlier. These versions of CodeQL were discontinued on 1 July 2026 alongside GitHub Enterprise Server 3.16, and will be unsupported by the next minor release of the CodeQL Action. <a href="https://redirect.github.com/github/codeql-action/pull/3956">#3956</a></li> <li>Update default CodeQL bundle version to <a href="https://github.com/github/codeql-action/releases/tag/codeql-bundle-v2.26.1">2.26.1</a>. <a href="https://redirect.github.com/github/codeql-action/pull/4019">#4019</a></li> </ul> <h2>4.37.0 - 08 Jul 2026</h2> <ul> <li>Update default CodeQL bundle version to <a href="https://github.com/github/codeql-action/releases/tag/codeql-bundle-v2.26.0">2.26.0</a>. <a href="https://redirect.github.com/github/codeql-action/pull/3995">#3995</a></li> <li>In addition to the existing input format, the <code>config-file</code> input for the <code>codeql-action/init</code> step will soon support a new <code>[owner/]repo[@ref][:path]</code> format. All components except the repository name are optional. If omitted, <code>owner</code> defaults to the same owner as the repository the analysis is running for, <code>ref</code> to <code>main</code>, and <code>path</code> to <code>.github/codeql-action.yaml</code>. Support for this format ships in this version of the CodeQL Action, but will only be enabled over the coming weeks. <a href="https://redirect.github.com/github/codeql-action/pull/3973">#3973</a></li> </ul> <h2>4.36.3 - 01 Jul 2026</h2> <!-- raw HTML omitted --> </blockquote> <p>... (truncated)</p> </details> <details> <summary>Commits</summary> <ul> <li><a href="https://github.com/github/codeql-action/commit/ff2f1c621b7f889edc0d3c761ac2e6a3f8cdb0dd"><code>ff2f1c6</code></a> Merge pull request <a href="https://redirect.github.com/github/codeql-action/issues/4093">#4093</a> from github/update-v4.37.7-be7a3dbb8</li> <li><a href="https://github.com/github/codeql-action/commit/951a133f96aa2114dd747e9e437305335d0bde16"><code>951a133</code></a> Update changelog for v4.37.7</li> <li><a href="https://github.com/github/codeql-action/commit/be7a3dbb8147b82cd6d27e0707105b36aa190fc1"><code>be7a3db</code></a> Merge pull request <a href="https://redirect.github.com/github/codeql-action/issues/4087">#4087</a> from github/dependabot/npm_and_yarn/npm-minor-0aa561...</li> <li><a href="https://github.com/github/codeql-action/commit/9310334b11405b305d9444edfa56cd86e2f1e4fe"><code>9310334</code></a> Merge pull request <a href="https://redirect.github.com/github/codeql-action/issues/4086">#4086</a> from github/mbg/thread-action-state-to-codeql</li> <li><a href="https://github.com/github/codeql-action/commit/b4d8a54218a8792de9af2f6f32e33af899ca5212"><code>b4d8a54</code></a> Rebuild</li> <li><a href="https://github.com/github/codeql-action/commit/ab5db2519c3344f2fa61c711fa2d6ad135829200"><code>ab5db25</code></a> Bump the npm-minor group across 1 directory with 8 updates</li> <li><a href="https://github.com/github/codeql-action/commit/38055a3c3cf3979323eaf70fc6c73a8690250bde"><code>38055a3</code></a> Drop <code>logger</code> from <code>databaseInitCluster</code> in interface</li> <li><a href="https://github.com/github/codeql-action/commit/1f87aed5e66849f0c43ae147377cc77f2d98ac99"><code>1f87aed</code></a> Merge pull request <a href="https://redirect.github.com/github/codeql-action/issues/4085">#4085</a> from github/update-bundle/codeql-bundle-v2.26.3</li> <li><a href="https://github.com/github/codeql-action/commit/dc1b98ad1c2f13ccf9fc33fb82f32fc76f944253"><code>dc1b98a</code></a> Make <code>logger</code> available to <code>getCodeQLForCmd</code></li> <li><a href="https://github.com/github/codeql-action/commit/6f0220ee37121218af472efbde25f06907a4da4f"><code>6f0220e</code></a> Merge pull request <a href="https://redirect.github.com/github/codeql-action/issues/4084">#4084</a> from github/navntoft/bump-undici</li> <li>Additional commits viewable in <a href="https://github.com/github/codeql-action/compare/5595ccaf912efad79be6eef63a5619ff05969be3...ff2f1c621b7f889edc0d3c761ac2e6a3f8cdb0dd">compare view</a></li> </ul> </details> <br /> [](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores) Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- <details> <summary>Dependabot commands and options</summary> <br /> You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot show <dependency name> ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself) </details> Copybara import of the project: -- b42df13028500dd6dc74cead0ea0a2e81d1f9f66 by dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>: Bump github/codeql-action/upload-sarif from 4.37.6 to 4.37.7 Bumps [github/codeql-action/upload-sarif](https://github.com/github/codeql-action) from 4.37.6 to 4.37.7. - [Release notes](https://github.com/github/codeql-action/releases) - [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md) - [Commits](github/codeql-action@5595cca...ff2f1c6) --- updated-dependencies: - dependency-name: github/codeql-action/upload-sarif dependency-version: 4.37.7 dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com> Merging this change closes #47909 PiperOrigin-RevId: 972436016
PiperOrigin-RevId: 972449593
…vative PiperOrigin-RevId: 972450089
…rnel_info XStat for XProf Imported from GitHub PR openxla/xla#47917 ## 📝 Summary of Changes - Populate ROCm `kernel_details` XStat with: - Architecture + accumulator VGPRs per work-item. - Static LDS from kernel symbol metadata. - Dynamic LDS derived from dispatch total minus static LDS. - XProf-compatible `regs`, `static_shared`, and `dynamic_shared` keys. - Adjust inline comments regarding existing `private_segment_size` and `group_segment_size` stats. ## 🎯 Justification On AMD GPU traces, XProf Kernel Stats page reads 0 for "Registers per thread" and "Shared Mem Bytes". This is because XProf parses these fields on specific keys in the `kernel_info` XStat, `regs`, `static_shared` and `dynamic_shared`, none of which we were populating from `rocm_collector`. ### Implementation detail: - Why do we need static/dynamic group segment size separately when we already have total group segment size? - XProf parses on static, dynamic group size separately. The group segment size stat provided by rocprofiler combines static and dynamic, so we need to manually decompose into static and dynamic attribution. - Why not add dynamic mem member (eg. `dynamic_group_segment_size`) to KernelDetails struct if we are already adding `static_group_segment_size`? - The static LDS size is fixed by the compiled kernel symbol and does not vary between dispatches of that symbol. Rocprofiler provides the total LDS allocation at dispatch time. The dynamic size is derived as (dispatch total - static LDS) and can vary between dispatches due to runtime allocation and padding. - Why not just change XProf to parse our `group_mem` instead of following this existing static + dynamic separation? - Would need to vendor gate in XProf as NVIDIA provides static, dynamic separately. Besides, adding separation for static vs dynamic LDS is a useful detail to have. - Rationale for defining registers per work item as arch VGPR + accum VGPR - `registers_per_work_item` sums `arch_vgpr_count + accum_vgpr_count` because both are per-lane allocations consuming the VGPR budget. SGPRs are per-wavefront (not per work item/thread) scalar registers in a separate register file and so are excluded. ### Evidence (XProf v2.23.1, profiled via JAX) VIEWING AMD TRACE GENERATED BEFORE PR CHANGES: <img width="1887" height="907" alt="image" src="https://github.com/user-attachments/assets/90a11231-29ec-4112-9276-3a9367579a78" /> VIEWING AMD TRACE AFTER PR CHANGES: <img width="1885" height="906" alt="image" src="https://github.com/user-attachments/assets/917bfb08-d66d-4827-b585-0a5582e4507a" /> ## 🚀 Kind of Contribution 🐛 Bug Fix ## 📊 Benchmark (for Performance Improvements) N/A ## 🧪 Unit Tests: `TEST(RocmCollectorTest, ToXStatDecomposesDispatchGroupMemory)` Verifies correct decomposition of total group-segment size into static and dynamic LDS, including underflow protection. ## 🧪 Execution Tests: N/A Copybara import of the project: -- 820a26cc795bab9176983bfce6e8c5e91324ca47 by Clarke Chong <clchong@amd.com>: Write VGPR and LDS kernel stats to kernel_info XStat for XProf Merging this change closes #47917 PiperOrigin-RevId: 972450467
ComputationPlacer::GetForPlatform now returns a default ComputationPlacer instance when no platform-specific placer is registered. This removes the need for computation_placer.cc to hardcode static registrations for common platforms (host, cuda, rocm, sycl), eliminates dependencies on platform-specific platform IDs, and allows removing alwayslink = True from the target. PiperOrigin-RevId: 972461341
PiperOrigin-RevId: 972462168
PiperOrigin-RevId: 972463068
This enables lowering of mixed FP8 dot operations to convolution. PiperOrigin-RevId: 972464941
Reverts d9a8da7 PiperOrigin-RevId: 972476125
PiperOrigin-RevId: 972477162
… force instructions to be pinned, kept in default memory, or operands prefetched starting at a specified point. PiperOrigin-RevId: 972481193
PiperOrigin-RevId: 972505706
Use a more generic name since we will add more non-LLVM backends. PiperOrigin-RevId: 972505937
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
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
11 participants
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.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )