cuda.core 1.0.0 release prep: support policy, docs, and release notes#2032
Draft
leofang wants to merge 4 commits intoNVIDIA:mainfrom
Draft
cuda.core 1.0.0 release prep: support policy, docs, and release notes#2032leofang wants to merge 4 commits intoNVIDIA:mainfrom
leofang wants to merge 4 commits intoNVIDIA:mainfrom
Conversation
Add support.rst covering versioning (SemVer), CUDA version support (dual major versions), Python version support (CPython EOL schedule), free-threading (experimental), and release cadence (bimonthly). Closes NVIDIA#2030
Contributor
leofang
added
documentation
- Update cuda.coop and cuda.compute URLs from the old
nvidia.github.io/cccl/python/{coop,compute} paths (now 404)
to the current unstable doc paths.
- Add nvFatbin and NVML to the cuda.bindings interface list.
- Update all three synced files: README.md, cuda_python/DESCRIPTION.rst,
and cuda_python/docs/source/index.rst.
leofang
changed the title
Add new features (green contexts, system.Device NVML APIs, system.typing module, NVML enum re-wrapping), breaking changes (tensor bridge behavior, system.Device renames, privatized helper classes, UUID format change, removed enums), and bug fixes (is_managed for pool alloc, nvJitLink log error handling, NVML event set init, Device.arch unknown, empty field values, runtime error messages, wheel size reduction).
- api.rst: replace pre-1.0 warning with stable-API statement and link to support policy. - install.rst: update free-threading version reference from 0.4.0 to 1.0.0. - nv-versions.json: add 1.0.0 entry for the version switcher dropdown.
leofang
changed the title
leofang
added
P0
mdboom
requested changes
May 6, 2026
| :meth:`Device.create_context`, then use :meth:`Context.create_stream` and | ||
| :attr:`Context.resources` to work within the partitioned resources. | ||
| (`#1976 <https://github.com/NVIDIA/cuda-python/pull/1976>`__) | ||
| - Added the :mod:`cuda.core.system` module for NVIDIA Management Library (NVML) |
Contributor
There was a problem hiding this comment.
This was added in 0.6.0. Below is just changes (additions) to it.
Suggested change
| - Added the :mod:`cuda.core.system` module for NVIDIA Management Library (NVML) | |
| - Changes to :mod:`cuda.core.system` module for NVIDIA Management Library (NVML) |
Comment on lines
+50
to
+53
| - Moved all :mod:`cuda.core.system` enums into a new :mod:`cuda.core.system.typing` | ||
| module. Imports from ``cuda.core.system`` continue to work but the canonical | ||
| location is now ``cuda.core.system.typing``. | ||
| (`#2022 <https://github.com/NVIDIA/cuda-python/pull/2022>`__) |
Contributor
There was a problem hiding this comment.
This is really part of the note above. The fact that we merged it and them immediately moved it is uninteresting to our users. It's also dangerously incorrect (their old locations will not continue to work, publicly).
Suggested change
| - Moved all :mod:`cuda.core.system` enums into a new :mod:`cuda.core.system.typing` | |
| module. Imports from ``cuda.core.system`` continue to work but the canonical | |
| location is now ``cuda.core.system.typing``. | |
| (`#2022 <https://github.com/NVIDIA/cuda-python/pull/2022>`__) |
Comment on lines
+249
to
+252
| - Removed ``BrandType``, ``NvlinkVersion``, ``PcieUtilCounter``, ``Pstates``, | ||
| and ``TemperatureSensors`` enums from ``cuda.core.system``; the underlying | ||
| values are now returned as plain strings or accessed through other APIs. | ||
| (`#2014 <https://github.com/NVIDIA/cuda-python/pull/2014>`__) |
Contributor
There was a problem hiding this comment.
Covered by the enum changes above.
Suggested change
| - Removed ``BrandType``, ``NvlinkVersion``, ``PcieUtilCounter``, ``Pstates``, | |
| and ``TemperatureSensors`` enums from ``cuda.core.system``; the underlying | |
| values are now returned as plain strings or accessed through other APIs. | |
| (`#2014 <https://github.com/NVIDIA/cuda-python/pull/2014>`__) |
Comment on lines
+277
to
+279
| - Fixed a potential crash when NVML event set creation failed, due to | ||
| ``__dealloc__`` freeing an uninitialized handle. | ||
| (`#1992 <https://github.com/NVIDIA/cuda-python/pull/1992>`__) |
Contributor
There was a problem hiding this comment.
Suggested change
| - Fixed a potential crash when NVML event set creation failed, due to | |
| ``__dealloc__`` freeing an uninitialized handle. | |
| (`#1992 <https://github.com/NVIDIA/cuda-python/pull/1992>`__) | |
| - Fixed a potential crash when NVML event set creation failed on Windows, due to | |
| ``__dealloc__`` freeing an uninitialized handle. | |
| (`#1992 <https://github.com/NVIDIA/cuda-python/pull/1992>`__) |
|
|
||
| - **Major**: Bumped when a new CUDA major release is out and support for the oldest CUDA major | ||
| version is dropped. Breaking API changes only happen at major-version boundaries. | ||
| - **Minor**: Bumped when new, backward-compatible features are added, or when a new Python minor |
Contributor
There was a problem hiding this comment.
Python calls these "feature releases", because "minor" sort of makes sense because it's the second number, it undersells the importance of them (and it doesn't follow semver anyway).
Suggested change
| - **Minor**: Bumped when new, backward-compatible features are added, or when a new Python minor | |
| - **Minor**: Bumped when new, backward-compatible features are added, or when a new Python feature |
| ``cuda.core`` supports all Python versions following the `CPython EOL schedule | ||
| <https://devguide.python.org/versions/>`_. As of writing, Python 3.10 – 3.14 are supported. | ||
|
|
||
| When a new Python minor version is released and the oldest supported version reaches EOL, |
Contributor
There was a problem hiding this comment.
Suggested change
| When a new Python minor version is released and the oldest supported version reaches EOL, | |
| When a new Python feature version is released and the oldest supported version reaches EOL, |
| - Supported CUDA versions | ||
| * - 1.x | ||
| - 12, 13 | ||
|
|
Contributor
There was a problem hiding this comment.
Do we want to add that certain features in cuda_core may require a specific minimum version of cuda_bindings? (I'm thinking of NVML support specifically, but that may not be the only case). Saying it supports "any 12 or 13" is a bit misleading.
As a minimum, I'm going to file another pull request to document the minimum required cuda_bindings version that will set next to the cuda.core.system docs.
mdboom
reviewed
May 6, 2026
Comment on lines
+45
to
+49
| - Re-wrapped NVML enums as human-readable ``StrEnum`` subclasses instead of raw | ||
| integer re-exports from ``cuda.bindings.nvml``. Added | ||
| :class:`~system.typing.GpuP2PCapsIndex`, :class:`~system.typing.GpuP2PStatus`, | ||
| and :class:`~system.typing.GpuTopologyLevel` enums. | ||
| (`#2014 <https://github.com/NVIDIA/cuda-python/pull/2014>`__) |
Contributor
There was a problem hiding this comment.
Suggested change
| - Re-wrapped NVML enums as human-readable ``StrEnum`` subclasses instead of raw | |
| integer re-exports from ``cuda.bindings.nvml``. Added | |
| :class:`~system.typing.GpuP2PCapsIndex`, :class:`~system.typing.GpuP2PStatus`, | |
| and :class:`~system.typing.GpuTopologyLevel` enums. | |
| (`#2014 <https://github.com/NVIDIA/cuda-python/pull/2014>`__) | |
| - Re-wrapped NVML enums as human-readable ``StrEnum`` subclasses instead of raw | |
| integer re-exports from ``cuda.bindings.nvml``. These are available in | |
| ``cuda.core.system.typing``. |
Contributor
There was a problem hiding this comment.
It will probably generate a merge conflict as a reminder, but not to selves: /IF/ we choose to merge #2037, we should make sure typing -> enums here.
Contributor
There was a problem hiding this comment.
#2037 not being merged, so we are good as-is here.
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 join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Summary
cuda_core/docs/source/support.rstcovering versioning (SemVer), CUDA version support (dual major versions), Python version support (CPython EOL schedule), experimental free-threading support, and release cadence (bimonthly target). Closes Document cuda.core support policy #2030.cuda.coopandcuda.compute(old paths return 404), and add missingnvFatbinandNVMLto thecuda.bindingsinterface list across all 3 synced files (README.md,cuda_python/DESCRIPTION.rst,cuda_python/docs/source/index.rst).api.rstpre-1.0 warning to stable-API statement,install.rstfree-threading version reference, and add 1.0.0 entry tonv-versions.json.Closes #2030