Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
185 changes: 185 additions & 0 deletions cuda_bindings_12/docs/source/release/12.9.11-notes.rst
Original file line number Diff line number Diff line change
@@ -0,0 +1,185 @@
.. SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
.. SPDX-License-Identifier: Apache-2.0

.. module:: cuda.bindings

``cuda-bindings`` 12.9.11 Release notes
=======================================

Unreleased

Breaking Changes
----------------

* ``nvml.NvlinkFirmwareInfo.firmware_version`` now exposes only valid entries,
rather than all 100 slots in the underlying buffer. Set this member to update
the valid-entry count; the separate ``num_valid_entries`` member has been
removed. Use ``len(info.firmware_version)`` to query the count.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* Reserved members of ``runtime`` struct wrappers are no longer exposed as
Python attributes. Accessing these members now raises ``AttributeError``:

* ``cudaArraySparseProperties.reserved``,
``cudaArrayMemoryRequirements.reserved``, ``cudaMemcpyNodeParams.reserved``,
``cudaFuncAttributes.reserved``, ``cudaMemPoolProps.reserved``,
``cudaMemPoolPtrExportData.reserved``, ``cudaDeviceProp.reserved``,
``cudaEglPlaneDesc_st.reserved``
* ``cudaIpcEventHandle_st.reserved``, ``cudaIpcMemHandle_st.reserved``,
``cudaMemFabricHandle_st.reserved`` (also removed from their ``_t`` aliases)
* ``cudaExternalSemaphoreSignalParams.params.nvSciSync.reserved``,
``cudaExternalSemaphoreSignalParams.params.reserved``,
``cudaExternalSemaphoreSignalParams.reserved``,
``cudaExternalSemaphoreWaitParams.params.nvSciSync.reserved``,
``cudaExternalSemaphoreWaitParams.params.reserved``,
``cudaExternalSemaphoreWaitParams.reserved``
* ``cudaGraphNodeParams.reserved0``, ``cudaGraphNodeParams.reserved1``,
``cudaGraphNodeParams.reserved2``, ``cudaGraphEdgeData_st.reserved``

Storage allocated by these wrappers is zero-initialized, including reserved
fields, so callers do not need to zero these fields explicitly.

To serialize IPC handles previously accessed through ``handle.reserved``,
copy their raw bytes using ``getPtr()`` and ``ctypes`` instead:

.. code-block:: python

import ctypes

from cuda.bindings import runtime

# handle was returned by runtime.cudaIpcGetMemHandle().
payload = ctypes.string_at(handle.getPtr(), runtime.CUDA_IPC_HANDLE_SIZE)
restored = runtime.cudaIpcMemHandle_t()
ctypes.memmove(restored.getPtr(), payload, runtime.CUDA_IPC_HANDLE_SIZE)

When restoring a payload received from another process, ensure it contains
exactly ``runtime.CUDA_IPC_HANDLE_SIZE`` bytes before copying it.

Enhancements
------------

* Expanded NVML coverage with ``device_get_gpu_uptime_v1``,
``system_get_conf_compute_settings_v2``, ``device_get_nvlink_info_v3``,
``device_get_nvlink_remap_table_info_v1``, ``device_set_nvlink_nvle_ready_v1``,
``device_lock_nvlink_remap_table_v1``,
``device_setup_nvlink_nvle_encryption_key_v1``, and
``device_get_nvlink_nvle_private_info_v1``, together with their struct wrappers.
These APIs require an NVML library that provides the corresponding functions.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* Buffer-like pointer arguments in ``nvfatbin``, ``nvjitlink``, and ``nvvm`` now
accept contiguous buffers of any dimensionality, including ctypes objects
and multidimensional NumPy arrays. Previously these buffers had to be 1D.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* Pointer arguments in ``cufile`` and ``nvml``, such as ``descr`` in
``cufile.handle_register`` and ``info`` in ``nvml.device_set_clock_offsets``,
now accept ``None`` as ``NULL``.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)

Bug Fixes
---------

* ``driver.cuLaunchHostFunc`` and ``runtime.cudaLaunchHostFunc`` now pass the
host function and its user data to CUDA unchanged. The previous internal
trampoline freed its state after the first invocation, causing a captured
host function to crash on the second launch of a CUDA graph. Callers must
retain the callback object and user data storage until all uses have completed,
including the lifetime of graphs and their executables; see
:ref:`host-function-lifetime`.
(`PR #3107 <https://github.com/NVIDIA/cuda-python/pull/3107>`_)
* Corrected the driver-version lookup for ``cuLogs*`` functions to CUDA 12.9,
preventing spurious ``FunctionNotFoundError`` failures on supported drivers.
(`PR #3377 <https://github.com/NVIDIA/cuda-python/pull/3377>`_,
`PR #3395 <https://github.com/NVIDIA/cuda-python/pull/3395>`_)
* Fixed crashes when registering or unregistering ``cuLogs*`` and asynchronous
notification callbacks concurrently under free-threaded Python.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* Nested ``driver`` and ``runtime`` struct and union views now retain their
owning wrapper, and struct-array getters in these modules return independent
copies, avoiding access to freed backing storage. Mutating a returned element
no longer updates the parent struct: read the array, modify its elements, then
assign it back (for example, ``attrs = cfg.attrs``, update ``attrs[0]``, then
``cfg.attrs = attrs``). ``cufile`` and ``nvml`` struct wrappers also retain
owners correctly when indexing and slicing arrays.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* ``cufile`` and ``nvml`` struct arrays are now zero-initialized, and assigning
to a member of a read-only array wrapper raises ``ValueError``.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* ``nvml.device_get_field_values`` initializes per-query status to
``ERROR_UNKNOWN`` so entries left unpopulated by NVML do not appear successful.
Integer field IDs now explicitly use scope ID zero.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* ``nvml.PwrModelMetricsPfpp1x`` and ``nvml.PwrModelMetricsDlppm1x`` now populate
their nested ``estimated_metrics`` arrays using each element's
``num_vf_points`` and ``num_estimated_metrics``, respectively.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)
* Moved cuFile and NVML dtype layout calculations to heap storage to avoid
placing large temporary structs on the stack during module initialization.
(`PR #3473 <https://github.com/NVIDIA/cuda-python/pull/3473>`_)
* Nested resource pointer conversion accepts both ``str`` and ``bytes``
options, matching the CUDA 13 bindings behavior.
* Coredump settings retain borrowed string buffers for the duration of the
call and free the buffers allocated for getters.
* ``get_cuda_native_handle`` preserves a ``KeyError`` raised inside a
registered getter instead of reporting an unknown type.
(`PR #3315 <https://github.com/NVIDIA/cuda-python/pull/3315>`_)

Deprecations
------------

* The NVML exception classes ``nvml.TimeoutError`` and ``nvml.MemoryError``
have been renamed to ``nvml.NvmlTimeoutError`` and ``nvml.NvmlMemoryError``
to avoid shadowing Python builtins. The old names remain available as
compatibility aliases. Both classes still derive from ``nvml.NvmlError``.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_)

End of Life
-----------

* Python 3.10 support dropped (end of life); the supported range is
Python 3.11-3.15.

Packaging and Source Builds
---------------------------

* Wheels omit ``.pyx`` implementation sources and ``.in`` generation templates,
following the CUDA 13 packaging policy. Cython declarations and includes
(``.pxd`` and ``.pxi``), and required headers, remain available for downstream
Cython builds. Source distributions retain the inputs needed to build the package.
* All binding sources, including the runtime's internal and Cython layers, are
now pre-generated. Source builds no longer parse CUDA headers or expand
templates, removing the ``pyclibrary`` build dependency and the
``CUDA_PYTHON_PARSER_CACHING`` setting.
(`PR #3429 <https://github.com/NVIDIA/cuda-python/pull/3429>`_,
`PR #3437 <https://github.com/NVIDIA/cuda-python/pull/3437>`_)
* Source builds use the monorepo's shared compiler and Cython-cache tooling and
check for CUDA 12.9 headers before Cythonization, reporting a clear error when
the selected toolkit's major.minor version does not match the generated sources.
(`PR #3173 <https://github.com/NVIDIA/cuda-python/pull/3173>`_)
* Development and test dependencies are now provided through dependency groups;
use ``python -m pip install -v -e . --group test`` from ``cuda_bindings_12``
for an editable installation with test dependencies.
(`PR #3315 <https://github.com/NVIDIA/cuda-python/pull/3315>`_)

Documentation and Maintenance
-----------------------------

* CUDA 12.9 maintenance sources now live in ``cuda_bindings_12`` on ``main``.
The historical ``12.9.x`` branch is frozen except for emergency fixes, which
must also be reconciled into ``main``. Main-tree CUDA 12 releases are built
and tested alongside CUDA 13. Old-layout CUDA 12 tags are released from a
validated workflow on the appropriate maintenance branch or historical tag,
not from ``main``.
* Corrected the CUDA 12 driver guidance to its existing minor-version
compatibility baseline: Linux 525.60.13 and Windows 528.33. This is a
documentation correction, not a new runtime version check. Newer features
and PTX compilation can still require a newer driver.

Known Issues
------------

* Updating from v12.6.2.post1 and older via ``pip install -U cuda-python`` might
not work. Uninstall with ``pip uninstall -y cuda-python``, then install the
CUDA 12 line with ``pip install "cuda-python~=12.9.0"``.
* ``nvml.system_get_process_name`` on WSL can return incorrect values. Set the
locale to ``"C"`` before calling ``nvml.device_get_compute_running_processes_v3``
(which sets the process names) and ``nvml.system_get_process_name``.
9 changes: 0 additions & 9 deletions cuda_bindings_12/docs/source/release/12.9.9-notes.rst
Original file line number Diff line number Diff line change
Expand Up @@ -345,15 +345,6 @@ Bugfixes
classes in ``cudla``, ``cufile``, and ``nvml``.
(`PR #2875 <https://github.com/NVIDIA/cuda-python/pull/2875>`_)

- ``driver.cuLaunchHostFunc`` and ``runtime.cudaLaunchHostFunc`` now pass the host
function and its user data to CUDA unchanged. They used to route the call through
an internal trampoline that freed its state after the first invocation, so a host
function captured into a CUDA graph crashed on the second launch of that graph.
(`Issue #790 <https://github.com/NVIDIA/cuda-python/issues/790>`_)
Callers must retain the callback object and user data storage for the lifetime
of the graph and its executables, including pending launches; see
:ref:`host-function-lifetime`.

Known issues
------------

Expand Down
31 changes: 10 additions & 21 deletions cuda_bindings_12/docs/source/release/12.X.Y-notes.rst
Original file line number Diff line number Diff line change
Expand Up @@ -13,31 +13,20 @@ Unreleased
Breaking Changes
----------------

* CUDA 12 bindings require Python 3.11 or newer. Python versions older than
3.11 are no longer supported; the supported range is Python 3.11-3.15.
Enhancements
------------

Bug Fixes
---------

* Nested resource pointer conversion accepts both ``str`` and ``bytes``
options, matching the CUDA 13 bindings behavior.
* Coredump settings retain borrowed string buffers for the duration of the
call and free the buffers allocated for getters.
* ``get_cuda_native_handle`` preserves a ``KeyError`` raised inside a
registered getter instead of reporting an unknown type.
Deprecations
------------

Packaging
---------

* Wheels omit ``.pyx`` implementation sources and ``.in`` generation templates,
following the CUDA 13 packaging policy. Cython declarations and includes
(``.pxd`` and ``.pxi``), and required headers, remain available for downstream
Cython builds. Source distributions retain the inputs needed to build the package.
Packaging and Source Builds
---------------------------

Documentation
-------------
Documentation and Maintenance
-----------------------------

* Corrected the CUDA 12 driver guidance to its existing minor-version
compatibility baseline: Linux 525.60.13 and Windows 528.33. This is a
documentation correction, not a new runtime version check. Newer features
and PTX compilation can still require a newer driver.
Known Issues
------------
Loading