Merge branch 'main' into ipc_suppressed_errors

Author: Andy-Jost

cuda_bindings/DESCRIPTION.rst

@@ -5,6 +5,10 @@
 cuda-bindings: Low-level CUDA interfaces
 ****************************************
 
+.. image:: https://img.shields.io/badge/NVIDIA-black?logo=nvidia
+   :target: https://www.nvidia.com/
+   :alt: NVIDIA
+
 `cuda.bindings <https://nvidia.github.io/cuda-python/cuda-bindings/>`_ is a standard set of low-level interfaces, providing full coverage of and 1:1 access to the CUDA host APIs from Python. Checkout the `Overview <https://nvidia.github.io/cuda-python/cuda-bindings/latest/overview.html>`_ for the workflow and performance results.
 
 * `Repository <https://github.com/NVIDIA/cuda-python/tree/main/cuda_bindings>`_

cuda_bindings/docs/source/install.rst

@@ -10,7 +10,7 @@ Runtime Requirements
 ``cuda.bindings`` supports the same platforms as CUDA. Runtime dependencies are:
 
 * Linux (x86-64, arm64) and Windows (x86-64)
-* Python 3.9 - 3.13
+* Python 3.9 - 3.14
 * Driver: Linux (580.65.06 or later) Windows (580.88 or later)
 * Optionally, NVRTC, nvJitLink, NVVM, and cuFile from CUDA Toolkit 13.x
 
@@ -20,6 +20,21 @@ Runtime Requirements
 
 Starting from v12.8.0, ``cuda-python`` becomes a meta package which currently depends only on ``cuda-bindings``; in the future more sub-packages will be added to ``cuda-python``. In the instructions below, we still use ``cuda-python`` as example to serve existing users, but everything is applicable to ``cuda-bindings`` as well.
 
+
+Free-threading Build Support
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+As of cuda-bindings 13.0.2 and 12.9.3, **experimental** packages for the `free-threaded interpreter`_ are shipped.
+
+1. Support for these builds is best effort, due to heavy use of `built-in
+   modules that are known to be thread-unsafe`_, such as ``ctypes``.
+2. For now, you are responsible for making sure that calls into the ``cuda-bindings``
+   library are thread-safe. This is subject to change.
+
+.. _built-in modules that are known to be thread-unsafe: https://github.com/python/cpython/issues/116738
+.. _free-threaded interpreter: https://docs.python.org/3/howto/free-threading-python.html
+
+
 Installing from PyPI
 --------------------
 

cuda_bindings/docs/source/release.rst

@@ -7,10 +7,10 @@ Release Notes
 .. toctree::
    :maxdepth: 3
 
-   13.X.Y <release/13.X.Y-notes.rst>
+   13.0.2 <release/13.0.2-notes.rst>
    13.0.1 <release/13.0.1-notes.rst>
    13.0.0 <release/13.0.0-notes.rst>
-   12.9.X <release/12.9.X-notes.rst>
+   12.9.3 <release/12.9.3-notes.rst>
    12.9.2 <release/12.9.2-notes.rst>
    12.9.1 <release/12.9.1-notes.rst>
    12.9.0 <release/12.9.0-notes.rst>

cuda_bindings/docs/source/release/12.9.3-notes.rst

@@ -3,20 +3,24 @@
 
 .. module:: cuda.bindings
 
-``cuda-bindings`` 12.9.X Release notes
+``cuda-bindings`` 12.9.3 Release notes
 ======================================
 
-Released on TBD
+Released on Oct 9, 2025
 
 
 Highlights
 ----------
 
+* This is the last release that officially supports Python 3.9.
+* Python 3.14 is supported.
+* **Experimental** free-threaded builds for Python 3.13/3.14 are made available. Any bugs can be reported to `our GitHub repo <https://github.com/NVIDIA/cuda-python>`_. More details are available in our :ref:`support` docs.
 * Automatic CUDA library path detection based on ``CUDA_HOME``, eliminating the need to manually set ``LIBRARY_PATH`` environment variables for installation.
 * The Python overhead of calling functions in CUDA bindings in ``driver``, ``runtime`` and ``nvrtc`` has been reduced by approximately 30%.
+* On Windows, the ``pywin32`` dependency has been removed. The necessary Windows API functions are now accessed directly.
 * Updated the ``cuda.bindings.runtime`` module to statically link against the CUDA Runtime library from CUDA Toolkit 12.9.1.
 * ``cyruntime.getLocalRuntimeVersion`` now uses pathfinder to find the CUDA runtime.
-* Experimental free-threaded builds are available on PyPI. More details are available in our :ref:`support` docs.
+
 
 Known issues
 ------------

cuda_bindings/docs/source/release/13.0.2-notes.rst

@@ -3,23 +3,25 @@
 
 .. module:: cuda.bindings
 
-``cuda-bindings`` 13.X.Y Release notes
+``cuda-bindings`` 13.0.2 Release notes
 ======================================
 
-Released on TBD
+Released on Oct 9, 2025
 
 
 Highlights
 ----------
 
+* This is the last release that officially supports Python 3.9.
+* Python 3.14 is supported.
+* **Experimental** free-threaded builds for Python 3.13/3.14 are made available. Any bugs can be reported to `our GitHub repo <https://github.com/NVIDIA/cuda-python>`_. More details are available in our :ref:`support` docs.
 * Migrated wheel dependencies from individual NVIDIA packages to the ``cuda-toolkit`` metapackage for improved dependency resolution and version constraints.
 * Automatic CUDA library path detection based on ``CUDA_HOME``, eliminating the need to manually set ``LIBRARY_PATH`` environment variables for installation.
 * The ``[all]`` optional dependencies now use ``cuda-toolkit`` with appropriate extras instead of individual packages. The NVCC compiler is no longer automatically installed with ``pip install cuda-python[all]`` as it was previously included only to access the NVVM library, which now has its own dedicated wheel. Users who need the NVCC compiler should explicitly install it with ``pip install cuda-toolkit[nvcc]==X.Y`` with the appropriate version for their needs.
 * The Python overhead of calling functions in CUDA bindings in ``driver``, ``runtime`` and ``nvrtc`` has been reduced by approximately 30%.
 * On Windows, the ``pywin32`` dependency has been removed. The necessary Windows API functions are now accessed directly.
 * Updated the ``cuda.bindings.runtime`` module to statically link against the CUDA Runtime library from CUDA Toolkit 13.0.1.
 * ``cyruntime.getLocalRuntimeVersion`` now uses pathfinder to find the CUDA runtime.
-* Experimental free-threaded builds are available on PyPI. More details are available in our :ref:`support` docs.
 
 
 Bug fixes

cuda_core/DESCRIPTION.rst

@@ -5,6 +5,10 @@
 cuda-core: Pythonic access to CUDA core functionalities
 *******************************************************
 
+.. image:: https://img.shields.io/badge/NVIDIA-black?logo=nvidia
+   :target: https://www.nvidia.com/
+   :alt: NVIDIA
+
 `cuda.core <https://nvidia.github.io/cuda-python/cuda-core/>`_ bridges Python's productivity with CUDA's performance through intuitive and pythonic APIs. The mission is to provide users full access to all of the core CUDA features in Python, such as runtime control, compiler and linker.
 
 * `Repository <https://github.com/NVIDIA/cuda-python/tree/main/cuda_core>`_

cuda_core/cuda/core/_version.py

@@ -2,4 +2,4 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 
-__version__ = "0.3.3a0"
+__version__ = "0.4.0"

cuda_core/cuda/core/experimental/_device.pyx

@@ -16,7 +16,7 @@ from cuda.core.experimental._context import Context, ContextOptions
 from cuda.core.experimental._event import Event, EventOptions
 from cuda.core.experimental._graph import GraphBuilder
 from cuda.core.experimental._memory import Buffer, DeviceMemoryResource, MemoryResource, _SynchronousMemoryResource
-from cuda.core.experimental._stream import IsStreamT, Stream, StreamOptions, default_stream
+from cuda.core.experimental._stream import IsStreamT, Stream, StreamOptions
 from cuda.core.experimental._utils.clear_error_support import assert_type
 from cuda.core.experimental._utils.cuda_utils import (
     ComputeCapability,
@@ -25,6 +25,7 @@ from cuda.core.experimental._utils.cuda_utils import (
     handle_return,
     runtime,
 )
+from cuda.core.experimental._stream cimport default_stream
 
 
 # TODO: I prefer to type these as "cdef object" and avoid accessing them from within Python,

cuda_core/cuda/core/experimental/_memory.pyx

@@ -12,6 +12,7 @@ from libc.string cimport memset, memcpy
 from cuda.bindings cimport cydriver
 
 from cuda.core.experimental._stream cimport Stream as cyStream
+from cuda.core.experimental._stream cimport default_stream
 from cuda.core.experimental._utils.cuda_utils cimport (
     _check_driver_error as raise_if_driver_error,
     check_or_create_options,
@@ -30,7 +31,7 @@ import platform
 import weakref
 
 from cuda.core.experimental._dlpack import DLDeviceType, make_py_capsule
-from cuda.core.experimental._stream import Stream, default_stream
+from cuda.core.experimental._stream import Stream
 from cuda.core.experimental._utils.cuda_utils import ( driver, Transaction, get_binding_version )
 
 if platform.system() == "Linux":

cuda_core/cuda/core/experimental/_stream.pxd

@@ -22,3 +22,6 @@ cdef class Stream:
     cpdef close(self)
     cdef int _get_context(self) except?-1 nogil
     cdef int _get_device_and_context(self) except?-1
+
+
+cdef Stream default_stream()