You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
.. SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2
+
.. SPDX-License-Identifier: Apache-2.0
3
+
4
+
.. currentmodule:: cuda.core
5
+
6
+
``cuda.core`` 0.6.0 Release Notes
7
+
==================================
8
+
9
+
New features
10
+
------------
11
+
12
+
- Added public access to default CUDA streams via module-level constants ``LEGACY_DEFAULT_STREAM`` and ``PER_THREAD_DEFAULT_STREAM``
13
+
14
+
Users can now access default streams directly from the ``cuda.core`` namespace:
15
+
16
+
.. code-block:: python
17
+
18
+
from cuda.core importLEGACY_DEFAULT_STREAM, PER_THREAD_DEFAULT_STREAM
19
+
20
+
# Use legacy default stream (synchronizes with all blocking streams)
21
+
LEGACY_DEFAULT_STREAM.sync()
22
+
23
+
# Use per-thread default stream (non-blocking, thread-local)
24
+
PER_THREAD_DEFAULT_STREAM.sync()
25
+
26
+
The legacy default stream synchronizes with all blocking streams in the same CUDA context, ensuring strict ordering but potentially limiting concurrency. The per-thread default stream is local to the calling thread and does not synchronize with other streams, enabling concurrent execution in multi-threaded applications.
27
+
28
+
This replaces the previous undocumented workaround of using ``Stream.from_handle(0)`` to access the legacy default stream.
0 commit comments