..
    MIT License

    Copyright (c) 2024-2026 Advanced Micro Devices, Inc.

    Permission is hereby granted, free of charge, to any person obtaining a copy
    of this software and associated documentation files (the "Software"), to deal
    in the Software without restriction, including without limitation the rights
    to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
    copies of the Software, and to permit persons to whom the Software is
    furnished to do so, subject to the following conditions:

    The above copyright notice and this permission notice shall be included in all
    copies or substantial portions of the Software.

    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
    IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
    AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
    OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
    SOFTWARE.

.. meta::
  :description: Overview of the libhipcxx synchronization primitives extended API, including system-wide atomics and atomic references for HIP.
  :keywords: libhipcxx, ROCm, HIP, C++, synchronization, atomic, atomic_ref


.. _libcudacxx-extended-api-synchronization:

Synchronization Primitives
===========================

This page covers the synchronization primitives extended API, providing thread-scoped atomics for coordinating host and device threads.

``cuda::latch``, ``cuda::barrier``, ``cuda::counting_semaphore``, ``cuda::binary_semaphore``, and
``cuda::pipeline`` are not supported in libhipcxx. See :ref:`libhipcxx-limitations`.

.. toctree::
   :hidden:
   :maxdepth: 1

   synchronization_primitives/atomic
   synchronization_primitives/atomic_ref

..
   Not supported in libhipcxx, see reference/libhipcxx-limitations.rst. The pages are
   also listed in exclude_patterns in docs/conf.py.

   synchronization_primitives/latch
   synchronization_primitives/barrier
   synchronization_primitives/counting_semaphore
   synchronization_primitives/binary_semaphore
   synchronization_primitives/pipeline

.. rubric:: Atomics

.. list-table::
   :widths: 25 45 30
   :header-rows: 1

   * - API
     - Description
     - Since

   * - :ref:`cuda::atomic <libcudacxx-extended-api-synchronization-atomic>`
     - System-wide `std::atomic <https://en.cppreference.com/w/cpp/atomic/atomic>`_ objects and operations
     - libhipcxx 2.7
   * - :ref:`cuda::atomic_ref <libcudacxx-extended-api-synchronization-atomic-ref>`
     - System-wide `std::atomic_ref <https://en.cppreference.com/w/cpp/atomic/atomic_ref>`_ objects and operations
     - libhipcxx 2.7

..
   Not supported in libhipcxx, see reference/libhipcxx-limitations.rst.

   .. rubric:: Latches

   .. list-table::
      :widths: 25 45 30
      :header-rows: 0

      * - :ref:`cuda::latch <libcudacxx-extended-api-synchronization-latch>`
        - System-wide `std::latch <https://en.cppreference.com/w/cpp/thread/latch>`_ single-phase asynchronous
          thread coordination mechanism
        - libhipcxx 2.7

   .. rubric:: Barriers

   .. list-table::
      :widths: 25 45 30
      :header-rows: 0

      * - :ref:`cuda::barrier <libcudacxx-extended-api-synchronization-barrier>`
        - System wide `std::barrier <https://en.cppreference.com/w/cpp/thread/barrier>`_ multi-phase asynchronous
          thread coordination mechanism
        - libhipcxx 2.7

   .. rubric:: Semaphores

   .. list-table::
      :widths: 25 45 30
      :header-rows: 0

      * - :ref:`cuda::counting_semaphore <libcudacxx-extended-api-synchronization-counting-semaphore>`
        - System wide `std::counting_semaphore <https://en.cppreference.com/w/cpp/thread/counting_semaphore>`_
          primitive for constraining concurrent access
        - libhipcxx 2.7
      * - :ref:`cuda::binary_semaphore <libcudacxx-extended-api-synchronization-counting-semaphore>`
        - System wide `std::binary_semaphore <https://en.cppreference.com/w/cpp/thread/counting_semaphore>`_
          primitive for mutual exclusion
        - libhipcxx 2.7

   .. rubric:: Pipelines

   The pipeline library is included in ROCm, but is not part of the open source libhipcxx distribution.

   .. list-table::
      :widths: 25 45 30
      :header-rows: 0

      * - :ref:`cuda::pipeline <libcudacxx-extended-api-synchronization-pipeline>`
        - Coordination mechanism for sequencing asynchronous operations
        - libhipcxx 2.7
      * - :ref:`cuda::pipeline_shared_state <libcudacxx-extended-api-synchronization-pipeline-pipeline-shared-state>`
        - :ref:`cuda::pipeline <libcudacxx-extended-api-synchronization-pipeline>` shared state object
        - libhipcxx 2.7
      * - :ref:`cuda::pipeline_role <libcudacxx-extended-api-synchronization-pipeline-pipeline-role>`
        - Defines producer/consumer role for a thread participating in a *pipeline*
        - libhipcxx 2.7
      * - :ref:`cuda::make_pipeline <libcudacxx-extended-api-synchronization-pipeline-pipeline-role>`
        - Creates a :ref:`cuda::pipeline <libcudacxx-extended-api-synchronization-pipeline>`
        - libhipcxx 2.7
      * - :ref:`cuda::pipeline_consumer_wait_prior <libcudacxx-extended-api-synchronization-pipeline-pipeline-consumer-wait-prior>`
        - Blocks the current thread until all operations committed up to a prior *pipeline stage* complete
        - libhipcxx 2.7
      * - :ref:`cuda::pipeline_producer_commit <libcudacxx-extended-api-synchronization-pipeline-pipeline-producer-commit>`
        - Binds operations previously issued by the current thread to a :ref:`cuda::barrier <libcudacxx-extended-api-synchronization-barrier>`
        - libhipcxx 2.7