Extended API

Fundamentals

Thread Scopes	Defines the kind of threads that can synchronize using a primitive. `(enum)` 1.0.0 / CUDA 10.2
Thread Groups	Concepts for groups of cooperating threads. `(concept)` 1.2.0 / CUDA 11.1

Shapes

`cuda::std::size_t`	Defines an extent of bytes. `(typedef)` 1.0.0 / CUDA 10.2
`cuda::aligned_size_t`	Defines an extent of bytes with a statically defined alignment. `(class template)` 1.2.0 / CUDA 11.1

Synchronization Primitives

Atomics

cuda::atomic System-wide cuda::std::atomic objects and operations. (class template)

1.0.0 / CUDA 10.2

Latches

cuda::latch System-wide cuda::std::latch single-phase asynchronous thread coordination mechanism. (class template)

1.1.0 / CUDA 11.0

Barriers

cuda::barrier System-wide cuda::std::barrier multi-phase asynchronous thread coordination mechanism. (class template)

1.1.0 / CUDA 11.0

Semaphores

`cuda::counting_semaphore`	System-wide `cuda::std::counting_semaphore` primitive for constraining concurrent access. `(class template)` 1.1.0 / CUDA 11.0
`cuda::binary_semaphore`	System-wide `cuda::std::binary_semaphore` primitive for mutual exclusion. `(class template)` 1.1.0 / CUDA 11.0

Pipelines

The pipeline library is included in the CUDA Toolkit, but is not part of the open source libcu++ distribution.

`cuda::pipeline`	Coordination mechanism for sequencing asynchronous operations. `(class template)` CUDA 11.1
`cuda::pipeline_shared_state`	`cuda::pipeline` shared state object. `(class template)` CUDA 11.1
`cuda::pipeline_role`	Defines producer/consumer role for a thread participating in a pipeline. `(enum)` CUDA 11.1
`cuda::make_pipeline`	Creates a `cuda::pipeline`. `(function template)` CUDA 11.1
`cuda::pipeline_consumer_wait_prior`	Blocks the current thread until all operations committed up to a prior pipeline stage complete. `(function template)` CUDA 11.1
`cuda::pipeline_producer_commit`	Binds operations previously issued by the current thread to a `cuda::barrier`. `(function template)` CUDA 11.1

Asynchronous Operations

cuda::memcpy_async Asynchronously copies one range to another. (function template)

1.1.0 / CUDA 11.0
1.2.0 / CUDA 11.1 (group & aligned overloads)