cuda::stream_t Class Reference

Proxy class for a CUDA stream. More...

#include <stream.hpp>

Collaboration diagram for cuda::stream_t:

Classes
class	enqueue_t
	A gadget through which commands are enqueued on the stream. More...

Public Types
enum	: bool {   doesnt_synchronizes_with_default_stream = false,   does_synchronize_with_default_stream = true }

Public Member Functions
stream::handle_t	handle () const noexcept
	The raw CUDA handle for a stream which this class wraps.
context::handle_t	context_handle () const noexcept
	The raw CUDA handle for the context in which the represented stream is defined.
device::id_t	device_id () const noexcept
	The raw CUDA ID for the device w.r.t. which the stream is defined.
device_t	device () const noexcept
	The device w.r.t. which the stream is defined.
context_t	context () const noexcept
	The context in which this stream was defined.
bool	is_owning () const noexcept
	True if this wrapper is responsible for telling CUDA to destroy the stream upon the wrapper's own destruction.
bool	synchronizes_with_default_stream () const
	When true, work running in the created stream may run concurrently with work in stream 0 (the NULL stream), and there is no implicit synchronization performed between it and stream 0.
stream::priority_t	priority () const
bool	has_work_remaining () const
	Determines whether all work on this stream has been completed. More...
bool	is_clear () const
	The opposite of has_work_remaining() More...
bool	query () const
	An alias for is_clear() - to conform to how the CUDA runtime API names this functionality.
void	synchronize () const
	Block or busy-wait until all previously-scheduled work on this stream has been completed.
	stream_t (const stream_t &other)=delete
	stream_t (stream_t &&other) noexcept
stream_t &	operator= (const stream_t &other)=delete
stream_t &	operator= (stream_t &&other) noexcept

Public Attributes
const enqueue_t	enqueue { *this }
	This data member is a gadget for use instead of a "class-local" namespace; we do not need it as a distinct object.

Friends
stream_t	stream::wrap (device::id_t device_id, context::handle_t context_handle, stream::handle_t stream_handle, bool take_ownership, bool hold_pc_refcount_unit) noexcept
bool	operator== (const stream_t &lhs, const stream_t &rhs) noexcept

Detailed Description

Proxy class for a CUDA stream.

Note

a stream is specific to a context, and thus also specific to a device.

This class is a "reference type", not a "value type". Therefore, making changes to properties of the stream is a const-respecting operation on this class.

Member Function Documentation

has_work_remaining()

bool cuda::stream_t::has_work_remaining ( ) const

inline

Determines whether all work on this stream has been completed.

Note

having work is not the same as being busy executing that work!

Todo:

What if there are incomplete operations, but they're all waiting on something on another queue? Should the queue count as "busy" then?

Returns

true if there is still work pending, false otherwise

is_clear()

bool cuda::stream_t::is_clear ( ) const

inline

The opposite of has_work_remaining()

Returns

true if there is no work pending, false if all previously-scheduled work has been completed

priority()

stream::priority_t cuda::stream_t::priority ( ) const

inline

Returns

the execution priority of a tasks on this stream (relative to other tasks in other streams on the same device

Friends And Related Function Documentation

operator==

bool operator== ( const stream_t &  lhs, const stream_t &  rhs  )

friend

Note

two stream proxies may be equal even though one is the owning reference and the other isn't, or if only one holds a primary context reference unit and the other doesn't.

---

The documentation for this class was generated from the following file:

• src/cuda/api/stream.hpp