stdgpu: stdgpu::mutex_array< Block, Allocator > — stdgpu Latest documentation (original) (raw)

mutex_array< Block, Allocator >#

stdgpu: stdgpu::mutex_array< Block, Allocator >

stdgpu Latest

Efficient STL-like Data Structures on the GPU

template<typename Block = mutex_default_type, typename Allocator = safe_device_allocator>
class stdgpu::mutex_array< Block, Allocator >

A class to model a mutex array on the GPU.

Template Parameters

Block	The internal bit block type
Allocator	The allocator type

Differences to std::mutex:

Mutexes must be modeled as containers since threads have to call the exact same object
Manual allocation and destruction of container required
No guaranteed valid state
Blocking lock is not supported

Public Member Functions
	mutex_array () noexcept=default

STDGPU_HOST_DEVICE bool	empty () const noexcept

STDGPU_HOST_DEVICE allocator_type	get_allocator () const noexcept

STDGPU_DEVICE_ONLY reference	operator[] (const index_t n)

STDGPU_DEVICE_ONLY const reference	operator[] (const index_t n) const

STDGPU_HOST_DEVICE index_t	size () const noexcept

bool	valid () const

template<typename ExecutionPolicy , STDGPU_DETAIL_OVERLOAD_IF(is_execution_policy_v< remove_cvref_t< ExecutionPolicy > >) >
bool	valid (ExecutionPolicy &&policy) const

Static Public Member Functions
static mutex_array	createDeviceObject (const index_t &size, const Allocator &allocator=Allocator())

template<typename ExecutionPolicy , STDGPU_DETAIL_OVERLOAD_IF(is_execution_policy_v< remove_cvref_t< ExecutionPolicy > >) >
static mutex_array	createDeviceObject (ExecutionPolicy &&policy, const index_t &size, const Allocator &allocator=Allocator())

template<typename ExecutionPolicy , STDGPU_DETAIL_OVERLOAD_IF(is_execution_policy_v< remove_cvref_t< ExecutionPolicy > >) >
static void	destroyDeviceObject (ExecutionPolicy &&policy, mutex_array &device_object)

static void	destroyDeviceObject (mutex_array &device_object)