INetworkBackend companion interface with CUDA-specific functionality. More...

#include <nvneural/CudaTypes.h>

Inheritance diagram for nvneural::INetworkBackendCuda:

Public Member Functions
virtual ICudaMemoryAllocator *	getAllocator () const noexcept=0
	Returns the CUDA memory allocator interface.

virtual CUcontext	getCudaContext () const noexcept=0
	Retrieves the active CUDA context.

virtual CUdevice	getCudaDevice () const noexcept=0
	Retrieves the active CUDA device.

virtual CUstream	getCudaStream () const noexcept=0
	Retrieves the active CUDA stream. More...

virtual ICudaRuntimeCompiler *	runtimeCompiler () const noexcept=0
	Retrieves a runtime CUDA compiler.

virtual NeuralResult	setAllocator (ICudaMemoryAllocator *pAllocator) noexcept=0
	Sets a new CUDA allocator instead of the default. More...

Public Member Functions inherited from nvneural::IRefObject
virtual RefCount	addRef () const noexcept=0
	Increments the object's reference count. More...

virtual const void *	queryInterface (TypeId interface) const noexcept=0
	This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

virtual void *	queryInterface (TypeId interface) noexcept=0
	Retrieves a new object interface pointer. More...

virtual RefCount	release () const noexcept=0
	Decrements the object's reference count and destroy the object if the reference count reaches zero. More...

Static Public Attributes
static const IRefObject::TypeId	typeID = 0x61f19c57a3032f9ul
	Interface TypeId for interfaceOf purposes.

Static Public Attributes inherited from nvneural::IRefObject
static const TypeId	typeID = 0x14ecc3f9de638e1dul
	Interface TypeId for InterfaceOf purposes.

Additional Inherited Members
Public Types inherited from nvneural::IRefObject
using	RefCount = std::uint32_t
	Typedef used to track the number of active references to an object.

using	TypeId = std::uint64_t
	Every interface must define a unique TypeId. This should be randomized.

Protected Member Functions inherited from nvneural::IRefObject
virtual	~IRefObject ()=default
	A protected destructor prevents accidental stack-allocation of IRefObjects or use with other smart pointer classes like std::unique_ptr.

Detailed Description

INetworkBackend companion interface with CUDA-specific functionality.

The CudaBackend class (createable with ClassRegistry and the NVNEURAL_INETWORKBACKENDCUDA_OBJECTCLASS object class) provides INetworkBackend for common operations and INetworkBackendCuda for things that require CUDA-specific details.

By convention this interface uses CUDA driver API types (cuThing) rather than runtime API types (cudaThing), but most of these types are trivially convertible between the two. See the CUDA documentation for details.

Member Function Documentation

◆ getCudaStream()

virtual CUstream nvneural::INetworkBackendCuda::getCudaStream ( ) const

pure virtualnoexcept

Retrieves the active CUDA stream.

Reminder: CUstream and cudaStream_t are equivalent types. CUDA runtime code can just cast the CUstream accordingly.

◆ setAllocator()

virtual NeuralResult nvneural::INetworkBackendCuda::setAllocator ( ICudaMemoryAllocator * pAllocator )

pure virtualnoexcept

Sets a new CUDA allocator instead of the default.

Warning: This invalidates ALL memory handles. Do not call this once device memory has been allocated.

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

Core/Inc/nvneural/CudaTypes.h

Public Member Functions

Static Public Attributes

Additional Inherited Members

Detailed Description

Member Function Documentation

◆ getCudaStream()

◆ setAllocator()