tvm::runtime::cuda_ipc::CUDAIPCMemory Class Reference

Managed reference to CUDAIPCMemoryObj. More...

#include <cuda_ipc_memory.h>

Inheritance diagram for tvm::runtime::cuda_ipc::CUDAIPCMemory:

Collaboration diagram for tvm::runtime::cuda_ipc::CUDAIPCMemory:

Public Member Functions
	CUDAIPCMemory (ffi::UnsafeInit tag, ffi::ObjectPtr< CUDAIPCMemoryObj > n)
	TVM_FFI_DEFINE_OBJECT_REF_METHODS_NOTNULLABLE (CUDAIPCMemory, ffi::ObjectRef, CUDAIPCMemoryObj)

Static Public Member Functions
static TVM_RUNTIME_DLL TensorMemoryPool *	GlobalMemoryPool ()
	Get the global singleton CUDAIPCMemory pool.
static TVM_RUNTIME_DLL CUDAIPCMemory	GetIPCMemoryFromDevicePtr (void *ptr)
	Given a local CUDA data pointer, return the CUDAIPCMemory object of the pointer.

Detailed Description

Managed reference to CUDAIPCMemoryObj.

See also

CUDAIPCMemory

Constructor & Destructor Documentation

CUDAIPCMemory()

tvm::runtime::cuda_ipc::CUDAIPCMemory::CUDAIPCMemory ( ffi::UnsafeInit  tag, ffi::ObjectPtr< CUDAIPCMemoryObj >  n  )

inlineexplicit

Member Function Documentation

GetIPCMemoryFromDevicePtr()

static TVM_RUNTIME_DLL CUDAIPCMemory tvm::runtime::cuda_ipc::CUDAIPCMemory::GetIPCMemoryFromDevicePtr ( void *  ptr )

static

Given a local CUDA data pointer, return the CUDAIPCMemory object of the pointer.

Note

The pointer's CUDAIPCMemory is expected to have been allocated through global function "cuda_ipc.alloc_storage". Or otherwise this function will raise exception.

GlobalMemoryPool()

static TVM_RUNTIME_DLL TensorMemoryPool * tvm::runtime::cuda_ipc::CUDAIPCMemory::GlobalMemoryPool ( )

static

Get the global singleton CUDAIPCMemory pool.

TVM_FFI_DEFINE_OBJECT_REF_METHODS_NOTNULLABLE()

tvm::runtime::cuda_ipc::CUDAIPCMemory::TVM_FFI_DEFINE_OBJECT_REF_METHODS_NOTNULLABLE	(	CUDAIPCMemory	,
		ffi::ObjectRef	,
		CUDAIPCMemoryObj
	)

---

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

• include/tvm/runtime/disco/cuda_ipc_memory.h