|
NVIDIA DeepStream SDK API Reference
|
8.0 Release
|
Go to the documentation of this file.
32 #ifndef DEEPSTREAM_TENSOR_HPP
33 #define DEEPSTREAM_TENSOR_HPP
42 class TensorImplementation;
85 const int64_t
shape[],
86 const int64_t strides[],
89 unsigned int device_id,
104 unsigned int rank()
const;
119 unsigned int bits()
const;
124 uint64_t
stride(
unsigned int d)
const;
159 uint64_t
size()
const;
Buffer wrap(NvBufSurfaceColorFormat format)
Wrap the tensor to a buffer.
uint64_t size() const
total size of the tensor in bytes
Tensor(unsigned int rank, DataType dtype, unsigned int bits, const int64_t shape[], const int64_t strides[], void *data, std::string format, unsigned int device_id, DeviceType device, Context *context)
Construct from common tensor configurations.
void * data() const
Retrieve the allocated data pointer.
TensorImplementation * impl_
DataType
Code for tensor data types.
std::vector< uint64_t > TensorShape
std::unique_ptr< Context > context_
DeviceType deviceType() const
Retrieve the device type.
unsigned int rank() const
Number of dimenstions of the tensor.
Tensor * toGPU(unsigned int device_id) const
Copy the tensor to GPU.
Tensor * clone() const
Clone a tensor.
virtual ~Tensor()
Destructor.
TensorShape shape() const
Shape of the tensor in vector.
uint64_t stride(unsigned int d) const
Retrieve the stride of a specific dimension.
unsigned int deviceId() const
Retrieve the device id.
Context object of the tensor, which associates the tensor with what creates it.
NvBufSurfaceColorFormat
Defines color formats for NvBufSurface.
unsigned int bits() const
Number of bits for each data.
static std::string format(const char *fmt,...)
DataType dtype() const
Data type of the tensor elements.