org.bytedeco.tvm

Class NDArray

java.lang.Object

org.bytedeco.javacpp.Pointer

org.bytedeco.tvm.ObjectRef

org.bytedeco.tvm.NDArray

All Implemented Interfaces:

AutoCloseable

@Namespace(value="tvm::runtime")
 @Properties(inherit=tvm_runtime.class)
public class NDArray
extends ObjectRef

\brief Managed NDArray. The array is backed by reference counted blocks.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	NDArray.Container \brief NDArray internal container type
static class	NDArray.ContainerBase \brief ContainerBase used to back the TVMArrayHandle
static class	NDArray.Internal

Nested classes/interfaces inherited from class org.bytedeco.javacpp.Pointer

Pointer.CustomDeallocator, Pointer.Deallocator, Pointer.NativeDeallocator, Pointer.ReferenceCounter

Field Summary

Fields inherited from class org.bytedeco.tvm.ObjectRef

_type_is_nullable

Fields inherited from class org.bytedeco.javacpp.Pointer

address, capacity, limit, position

Constructor Summary

Constructors

Constructor and Description
NDArray() \brief default constructor
NDArray(long size) Native array allocator.
NDArray(ObjectPtr data) \brief constructor.
NDArray(Pointer p) Pointer cast constructor.

Method Summary

Modifier and Type	Method and Description
static boolean	AbilityOfZeroCopyForDLTensor(DLTensor tensor, DLDevice dev) \brief Check conditions for construction NDArray over DLTensor without copying.
DLTensor	accessDLTensor()
void	CopyFrom(DLTensor other) \brief Copy data content from another array.
void	CopyFrom(NDArray other)
void	CopyFromBytes(Pointer data, long nbytes) \brief Copy data content from a byte buffer.
static void	CopyFromTo(DLTensor from, DLTensor to)
static void	CopyFromTo(DLTensor from, DLTensor to, TVMStreamHandle stream) \brief Function to copy data from one array to another.
NDArray	CopyTo(DLDevice dev)
NDArray	CopyTo(DLDevice dev, TVMStringOptional mem_scope) \brief Copy the data to another device.
void	CopyTo(DLTensor other) \brief Copy data content into another array.
void	CopyTo(NDArray other)
void	CopyToBytes(Pointer data, long nbytes) \brief Copy data content into another array.
NDArray	CreateView(ShapeTuple shape, DLDataType dtype)
NDArray	CreateView(ShapeTuple shape, DLDataType dtype, long relative_byte_offset) \brief Create a NDArray that shares the data memory with the current one.
DataType	DataType()
static NDArray	Empty(ShapeTuple shape, DLDataType dtype, DLDevice dev)
static NDArray	Empty(ShapeTuple shape, DLDataType dtype, DLDevice dev, TVMStringOptional mem_scope) \brief Create an empty NDArray.
static NDArray	FromDLPack(DLManagedTensor tensor) \brief Create a NDArray backed by a dlpack tensor.
static NDArray	FromExternalDLTensor(DLTensor dl_tensor) \brief Create a NDArray backed by an external DLTensor without memory copying.
NDArray	getPointer(long i)
boolean	IsContiguous()
boolean	Load(Stream stream) \brief Load NDArray from stream
static NDArray	NewFromDLTensor(DLTensor dl_tensor, DLDevice dev) \brief Create new NDArray, data is copied from DLTensor.
NDArray	position(long position)
void	Save(Stream stream) \brief Save NDArray to stream
ShapeTuple	Shape()
DLManagedTensor	ToDLPack() \brief Create a reference view of NDArray that represents as DLManagedTensor.
int	use_count()

Methods inherited from class org.bytedeco.tvm.ObjectRef

_type_is_nullable, access, defined, equals, get, lessThan, notEquals, same_as, unique

Methods inherited from class org.bytedeco.javacpp.Pointer

address, asBuffer, asByteBuffer, availablePhysicalBytes, calloc, capacity, capacity, close, deallocate, deallocate, deallocateReferences, deallocator, deallocator, equals, fill, formatBytes, free, getDirectBufferAddress, getPointer, getPointer, getPointer, hashCode, interruptDeallocatorThread, isNull, isNull, limit, limit, malloc, maxBytes, maxPhysicalBytes, memchr, memcmp, memcpy, memmove, memset, offsetAddress, offsetof, offsetof, parseBytes, physicalBytes, physicalBytesInaccurate, position, put, realloc, referenceCount, releaseReference, retainReference, setNull, sizeof, sizeof, toString, totalBytes, totalCount, totalPhysicalBytes, withDeallocator, zero

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Constructor Detail

NDArray

public NDArray(Pointer p)

Pointer cast constructor. Invokes Pointer(Pointer).

NDArray

public NDArray(long size)

Native array allocator. Access with Pointer.position(long).

NDArray

public NDArray()

\brief default constructor

NDArray

public NDArray(@ByVal
               ObjectPtr data)

\brief constructor.

Parameters:

data - ObjectPtr to the data container.

Method Detail

position

public NDArray position(long position)

Overrides:

position in class ObjectRef

getPointer

public NDArray getPointer(long i)

Overrides:

getPointer in class ObjectRef

use_count

public int use_count()

Overrides:

use_count in class ObjectRef

Returns:

the reference counter \note this number is approximate in multi-threaded setting.

accessDLTensor

@Const
 @Name(value="operator ->")
public DLTensor accessDLTensor()

Returns:

Pointer to content of DLTensor

IsContiguous

@Cast(value="bool")
public boolean IsContiguous()

Returns:

Whether the tensor is contiguous

CopyFrom

public void CopyFrom(@Const
                     DLTensor other)

\brief Copy data content from another array.

Parameters:

other - The source array to be copied from. \note The copy may happen asynchronously if it involves a GPU context. TVMSynchronize is necessary.

CopyFrom

public void CopyFrom(@Const @ByRef
                     NDArray other)

CopyFromBytes

public void CopyFromBytes(@Const
                          Pointer data,
                          @Cast(value="size_t")
                          long nbytes)

\brief Copy data content from a byte buffer.

Parameters:

data - The source bytes to be copied from.

nbytes - The size of the buffer in bytes Must be equal to the size of the NDArray. \note The copy always triggers a TVMSynchronize.

CopyTo

public void CopyTo(DLTensor other)

\brief Copy data content into another array.

Parameters:

other - The source array to be copied from. \note The copy may happen asynchronously if it involves a GPU context. TVMSynchronize is necessary.

CopyTo

public void CopyTo(@Const @ByRef
                   NDArray other)

CopyToBytes

public void CopyToBytes(Pointer data,
                        @Cast(value="size_t")
                        long nbytes)

\brief Copy data content into another array.

Parameters:

data - The source bytes to be copied from.

nbytes - The size of the data buffer. Must be equal to the size of the NDArray. \note The copy always triggers a TVMSynchronize.

CopyTo

@ByVal
public NDArray CopyTo(@Cast(value="const tvm::Device*") @ByRef
                              DLDevice dev,
                              @ByVal(nullValue="tvm::runtime::Optional<tvm::runtime::String>(tvm::NullOpt)")
                              TVMStringOptional mem_scope)

\brief Copy the data to another device.

Parameters:

dev - The target device.

mem_scope - The memory scope of the target array.

Returns:

The array under another device. \note The copy always triggers a TVMSynchronize.

CopyTo

@ByVal
public NDArray CopyTo(@Cast(value="const tvm::Device*") @ByRef
                              DLDevice dev)

Load

@Cast(value="bool")
public boolean Load(Stream stream)

\brief Load NDArray from stream

Parameters:

stream - The input data stream

Returns:

Whether load is successful

Save

public void Save(Stream stream)

\brief Save NDArray to stream

Parameters:

stream - The output data stream

CreateView

@ByVal
public NDArray CreateView(@ByVal
                                  ShapeTuple shape,
                                  @ByVal
                                  DLDataType dtype,
                                  @Cast(value="uint64_t")
                                  long relative_byte_offset)

\brief Create a NDArray that shares the data memory with the current one.

Parameters:

shape - The shape of the new array.

dtype - The data type of the new array.

relative_byte_offset - The offset of the output NDArray, relative to the current byte offset. By default, the offset of the view is the same as the offset of the current array. \note The new array must not allow access of addresses which would be out of bounds in the current array. If the new array is larger than the current array, or if the relative_byte_offset would place the end of the new array outside the bounds of the current array, this function will raise an exception.

CreateView

@ByVal
public NDArray CreateView(@ByVal
                                  ShapeTuple shape,
                                  @ByVal
                                  DLDataType dtype)

ToDLPack

public DLManagedTensor ToDLPack()

\brief Create a reference view of NDArray that represents as DLManagedTensor.

Returns:

A DLManagedTensor

Empty

@ByVal
public static NDArray Empty(@ByVal
                                    ShapeTuple shape,
                                    @ByVal
                                    DLDataType dtype,
                                    @ByVal @Cast(value="tvm::Device*")
                                    DLDevice dev,
                                    @ByVal(nullValue="tvm::runtime::Optional<tvm::runtime::String>(tvm::NullOpt)")
                                    TVMStringOptional mem_scope)

\brief Create an empty NDArray.

Parameters:

shape - The shape of the new array.

dtype - The data type of the new array.

dev - The device of the array.

mem_scope - The memory scope of the array.

Returns:

The created Array

Empty

@ByVal
public static NDArray Empty(@ByVal
                                    ShapeTuple shape,
                                    @ByVal
                                    DLDataType dtype,
                                    @ByVal @Cast(value="tvm::Device*")
                                    DLDevice dev)

FromExternalDLTensor

@ByVal
public static NDArray FromExternalDLTensor(@Const @ByRef
                                                   DLTensor dl_tensor)

\brief Create a NDArray backed by an external DLTensor without memory copying. If DLTensor is not contiguous or has bad aligned data, It fails. This allows us to create a NDArray using the memory allocated by an external source. Responsibility for memory retaining lies with the external source.

Parameters:

dl_tensor - The DLTensor for NDArray base.

Returns:

The created NDArray view.

NewFromDLTensor

@ByVal
public static NDArray NewFromDLTensor(DLTensor dl_tensor,
                                              @Cast(value="const tvm::Device*") @ByRef
                                              DLDevice dev)

\brief Create new NDArray, data is copied from DLTensor.

Parameters:

dl_tensor - The DLTensor to copy from.

dev - device location of the created NDArray.

Returns:

The created NDArray view.

FromDLPack

@ByVal
public static NDArray FromDLPack(DLManagedTensor tensor)

\brief Create a NDArray backed by a dlpack tensor. This allows us to create a NDArray using the memory allocated by an external deep learning framework that is DLPack compatible. The memory is retained until the NDArray went out of scope.

Parameters:

tensor - The DLPack tensor to copy from.

Returns:

The created NDArray view.

CopyFromTo

public static void CopyFromTo(@Const
                              DLTensor from,
                              DLTensor to,
                              TVMStreamHandle stream)

\brief Function to copy data from one array to another.

Parameters:

from - The source array.

to - The target array.

stream - The stream used in copy.

CopyFromTo

public static void CopyFromTo(@Const
                              DLTensor from,
                              DLTensor to)

Shape

@ByVal
public ShapeTuple Shape()

DataType

@ByVal
public DataType DataType()

AbilityOfZeroCopyForDLTensor

@Cast(value="bool")
public static boolean AbilityOfZeroCopyForDLTensor(DLTensor tensor,
                                                                        @Cast(value="const tvm::Device*") @ByRef
                                                                        DLDevice dev)

\brief Check conditions for construction NDArray over DLTensor without copying. There are three conditions to check: 1. Destination device is the same as DLTensor device 2. Destination device id is the same as DLTensor device id 3. Memory in DLTensor is aligned as expected for NDArray

Parameters:

tensor - the DLTensor.

dev - destination device.

Returns:

true if all conditions are satisfied.