tvm.runtime.ndarray

Runtime NDArray API

class tvm.runtime.ndarray.NDArray

Lightweight NDArray class of TVM runtime.

Strictly this is only an Array Container (a buffer object) No arthimetic operations are defined. All operations are performed by TVM functions.

The goal is not to re-build yet another array library. Instead, this is a minimal data structure to demonstrate how can we use TVM in existing project which might have their own array containers.

property dtype: Type of this array

property device: Device of this array

same_as(other)

Check object identity equality

Parameters:: other (object) – The other object to compare to
Returns:: same – Whether other is same as self.
Return type:: bool

copyfrom(source_array)

Perform a synchronous copy from the array.

Parameters:: source_array (array_like) – The data source we should like to copy from.
Returns:: arr – Reference to self.
Return type:: NDArray

asnumpy(): Convert this array to numpy array. This API will be deprecated in TVM v0.8 release. Please use numpy instead.

numpy()

Convert this array to numpy array

Returns:: np_arr – The corresponding numpy array.
Return type:: numpy.ndarray

copyto(target, mem_scope=None)

Copy array to target

Parameters:

target (NDArray) – The target array to be copied, must have same shape as this array.
mem_scope (Optional[str]) – The memory scope of the array.

tvm.runtime.ndarray.device(dev_type, dev_id=0)

Construct a TVM device with given device type and id.

Parameters:

dev_type (int or str) – The device type mask or name of the device.
dev_id (int, optional) – The integer device id

Returns:

dev – The corresponding device.

Return type:

tvm.runtime.Device

Examples

Device can be used to create reflection of device by string representation of the device type.

assert tvm.device("cpu", 1) == tvm.cpu(1)
assert tvm.device("cuda", 0) == tvm.cuda(0)

tvm.runtime.ndarray.numpyasarray(np_data): Return a TVMArray representation of a numpy array.

tvm.runtime.ndarray.empty(shape, dtype='float32', device=cpu(0), mem_scope=None)

Create an empty array given shape and device

Parameters:

shape (Union[tvm.runtime.ShapeTuple, Sequence[SupportsInt]]) – The shape of the array.
dtype (type or str) – The data type of the array.
device (Device) – The device of the array.
mem_scope (Optional[str]) – The memory scope of the array.

Returns:

arr – The array tvm supported.

Return type:

tvm.nd.NDArray

tvm.runtime.ndarray.from_dlpack(dltensor)

Produces an array from an object with __dlpack__ method or a DLPack tensor w/o memory copy. Retreives the underlying DLPack tensor’s pointer to create an array from the data. Removes the original DLPack tensor’s destructor as now the array is responsible for destruction.

Parameters:: dltensor (object with __dlpack__ attribute or a DLPack capsule)
Returns:: arr – The array view of the tensor data.
Return type:: tvm.nd.NDArray

tvm.runtime.ndarray.cpu(dev_id=0)

Construct a CPU device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.cuda(dev_id=0)

Construct a CUDA GPU device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.gpu(dev_id=0)

Construct a CUDA GPU device

deprecated:: 0.9.0 Use tvm.cuda() instead.

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.rocm(dev_id=0)

Construct a ROCM device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.opencl(dev_id=0)

Construct a OpenCL device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.metal(dev_id=0)

Construct a metal device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.vpi(dev_id=0)

Construct a VPI simulated device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.vulkan(dev_id=0)

Construct a Vulkan device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.ext_dev(dev_id=0)

Construct a extension device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

Note

This API is reserved for quick testing of new device by plugin device API as ext_dev.

tvm.runtime.ndarray.hexagon(dev_id=0)

Construct a Hexagon device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.webgpu(dev_id=0)

Construct a webgpu device.

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.cl(dev_id=0)

Construct a OpenCL device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.mtl(dev_id=0)

Construct a metal device

Parameters:: dev_id (int, optional) – The integer device id
Returns:: dev – The created device
Return type:: Device

tvm.runtime.ndarray.array(arr, device=cpu(0), mem_scope=None)

Create an array from source arr.

Parameters:

arr (numpy.ndarray) – The array to be copied from
device (Device, optional) – The device to create the array
mem_scope (Optional[str]) – The memory scope of the array

Returns:

ret – The created array

Return type:

NDArray