Merge move_tensor_elementwise_impl_unary into move_tensor_elementwise_impl_unary_par_2

Author: vlad-perevezentsev

dpctl_ext/tensor/__init__.py

@@ -27,16 +27,16 @@
 # *****************************************************************************
 
 
-from dpctl.tensor._search_functions import where
-
-from dpctl_ext.tensor._copy_utils import (
+from ._accumulation import cumulative_logsumexp, cumulative_prod, cumulative_sum
+from ._clip import clip
+from ._copy_utils import (
     asnumpy,
     astype,
     copy,
     from_numpy,
     to_numpy,
 )
-from dpctl_ext.tensor._ctors import (
+from ._ctors import (
     arange,
     asarray,
     empty,
@@ -53,36 +53,6 @@
     zeros,
     zeros_like,
 )
-from dpctl_ext.tensor._indexing_functions import (
-    extract,
-    nonzero,
-    place,
-    put,
-    put_along_axis,
-    take,
-    take_along_axis,
-)
-from dpctl_ext.tensor._manipulation_functions import (
-    broadcast_arrays,
-    broadcast_to,
-    concat,
-    expand_dims,
-    flip,
-    moveaxis,
-    permute_dims,
-    repeat,
-    roll,
-    squeeze,
-    stack,
-    swapaxes,
-    tile,
-    unstack,
-)
-from dpctl_ext.tensor._reshape import reshape
-from dpctl_ext.tensor._utility_functions import all, any, diff
-
-from ._accumulation import cumulative_logsumexp, cumulative_prod, cumulative_sum
-from ._clip import clip
 from ._elementwise_funcs import (
     abs,
     acos,
@@ -112,6 +82,31 @@
     negative,
     positive,
 )
+from ._indexing_functions import (
+    extract,
+    nonzero,
+    place,
+    put,
+    put_along_axis,
+    take,
+    take_along_axis,
+)
+from ._manipulation_functions import (
+    broadcast_arrays,
+    broadcast_to,
+    concat,
+    expand_dims,
+    flip,
+    moveaxis,
+    permute_dims,
+    repeat,
+    roll,
+    squeeze,
+    stack,
+    swapaxes,
+    tile,
+    unstack,
+)
 from ._reduction import (
     argmax,
     argmin,
@@ -123,6 +118,8 @@
     reduce_hypot,
     sum,
 )
+from ._reshape import reshape
+from ._search_functions import where
 from ._searchsorted import searchsorted
 from ._set_functions import (
     isin,
@@ -133,6 +130,7 @@
 )
 from ._sorting import argsort, sort, top_k
 from ._type_utils import can_cast, finfo, iinfo, isdtype, result_type
+from ._utility_functions import all, any, diff
 
 __all__ = [
     "abs",

dpctl_ext/tensor/_accumulation.py

@@ -35,14 +35,14 @@
 import dpctl_ext.tensor as dpt_ext
 import dpctl_ext.tensor._tensor_accumulation_impl as tai
 import dpctl_ext.tensor._tensor_impl as ti
-from dpctl_ext.tensor._type_utils import (
+
+from ._numpy_helper import normalize_axis_index
+from ._type_utils import (
     _default_accumulation_dtype,
     _default_accumulation_dtype_fp_types,
     _to_device_supported_dtype,
 )
 
-from ._numpy_helper import normalize_axis_index
-
 
 def _accumulate_common(
     x,

dpctl_ext/tensor/_clip.py

@@ -35,21 +35,21 @@
 # when dpnp fully migrates dpctl/tensor
 import dpctl_ext.tensor as dpt_ext
 import dpctl_ext.tensor._tensor_impl as ti
-from dpctl_ext.tensor._copy_utils import (
+
+from ._copy_utils import (
     _empty_like_orderK,
     _empty_like_pair_orderK,
     _empty_like_triple_orderK,
 )
-from dpctl_ext.tensor._manipulation_functions import _broadcast_shape_impl
-from dpctl_ext.tensor._type_utils import _can_cast
-
+from ._manipulation_functions import _broadcast_shape_impl
 from ._scalar_utils import (
     _get_dtype,
     _get_queue_usm_type,
     _get_shape,
     _validate_dtype,
 )
 from ._type_utils import (
+    _can_cast,
     _resolve_one_strong_one_weak_types,
     _resolve_one_strong_two_weak_types,
 )

dpctl_ext/tensor/_copy_utils.py

@@ -42,9 +42,9 @@
 # when dpnp fully migrates dpctl/tensor
 import dpctl_ext.tensor as dpt_ext
 import dpctl_ext.tensor._tensor_impl as ti
-from dpctl_ext.tensor._type_utils import _dtype_supported_by_device_impl
 
 from ._numpy_helper import normalize_axis_index
+from ._type_utils import _dtype_supported_by_device_impl
 
 __doc__ = (
     "Implementation module for copy- and cast- operations on "
@@ -299,7 +299,7 @@ def _prepare_indices_arrays(inds, q, usm_type):
     inds = tuple(
         map(
             lambda ind: (
-                ind if ind.dtype == ind_dt else dpt.astype(ind, ind_dt)
+                ind if ind.dtype == ind_dt else dpt_ext.astype(ind, ind_dt)
             ),
             inds,
         )

dpctl_ext/tensor/_ctors.py

@@ -42,7 +42,8 @@
 # when dpnp fully migrates dpctl/tensor
 import dpctl_ext.tensor as dpt_ext
 import dpctl_ext.tensor._tensor_impl as ti
-from dpctl_ext.tensor._copy_utils import (
+
+from ._copy_utils import (
     _empty_like_orderK,
     _from_numpy_empty_like_orderK,
 )
@@ -1440,7 +1441,7 @@ def linspace(
     )
     _manager.add_event_pair(hev, la_ev)
 
-    return res if int_dt is None else dpt.astype(res, int_dt)
+    return res if int_dt is None else dpt_ext.astype(res, int_dt)
 
 
 def meshgrid(*arrays, indexing="xy"):

dpctl_ext/tensor/_indexing_functions.py

@@ -190,7 +190,7 @@ def place(arr, mask, vals):
     if vals.dtype == arr.dtype:
         rhs = vals
     else:
-        rhs = dpt.astype(vals, arr.dtype)
+        rhs = dpt_ext.astype(vals, arr.dtype)
     hev, pl_ev = ti._place(
         dst=arr,
         cumsum=cumsum,

dpctl_ext/tensor/_reshape.py

@@ -31,16 +31,17 @@
 import dpctl.tensor as dpt
 import dpctl.utils
 import numpy as np
-from dpctl.tensor._tensor_impl import (
-    _copy_usm_ndarray_for_reshape,
-    _ravel_multi_index,
-    _unravel_index,
-)
 
 # TODO: revert to `import dpctl.tensor...`
 # when dpnp fully migrates dpctl/tensor
 import dpctl_ext.tensor as dpt_ext
 
+from ._tensor_impl import (
+    _copy_usm_ndarray_for_reshape,
+    _ravel_multi_index,
+    _unravel_index,
+)
+
 __doc__ = "Implementation module for :func:`dpctl.tensor.reshape`."
 
 

dpctl_ext/tensor/_search_functions.py

@@ -34,9 +34,9 @@
 # when dpnp fully migrates dpctl/tensor
 import dpctl_ext.tensor as dpt_ext
 import dpctl_ext.tensor._tensor_impl as ti
-from dpctl_ext.tensor._manipulation_functions import _broadcast_shape_impl
 
 from ._copy_utils import _empty_like_orderK, _empty_like_triple_orderK
+from ._manipulation_functions import _broadcast_shape_impl
 from ._scalar_utils import (
     _get_dtype,
     _get_queue_usm_type,

dpctl_ext/tensor/libtensor/include/kernels/clip.hpp

@@ -35,10 +35,10 @@
 #pragma once
 #include <algorithm>
 #include <cmath>
-#include <complex>
 #include <cstddef>
 #include <cstdint>
 #include <type_traits>
+#include <vector>
 
 #include <sycl/sycl.hpp>
 

dpctl_ext/tensor/libtensor/include/kernels/constructors.hpp

@@ -33,9 +33,10 @@
 //===----------------------------------------------------------------------===//
 
 #pragma once
+
 #include <array>
-#include <complex>
 #include <cstddef>
+#include <type_traits>
 #include <vector>
 
 #include <sycl/sycl.hpp>
@@ -130,7 +131,7 @@ sycl::event lin_space_step_impl(sycl::queue &exec_q,
 }
 
 // Constructor to populate tensor with linear sequence defined by
-// start and and data
+// start and data
 
 template <typename Ty, typename wTy>
 class LinearSequenceAffineFunctor
@@ -191,7 +192,7 @@ class LinearSequenceAffineFunctor
  *
  * @param exec_q  Sycl queue to which kernel is submitted for execution.
  * @param nelems  Length of the sequence.
- * @param start_v Stating value of the sequence.
+ * @param start_v Starting value of the sequence.
  * @param end_v   End-value of the sequence.
  * @param include_endpoint  Whether the end-value is included in the sequence.
  * @param array_data Kernel accessible USM pointer to the start of array to be