Handle dpnp functions and tests to run on CUDA devices (#2075)

This PR suggests updating some function implementations and tests to run
on CUDA devices
The PR includes:
1. Raise NotImplemetedError for unsupported functions
2. Create **skipped_tests_cuda.tbl** for unsupported functions
3. Skipped tests with a bug ticket for functions that should run on CUDA
4. Implement **is_cuda_backend()** and **is_cuda_device()** to determine
that the object has a CUDA backend and that the tests run on CUDA
---------

Co-authored-by: Vladislav Perevezentsev <[email protected]>

Author: npolina4

dpnp/dpnp_iface.py

@@ -68,6 +68,7 @@
     "get_result_array",
     "get_usm_ndarray",
     "get_usm_ndarray_or_scalar",
+    "is_cuda_backend",
     "is_supported_array_or_scalar",
     "is_supported_array_type",
     "synchronize_array_data",
@@ -681,6 +682,41 @@ def get_usm_ndarray_or_scalar(a):
     return a if dpnp.isscalar(a) else get_usm_ndarray(a)
 
 
+def is_cuda_backend(obj=None):
+    """
+    Checks that object has a CUDA backend.
+
+    Parameters
+    ----------
+    obj : {Device, SyclDevice, SyclQueue, dpnp.ndarray, usm_ndarray, None},
+          optional
+        An input object with sycl_device property to check device backend.
+        If `obj` is ``None``, device backend will be checked for the default
+        queue.
+        Default: ``None``.
+
+    Returns
+    -------
+    out : bool
+        Return ``True`` if data of the input object resides on a CUDA backend,
+        otherwise ``False``.
+
+    """
+
+    if obj is None:
+        sycl_device = dpctl.select_default_device()
+    elif isinstance(obj, dpctl.SyclDevice):
+        sycl_device = obj
+    else:
+        sycl_device = getattr(obj, "sycl_device", None)
+    if (
+        sycl_device is not None
+        and sycl_device.backend == dpctl.backend_type.cuda
+    ):
+        return True
+    return False
+
+
 def is_supported_array_or_scalar(a):
     """
     Return ``True`` if `a` is a scalar or an array of either

dpnp/dpnp_iface_indexing.py

@@ -128,6 +128,7 @@ def choose(x1, choices, out=None, mode="raise"):
     :obj:`dpnp.take_along_axis` : Preferable if choices is an array.
 
     """
+
     x1_desc = dpnp.get_dpnp_descriptor(x1, copy_when_nondefault_queue=False)
 
     choices_list = []
@@ -137,6 +138,11 @@ def choose(x1, choices, out=None, mode="raise"):
         )
 
     if x1_desc:
+        if dpnp.is_cuda_backend(x1_desc.get_array()):
+            raise NotImplementedError(
+                "Running on CUDA is currently not supported"
+            )
+
         if any(not desc for desc in choices_list):
             pass
         elif out is not None:

dpnp/dpnp_iface_libmath.py

@@ -82,6 +82,10 @@ def erf(in_array1):
         in_array1, copy_when_strides=False, copy_when_nondefault_queue=False
     )
     if x1_desc:
+        if dpnp.is_cuda_backend(x1_desc.get_array()):
+            raise NotImplementedError(
+                "Running on CUDA is currently not supported"
+            )
         return dpnp_erf(x1_desc).get_pyobj()
 
     result = create_output_descriptor_py(

dpnp/dpnp_iface_mathematical.py

@@ -2945,8 +2945,16 @@ def modf(x1, **kwargs):
     """
 
     x1_desc = dpnp.get_dpnp_descriptor(x1, copy_when_nondefault_queue=False)
-    if x1_desc and not kwargs:
-        return dpnp_modf(x1_desc)
+    if x1_desc:
+        if dpnp.is_cuda_backend(x1_desc.get_array()):
+            raise NotImplementedError(
+                "Running on CUDA is currently not supported"
+            )
+
+        if kwargs:
+            pass
+        else:
+            return dpnp_modf(x1_desc)
 
     return call_origin(numpy.modf, x1, **kwargs)
 

dpnp/dpnp_iface_sorting.py

@@ -192,6 +192,11 @@ def partition(x1, kth, axis=-1, kind="introselect", order=None):
 
     x1_desc = dpnp.get_dpnp_descriptor(x1, copy_when_nondefault_queue=False)
     if x1_desc:
+        if dpnp.is_cuda_backend(x1_desc.get_array()):
+            raise NotImplementedError(
+                "Running on CUDA is currently not supported"
+            )
+
         if not isinstance(kth, int):
             pass
         elif x1_desc.ndim == 0: