Add call_kernel for the old kernel

numba_dpex/__init__.py

@@ -17,6 +17,7 @@
 from numba import __version__ as numba_version
 from numba.np.ufunc.decorators import Vectorize
 
+from numba_dpex.core.kernel_interface.launcher import call_kernel
 from numba_dpex.vectorizers import Vectorize as DpexVectorize
 
 from .numba_patches import patch_arrayexpr_tree_to_ir, patch_is_ufunc
@@ -145,4 +146,4 @@ def parse_sem_version(version_string: str) -> Tuple[int, int, int]:
 __version__ = get_versions()["version"]
 del get_versions
 
-__all__ = types.__all__ + ["Range", "NdRange"]
+__all__ = types.__all__ + ["Range", "NdRange", "call_kernel"]

numba_dpex/core/kernel_interface/launcher.py

@@ -0,0 +1,18 @@
+"""Launcher package to provide the same way of calling kernel as experimental
+one."""
+
+
+def call_kernel(kernel_fn, index_space, *kernel_args) -> None:
+    """Syntax sugar for calling kernel the same way as experimental one.
+    It is a temporary glue for the experimental kernel migration.
+
+    Args:
+        kernel_fn (numba_dpex.experimental.KernelDispatcher): A
+        numba_dpex.kernel decorated function that is compiled to a
+        KernelDispatcher by numba_dpex.
+        index_space (Range | NdRange): A numba_dpex.Range or numba_dpex.NdRange
+        type object that specifies the index space for the kernel.
+        kernel_args : List of objects that are passed to the numba_dpex.kernel
+        decorated function.
+    """
+    kernel_fn[index_space](*kernel_args)

numba_dpex/tests/kernel_tests/test_atomic_op.py

@@ -56,7 +56,7 @@ def f(a):
 def test_kernel_atomic_simple(input_arrays, kernel_result_pair):
     a, dtype = input_arrays()
     kernel, expected = kernel_result_pair
-    kernel[dpex.Range(global_size)](a)
+    dpex.call_kernel(kernel, dpex.Range(global_size), a)
     assert a[0] == expected
 
 
@@ -96,7 +96,7 @@ def test_kernel_atomic_local(input_arrays, return_list_of_op):
     op_type, expected = return_list_of_op
     f = get_func_local(op_type, dtype)
     kernel = dpex.kernel(f)
-    kernel[dpex.NdRange(dpex.Range(N), dpex.Range(N))](a)
+    dpex.call_kernel(kernel, dpex.NdRange(dpex.Range(N), dpex.Range(N)), a)
     assert a[0] == expected
 
 
@@ -134,7 +134,7 @@ def test_kernel_atomic_multi_dim(
     dim = return_list_of_dim
     kernel = get_kernel_multi_dim(op_type, len(dim))
     a = np.zeros(dim, dtype=return_dtype)
-    kernel[dpex.Range(global_size)](a)
+    dpex.call_kernel(kernel, dpex.Range(global_size), a)
     assert a[0] == expected
 
 

numba_dpex/tests/kernel_tests/test_barrier.py

@@ -25,7 +25,7 @@ def twice(A):
     orig = dpt.asnumpy(arr)
     global_size = (N,)
     local_size = (N // 2,)
-    twice[NdRange(global_size, local_size)](arr)
+    dpex.call_kernel(twice, NdRange(global_size, local_size), arr)
     after = dpt.asnumpy(arr)
     # The computation is correct?
     np.testing.assert_allclose(orig * 2, after)
@@ -43,7 +43,7 @@ def twice(A):
     N = 256
     arr = dpt.arange(N, dtype=dpt.float32)
     orig = dpt.asnumpy(arr)
-    twice[Range(N)](arr)
+    dpex.call_kernel(twice, Range(N), arr)
     after = dpt.asnumpy(arr)
     # The computation is correct?
     np.testing.assert_allclose(orig * 2, after)
@@ -66,7 +66,9 @@ def reverse_array(A):
 
     arr = dpt.arange(blocksize, dtype=dpt.float32)
     orig = dpt.asnumpy(arr)
-    reverse_array[NdRange(Range(blocksize), Range(blocksize))](arr)
+    dpex.call_kernel(
+        reverse_array, NdRange(Range(blocksize), Range(blocksize)), arr
+    )
     after = dpt.asnumpy(arr)
     expected = orig[::-1] + orig
     np.testing.assert_allclose(expected, after)

numba_dpex/tests/kernel_tests/test_complex_array.py

@@ -48,7 +48,7 @@ def test_numeric_kernel_arg_complex_scalar(input_arrays):
     a, b, _ = input_arrays
     s = a.dtype.type(2 + 1j)
 
-    kernel_scalar[dpex.Range(N)](a, b, s)
+    dpex.call_kernel(kernel_scalar, dpex.Range(N), a, b, s)
 
     nb = dpnp.asnumpy(b)
     nexpected = numpy.full_like(nb, fill_value=2 + 1j)
@@ -65,7 +65,7 @@ def test_numeric_kernel_arg_complex_array(input_arrays):
 
     a, b, c = input_arrays
 
-    kernel_array[dpex.Range(N)](a, b, c)
+    dpex.call_kernel(kernel_array, dpex.Range(N), a, b, c)
 
     nb = dpnp.asnumpy(b)
     nexpected = numpy.full_like(nb, fill_value=0 + 0j)

numba_dpex/tests/kernel_tests/test_dpnp_ndarray_args.py

@@ -85,7 +85,9 @@ def test_setting_private_from_dpnp_ndarray():
     global_range = ndpx.Range(N_POINTS // N_POINTS_PER_WORK_ITEM)
     local_range = ndpx.Range(LOCAL_SIZE)
     try:
-        _kernel[ndpx.NdRange(global_range, local_range)](COEFFICIENTS)
+        ndpx.call_kernel(
+            _kernel, ndpx.NdRange(global_range, local_range), COEFFICIENTS
+        )
     except Exception as e:
         assert (
             False

numba_dpex/tests/kernel_tests/test_func.py

@@ -24,6 +24,6 @@ def test_func_call_from_kernel():
     a = dpnp.ones(1024)
     b = dpnp.ones(1024)
 
-    f[dpex.Range(1024)](a, b)
+    dpex.call_kernel(f, dpex.Range(1024), a, b)
     nb = dpnp.asnumpy(b)
     assert numpy.all(nb == 2)

numba_dpex/tests/kernel_tests/test_func_specialization.py

@@ -35,7 +35,7 @@ def kernel_function(a, b):
     a = dpnp.ones(N)
     b = dpnp.ones(N)
 
-    k[dpex.Range(N)](a, b)
+    dpex.call_kernel(k, dpex.Range(N), a, b)
 
     assert np.array_equal(dpnp.asnumpy(b), dpnp.asnumpy(a) + 1)
 
@@ -56,7 +56,7 @@ def kernel_function(a, b):
     a = dpnp.ones(N, dtype=dpnp.int32)
     b = dpnp.ones(N, dtype=dpnp.int32)
 
-    k[dpex.Range(N)](a, b)
+    dpex.call_kernel(k, dpex.Range(N), a, b)
 
     assert np.array_equal(dpnp.asnumpy(b), dpnp.asnumpy(a) + 1)
 
@@ -65,7 +65,7 @@ def kernel_function(a, b):
     b = dpnp.ones(N, dtype=dpnp.int64)
 
     with pytest.raises(Exception) as e:
-        k[dpex.Range(N)](a, b)
+        dpex.call_kernel(k, dpex.Range(N), a, b)
 
     assert " >>> <unknown function>(int64)" in e.value.args[0]
 
@@ -86,15 +86,15 @@ def kernel_function(a, b):
     a = dpnp.ones(N, dtype=dpnp.int32)
     b = dpnp.ones(N, dtype=dpnp.int32)
 
-    k[dpex.Range(N)](a, b)
+    dpex.call_kernel(k, dpex.Range(N), a, b)
 
     assert np.array_equal(dpnp.asnumpy(b), dpnp.asnumpy(a) + 1)
 
     # Test with float32, should work
     a = dpnp.ones(N, dtype=dpnp.float32)
     b = dpnp.ones(N, dtype=dpnp.float32)
 
-    k[dpex.Range(N)](a, b)
+    dpex.call_kernel(k, dpex.Range(N), a, b)
 
     assert np.array_equal(dpnp.asnumpy(b), dpnp.asnumpy(a) + 1)
 
@@ -103,6 +103,6 @@ def kernel_function(a, b):
     b = dpnp.ones(N, dtype=dpnp.int64)
 
     with pytest.raises(Exception) as e:
-        k[dpex.Range(N)](a, b)
+        dpex.call_kernel(k, dpex.Range(N), a, b)
 
     assert " >>> <unknown function>(int64)" in e.value.args[0]

numba_dpex/tests/kernel_tests/test_invalid_kernel_args.py

@@ -26,4 +26,4 @@ def test_passing_numpy_arrays_as_kernel_args():
     c = numpy.zeros(N)
 
     with pytest.raises(UnsupportedKernelArgumentError):
-        vecadd_kernel[dpex.Range(N)](a, b, c)
+        dpex.call_kernel(vecadd_kernel, dpex.Range(N), a, b, c)

numba_dpex/tests/kernel_tests/test_kernel_has_return_value_error.py

@@ -31,4 +31,4 @@ def test_return(sig):
 
     with pytest.raises(dpex.core.exceptions.KernelHasReturnValueError):
         kernel_fn = dpex.kernel(sig)(f)
-        kernel_fn[dpex.Range(a.size)](a)
+        dpex.call_kernel(kernel_fn, dpex.Range(a.size), a)

numba_dpex/tests/kernel_tests/test_kernel_specialization.py

@@ -60,7 +60,9 @@ def test_missing_specialization_error():
     c = dpt.zeros(1024, dtype=dpt.int32)
 
     with pytest.raises(MissingSpecializationError):
-        specialized_kernel1(data_parallel_sum)[Range(1024)](a, b, c)
+        dpex.call_kernel(
+            specialized_kernel1(data_parallel_sum), Range(1024), a, b, c
+        )
 
 
 def test_execution_of_specialized_kernel():
@@ -69,7 +71,9 @@ def test_execution_of_specialized_kernel():
     b = dpt.ones(1024, dtype=dpt.int64)
     c = dpt.zeros(1024, dtype=dpt.int64)
 
-    specialized_kernel1(data_parallel_sum)[Range(1024)](a, b, c)
+    dpex.call_kernel(
+        specialized_kernel1(data_parallel_sum), Range(1024), a, b, c
+    )
 
     npc = dpt.asnumpy(c)
     import numpy as np

numba_dpex/tests/kernel_tests/test_math_functions.py

@@ -43,7 +43,7 @@ def f(a, b):
         i = dpex.get_global_id(0)
         b[i] = uop(a[i])
 
-    f[dpex.Range(a.size)](a, b)
+    dpex.call_kernel(f, dpex.Range(a.size), a, b)
 
     expected = dpnp_uop(a)
 

numba_dpex/tests/kernel_tests/test_ndrange_exceptions.py

@@ -34,4 +34,4 @@ def test_ndrange_config_error(error, ranges):
 
     with pytest.raises(error):
         range = NdRange(ranges[0], ranges[1])
-        kernel_vector_sum[range](a, b, c)
+        ndpx.call_kernel(kernel_vector_sum, range, a, b, c)

numba_dpex/tests/kernel_tests/test_print.py

@@ -42,7 +42,7 @@ def print_scalar_val(s):
 
     a = input_arrays
 
-    print_scalar_val[dpex.Range(1)](a)
+    dpex.call_kernel(print_scalar_val, dpex.Range(1), a)
     captured = capfd.readouterr()
     assert "printing ... 10" in captured.out
 
@@ -60,7 +60,7 @@ def print_scalar_val(s):
 
     a = input_arrays
 
-    print_scalar_val[dpex.Range(1)](a)
+    dpex.call_kernel(print_scalar_val, dpex.Range(1), a)
     captured = capfd.readouterr()
 
     assert "10" in captured.out
@@ -85,7 +85,7 @@ def print_string(a):
     a = input_arrays
 
     with pytest.raises(LoweringError):
-        print_string[dpex.Range(1)](a)
+        dpex.call_kernel(print_string, dpex.Range(1), a)
 
 
 @skip_on_gpu
@@ -101,4 +101,4 @@ def print_string(a):
     a = input_arrays
 
     with pytest.raises(LoweringError):
-        print_string[dpex.Range(1)](a)
+        dpex.call_kernel(print_string, dpex.Range(1), a)

numba_dpex/tests/kernel_tests/test_private_memory_allocation.py

@@ -23,7 +23,7 @@ def kernel_with_private_memory_allocation(A):
 def test_private_memory_allocation():
     N = 64
     arr = dpnp.zeros(N, dtype=dpnp.float32)
-    kernel_with_private_memory_allocation[dpex.Range(N)](arr)
+    dpex.call_kernel(kernel_with_private_memory_allocation, dpex.Range(N), arr)
 
     nparr = dpnp.asnumpy(arr)
 

numba_dpex/tests/kernel_tests/test_scalar_arg_types.py

@@ -49,7 +49,7 @@ def test_numeric_kernel_arg_types1(input_arrays):
     a, b = input_arrays
     s = a.dtype.type(2)
 
-    scaling_kernel[dpex.Range(N)](a, b, s)
+    dpex.call_kernel(scaling_kernel, dpex.Range(N), a, b, s)
 
     nb = dpnp.asnumpy(b)
     nexpected = numpy.full_like(nb, fill_value=2)
@@ -65,14 +65,14 @@ def test_bool_kernel_arg_type(input_arrays):
     """
     a, b = input_arrays
 
-    kernel_with_bool_arg[dpex.Range(a.size)](a, b, True)
+    dpex.call_kernel(kernel_with_bool_arg, dpex.Range(a.size), a, b, True)
 
     nb = dpnp.asnumpy(b)
     nexpected_true = numpy.full_like(nb, fill_value=2)
 
     assert numpy.allclose(nb, nexpected_true)
 
-    kernel_with_bool_arg[dpex.Range(a.size)](a, b, False)
+    dpex.call_kernel(kernel_with_bool_arg, dpex.Range(a.size), a, b, False)
 
     nb = dpnp.asnumpy(b)
     nexpected_false = numpy.zeros_like(nb)

numba_dpex/tests/kernel_tests/test_sycl_usm_array_iface_interop.py

@@ -88,7 +88,7 @@ def test_kernel_valid_usm_obj(dtype):
     C = DuckUSMArray(shape=buffC.shape, dtype=dtype, host_buffer=buffC)
 
     try:
-        vecadd[dpex.Range(N)](A, B, C)
+        dpex.call_kernel(vecadd, dpex.Range(N), A, B, C)
     except Exception:
         pytest.fail(
             "Could not pass Python object with sycl_usm_array_interface"
@@ -112,4 +112,4 @@ def test_kernel_invalid_usm_obj(dtype):
     C = PseudoDuckUSMArray()
 
     with pytest.raises(Exception):
-        vecadd[dpex.Range(N)](A, B, C)
+        dpex.call_kernel(vecadd, dpex.Range(N), A, B, C)

numba_dpex/tests/kernel_tests/test_usm_ndarray_interop.py

@@ -52,7 +52,7 @@ def data_parallel_sum(a, b, c):
 
     c = dpt.empty_like(a)
 
-    data_parallel_sum[dpex.Range(N, N)](a, b, c)
+    dpex.call_kernel(data_parallel_sum, dpex.Range(N, N), a, b, c)
 
     na = dpt.asnumpy(a)
     nb = dpt.asnumpy(b)

numba_dpex/tests/misc/test_warnings.py

@@ -17,7 +17,7 @@ def test_opt_warning():
     config.DPEX_OPT = 3
 
     with pytest.warns(UserWarning):
-        foo[dpex.Range(10)](dpnp.arange(10))
+        dpex.call_kernel(foo, dpex.Range(10), dpnp.arange(10))
 
     config.DPEX_OPT = bkp
 
@@ -27,7 +27,7 @@ def test_inline_threshold_eq_3_warning():
     config.INLINE_THRESHOLD = 3
 
     with pytest.warns(UserWarning):
-        foo[dpex.Range(10)](dpnp.arange(10))
+        dpex.call_kernel(foo, dpex.Range(10), dpnp.arange(10))
 
     config.INLINE_THRESHOLD = bkp
 
@@ -37,7 +37,7 @@ def test_inline_threshold_negative_val_warning_():
     config.INLINE_THRESHOLD = -1
 
     with pytest.warns(UserWarning):
-        foo[dpex.Range(10)](dpnp.arange(10))
+        dpex.call_kernel(foo, dpex.Range(10), dpnp.arange(10))
 
     config.INLINE_THRESHOLD = bkp
 
@@ -47,12 +47,12 @@ def test_inline_threshold_gt_3_warning():
     config.INLINE_THRESHOLD = 4
 
     with pytest.warns(UserWarning):
-        foo[dpex.Range(10)](dpnp.arange(10))
+        dpex.call_kernel(foo, dpex.Range(10), dpnp.arange(10))
 
     config.INLINE_THRESHOLD = bkp
 
 
 def test_no_warning():
     with warnings.catch_warnings():
         warnings.simplefilter("error")
-        foo[dpex.Range(10)](dpnp.arange(10))
+        dpex.call_kernel(foo, dpex.Range(10), dpnp.arange(10))