Move everything to cuda.core.experimental

cuda_core/cuda/core/__init__.py

@@ -2,9 +2,4 @@
 #
 # SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
-from cuda.core._device import Device
-from cuda.core._event import EventOptions
-from cuda.core._launcher import LaunchConfig, launch
-from cuda.core._program import Program
-from cuda.core._stream import Stream, StreamOptions
 from cuda.core._version import __version__

cuda_core/cuda/core/experimental/__init__.py

@@ -0,0 +1,9 @@
+# Copyright (c) 2024, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+#
+# SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
+
+from cuda.core.experimental._device import Device
+from cuda.core.experimental._event import EventOptions
+from cuda.core.experimental._launcher import LaunchConfig, launch
+from cuda.core.experimental._program import Program
+from cuda.core.experimental._stream import Stream, StreamOptions

cuda_core/cuda/core/_context.py → cuda_core/cuda/core/experimental/_context.py

@@ -5,7 +5,7 @@
 from dataclasses import dataclass
 
 from cuda import cuda, cudart
-from cuda.core._utils import handle_return
+from cuda.core.experimental._utils import handle_return
 
 
 @dataclass

cuda_core/cuda/core/_device.py → cuda_core/cuda/core/experimental/_device.py

@@ -7,11 +7,11 @@
 import warnings
 
 from cuda import cuda, cudart
-from cuda.core._utils import handle_return, ComputeCapability, CUDAError, \
+from cuda.core.experimental._utils import handle_return, ComputeCapability, CUDAError, \
                              precondition
-from cuda.core._context import Context, ContextOptions
-from cuda.core._memory import _DefaultAsyncMempool, Buffer, MemoryResource
-from cuda.core._stream import default_stream, Stream, StreamOptions
+from cuda.core.experimental._context import Context, ContextOptions
+from cuda.core.experimental._memory import _DefaultAsyncMempool, Buffer, MemoryResource
+from cuda.core.experimental._stream import default_stream, Stream, StreamOptions
 
 
 _tls = threading.local()
@@ -125,7 +125,7 @@ def set_current(self, ctx: Context=None) -> Union[Context, None]:
         Entry point of this object. Users always start a code by
         calling this method, e.g.
 
-        >>> from cuda.core import Device
+        >>> from cuda.core.experimental import Device
         >>> dev0 = Device(0)
         >>> dev0.set_current()
         >>> # ... do work on device 0 ...

cuda_core/cuda/core/_event.py → cuda_core/cuda/core/experimental/_event.py

@@ -6,9 +6,9 @@
 from typing import Optional
 
 from cuda import cuda
-from cuda.core._utils import check_or_create_options
-from cuda.core._utils import CUDAError
-from cuda.core._utils import handle_return
+from cuda.core.experimental._utils import check_or_create_options
+from cuda.core.experimental._utils import CUDAError
+from cuda.core.experimental._utils import handle_return
 
 
 @dataclass

cuda_core/cuda/core/_kernel_arg_handler.pyx → cuda_core/cuda/core/experimental/_kernel_arg_handler.pyx

@@ -15,7 +15,7 @@ import ctypes
 
 import numpy
 
-from cuda.core._memory import Buffer
+from cuda.core.experimental._memory import Buffer
 
 
 ctypedef cpp_complex.complex[float] cpp_single_complex

cuda_core/cuda/core/_launcher.py → cuda_core/cuda/core/experimental/_launcher.py

@@ -8,11 +8,11 @@
 import numpy as np
 
 from cuda import cuda, cudart
-from cuda.core._kernel_arg_handler import ParamHolder
-from cuda.core._memory import Buffer
-from cuda.core._module import Kernel
-from cuda.core._stream import Stream
-from cuda.core._utils import CUDAError, check_or_create_options, handle_return
+from cuda.core.experimental._kernel_arg_handler import ParamHolder
+from cuda.core.experimental._memory import Buffer
+from cuda.core.experimental._module import Kernel
+from cuda.core.experimental._stream import Stream
+from cuda.core.experimental._utils import CUDAError, check_or_create_options, handle_return
 
 
 @dataclass

cuda_core/cuda/core/_memory.py → cuda_core/cuda/core/experimental/_memory.py

@@ -9,9 +9,9 @@
 import warnings
 
 from cuda import cuda
-from cuda.core._dlpack import DLDeviceType, make_py_capsule
-from cuda.core._stream import default_stream
-from cuda.core._utils import handle_return
+from cuda.core.experimental._dlpack import DLDeviceType, make_py_capsule
+from cuda.core.experimental._stream import default_stream
+from cuda.core.experimental._utils import handle_return
 
 
 PyCapsule = TypeVar("PyCapsule")

cuda_core/cuda/core/_memoryview.pyx → cuda_core/cuda/core/experimental/_memoryview.pyx

@@ -12,7 +12,7 @@ from typing import Any, Optional
 from cuda import cuda
 import numpy
 
-from cuda.core._utils import handle_return
+from cuda.core.experimental._utils import handle_return
 
 
 # TODO(leofang): support NumPy structured dtypes

cuda_core/cuda/core/_module.py → cuda_core/cuda/core/experimental/_module.py

@@ -3,7 +3,7 @@
 # SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
 from cuda import cuda, cudart
-from cuda.core._utils import handle_return
+from cuda.core.experimental._utils import handle_return
 
 
 _backend = {

cuda_core/cuda/core/_program.py → cuda_core/cuda/core/experimental/_program.py

@@ -3,8 +3,8 @@
 # SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
 from cuda import nvrtc
-from cuda.core._utils import handle_return
-from cuda.core._module import ObjectCode
+from cuda.core.experimental._utils import handle_return
+from cuda.core.experimental._module import ObjectCode
 
 
 class Program:

cuda_core/cuda/core/_stream.py → cuda_core/cuda/core/experimental/_stream.py

@@ -9,13 +9,13 @@
 from typing import Optional, Tuple, TYPE_CHECKING, Union
 
 if TYPE_CHECKING:
-    from cuda.core._device import Device
+    from cuda.core.experimental._device import Device
 from cuda import cuda, cudart
-from cuda.core._context import Context
-from cuda.core._event import Event, EventOptions
-from cuda.core._utils import check_or_create_options
-from cuda.core._utils import get_device_from_ctx
-from cuda.core._utils import handle_return
+from cuda.core.experimental._context import Context
+from cuda.core.experimental._event import Event, EventOptions
+from cuda.core.experimental._utils import check_or_create_options
+from cuda.core.experimental._utils import get_device_from_ctx
+from cuda.core.experimental._utils import handle_return
 
 
 @dataclass
@@ -181,7 +181,7 @@ def device(self) -> Device:
         # Note that Stream.device.context might not necessarily agree with
         # Stream.context, in cases where a different CUDA context is set
         # current after a stream was created.
-        from cuda.core._device import Device  # avoid circular import
+        from cuda.core.experimental._device import Device  # avoid circular import
         if self._device_id is None:
             # Get the stream context first
             if self._ctx_handle is None:

cuda_core/cuda/core/utils.py → cuda_core/cuda/core/experimental/utils.py

@@ -2,4 +2,4 @@
 #
 # SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
-from cuda.core._memoryview import StridedMemoryView, viewable
+from cuda.core.experimental._memoryview import StridedMemoryView, viewable

cuda_core/docs/source/api.rst

@@ -1,7 +1,7 @@
-.. module:: cuda.core
+.. module:: cuda.core.experimental
 
-``cuda.core`` API Reference
-===========================
+``cuda.core.experimental`` API Reference
+========================================
 
 CUDA runtime
 ------------

cuda_core/examples/saxpy.py

@@ -4,9 +4,9 @@
 
 import sys
 
-from cuda.core import Device
-from cuda.core import LaunchConfig, launch
-from cuda.core import Program
+from cuda.core.experimental import Device
+from cuda.core.experimental import LaunchConfig, launch
+from cuda.core.experimental import Program
 
 import cupy as cp
 

cuda_core/examples/vector_add.py

@@ -2,9 +2,9 @@
 #
 # SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
-from cuda.core import Device
-from cuda.core import LaunchConfig, launch
-from cuda.core import Program
+from cuda.core.experimental import Device
+from cuda.core.experimental import LaunchConfig, launch
+from cuda.core.experimental import Program
 
 import cupy as cp
 

cuda_core/pyproject.toml

@@ -46,8 +46,8 @@ dependencies = [
 ]
 
 
-[tool.setuptools]
-packages = ["cuda", "cuda.core"]
+[tool.setuptools.packages.find]
+include = ["cuda.core*"]
 
 
 [tool.setuptools.dynamic]

cuda_core/setup.py

@@ -5,24 +5,24 @@
 import os
 
 from Cython.Build import cythonize
-from setuptools import setup, Extension, find_packages
+from setuptools import setup, Extension
 from setuptools.command.build_ext import build_ext as _build_ext
 
 
 ext_modules = (
     Extension(
-        "cuda.core._dlpack",
-        sources=["cuda/core/_dlpack.pyx"],
+        "cuda.core.experimental._dlpack",
+        sources=["cuda/core/experimental/_dlpack.pyx"],
         language="c++",
     ),
     Extension(
-        "cuda.core._memoryview",
-        sources=["cuda/core/_memoryview.pyx"],
+        "cuda.core.experimental._memoryview",
+        sources=["cuda/core/experimental/_memoryview.pyx"],
         language="c++",
     ),
     Extension(
-        "cuda.core._kernel_arg_handler",
-        sources=["cuda/core/_kernel_arg_handler.pyx"],
+        "cuda.core.experimental._kernel_arg_handler",
+        sources=["cuda/core/experimental/_kernel_arg_handler.pyx"],
         language="c++",
     ),
 )
@@ -39,11 +39,6 @@ def build_extensions(self):
     ext_modules=cythonize(ext_modules,
         verbose=True, language_level=3,
         compiler_directives={'embedsignature': True}),
-    packages=find_packages(include=['cuda.core', 'cuda.core.*']),
-    package_data=dict.fromkeys(
-        find_packages(include=["cuda.core.*"]),
-        ["*.pxd", "*.pyx", "*.py"],
-    ),
     cmdclass = {'build_ext': build_ext,},
     zip_safe=False,
 )