Is this a duplicate?
Is this for new documentation, or an update to existing docs?
Update
Describe the incorrect/future/missing documentation
The rendered page for PinnedMemoryResource.allocate is a single sentence. It does not say what the required keyword-only stream argument must be, what the method returns, or that the method raises RuntimeError when the device does not support the requested host memory pool. The one-line docstring on the override in cuda_core/cuda/core/_memory/_pinned_memory_resource.pyx is all autodoc has to render.
from cuda.core import PinnedMemoryResource
doc = PinnedMemoryResource.allocate.__doc__
print(doc)
print("documents the 'stream' parameter:", "stream :" in doc)
print("documents a return value:", "Returns" in doc)
PinnedMemoryResource.allocate(self, size_t size, *, stream: Stream | GraphBuilder) -> Buffer
Allocate a host-pinned buffer asynchronously on the supplied stream.
documents the 'stream' parameter: False
documents a return value: False
Expected: the page states that stream is required and has no default, what allocate returns, and that it raises RuntimeError on a device without the required host memory pool support. Either the full numpydoc body or a cross-reference to the base class would fill the gap, and both shapes are already used elsewhere in cuda.core.
If this is a correction, please provide a link to the incorrect documentation. If this is a new documentation request, please link to where you have looked.
https://nvidia.github.io/cuda-python/cuda-core/latest/generated/cuda.core.PinnedMemoryResource.html
Is this a duplicate?
Is this for new documentation, or an update to existing docs?
Update
Describe the incorrect/future/missing documentation
The rendered page for
PinnedMemoryResource.allocateis a single sentence. It does not say what the required keyword-onlystreamargument must be, what the method returns, or that the method raisesRuntimeErrorwhen the device does not support the requested host memory pool. The one-line docstring on the override incuda_core/cuda/core/_memory/_pinned_memory_resource.pyxis all autodoc has to render.Expected: the page states that
streamis required and has no default, whatallocatereturns, and that it raisesRuntimeErroron a device without the required host memory pool support. Either the full numpydoc body or a cross-reference to the base class would fill the gap, and both shapes are already used elsewhere incuda.core.If this is a correction, please provide a link to the incorrect documentation. If this is a new documentation request, please link to where you have looked.
https://nvidia.github.io/cuda-python/cuda-core/latest/generated/cuda.core.PinnedMemoryResource.html