cyx-6 commented on code in PR #169:
URL: https://github.com/apache/tvm-ffi/pull/169#discussion_r2543881908


##########
docs/guides/kernel_library_guide.rst:
##########
@@ -0,0 +1,167 @@
+.. Licensed to the Apache Software Foundation (ASF) under one
+.. or more contributor license agreements.  See the NOTICE file
+.. distributed with this work for additional information
+.. regarding copyright ownership.  The ASF licenses this file
+.. to you under the Apache License, Version 2.0 (the
+.. "License"); you may not use this file except in compliance
+.. with the License.  You may obtain a copy of the License at
+..
+..   http://www.apache.org/licenses/LICENSE-2.0
+..
+.. Unless required by applicable law or agreed to in writing,
+.. software distributed under the License is distributed on an
+.. "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+.. KIND, either express or implied.  See the License for the
+.. specific language governing permissions and limitations
+.. under the License.
+
+====================
+Kernel Library Guide
+====================
+
+This guide serves as a quick start for shipping python version and machine 
learning(ML) framework agnostic kernel libraries with TVM FFI. With the help of 
TVM FFI, we can connect the kernel libraries to multiple ML framework, such as 
PyTorch, XLA, JAX, together with the minimal efforts.
+
+Tensor
+======
+
+Almost all kernel libraries are about tensor computation and manipulation. For 
better adaptation to different ML frameworks, TVM FFI provides a minimal set of 
data structures to represent tensors from ML frameworks, including the tensor 
basic attributes and storage pointer.
+To be specific, in TVM FFI, two types of tensor constructs, 
:cpp:class:`~tvm::ffi::Tensor` and :cpp:class:`~tvm::ffi::TensorView`, can be 
used to represent a tensor from ML frameworks.
+
+Tensor and TensorView
+---------------------
+
+Though both :cpp:class:`~tvm::ffi::Tensor` and 
:cpp:class:`~tvm::ffi::TensorView` are designed to represent tensors from ML 
frameworks that interact with the TVM FFI ABI. They are backed by the 
`DLTensor` in DLPack in practice. The main difference is whether it is an 
owning tensor structure.
+
+:cpp:class:`tvm::ffi::Tensor`
+ :cpp:class:`~tvm::ffi::Tensor` is a completely owning tensor with reference 
counting. It can be created and passed between C++ and Python side safely. When 
the counting reference goes to zero, its underlying deleter function will be 
called to free the tensor storage.

Review Comment:
   updated



##########
docs/guides/kernel_library_guide.rst:
##########
@@ -0,0 +1,167 @@
+.. Licensed to the Apache Software Foundation (ASF) under one
+.. or more contributor license agreements.  See the NOTICE file
+.. distributed with this work for additional information
+.. regarding copyright ownership.  The ASF licenses this file
+.. to you under the Apache License, Version 2.0 (the
+.. "License"); you may not use this file except in compliance
+.. with the License.  You may obtain a copy of the License at
+..
+..   http://www.apache.org/licenses/LICENSE-2.0
+..
+.. Unless required by applicable law or agreed to in writing,
+.. software distributed under the License is distributed on an
+.. "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+.. KIND, either express or implied.  See the License for the
+.. specific language governing permissions and limitations
+.. under the License.
+
+====================
+Kernel Library Guide
+====================
+
+This guide serves as a quick start for shipping python version and machine 
learning(ML) framework agnostic kernel libraries with TVM FFI. With the help of 
TVM FFI, we can connect the kernel libraries to multiple ML framework, such as 
PyTorch, XLA, JAX, together with the minimal efforts.
+
+Tensor
+======
+
+Almost all kernel libraries are about tensor computation and manipulation. For 
better adaptation to different ML frameworks, TVM FFI provides a minimal set of 
data structures to represent tensors from ML frameworks, including the tensor 
basic attributes and storage pointer.
+To be specific, in TVM FFI, two types of tensor constructs, 
:cpp:class:`~tvm::ffi::Tensor` and :cpp:class:`~tvm::ffi::TensorView`, can be 
used to represent a tensor from ML frameworks.
+
+Tensor and TensorView
+---------------------
+
+Though both :cpp:class:`~tvm::ffi::Tensor` and 
:cpp:class:`~tvm::ffi::TensorView` are designed to represent tensors from ML 
frameworks that interact with the TVM FFI ABI. They are backed by the 
`DLTensor` in DLPack in practice. The main difference is whether it is an 
owning tensor structure.
+
+:cpp:class:`tvm::ffi::Tensor`
+ :cpp:class:`~tvm::ffi::Tensor` is a completely owning tensor with reference 
counting. It can be created and passed between C++ and Python side safely. When 
the counting reference goes to zero, its underlying deleter function will be 
called to free the tensor storage.
+
+:cpp:class:`tvm::ffi::TensorView`
+ :cpp:class:`~tvm::ffi::TensorView` is a non-owning view of an existing 
tensor, pointing to an existing tensor (e.g., a tensor allocated by PyTorch).
+
+It is **recommended** to use :cpp:class:`~tvm::ffi::TensorView` when possible, 
that helps us to support more cases, including cases where only view but not 
strong reference are passed, like XLA buffer.
+It is also more lightweight. However, since :cpp:class:`~tvm::ffi::TensorView` 
is a non-owning view, it is the user's responsibility to ensure the lifetime of 
underlying tensor data and attributes of the viewed tensor object.

Review Comment:
   removed



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to