From 4678b1740f886fae3e07ceb986e83848b9cc4fbe Mon Sep 17 00:00:00 2001 From: geekun Date: Sat, 29 Aug 2020 17:24:44 +0800 Subject: [PATCH] refresh the guide of the custom operator --- tutorials/source_en/use/custom_operator.md | 10 ++++++---- tutorials/source_zh_cn/use/custom_operator.md | 10 ++++++---- 2 files changed, 12 insertions(+), 8 deletions(-) diff --git a/tutorials/source_en/use/custom_operator.md b/tutorials/source_en/use/custom_operator.md index 7e280e7b..0f05c9e4 100644 --- a/tutorials/source_en/use/custom_operator.md +++ b/tutorials/source_en/use/custom_operator.md @@ -29,7 +29,9 @@ The related concepts are as follows: - Operator implementation: describes the implementation of the internal computation logic for an operator through the DSL API provided by the Tensor Boost Engine (TBE). The TBE supports the development of custom operators based on the Ascend AI chip. You can apply for Open Beta Tests (OBTs) by visiting . - Operator information: describes basic information about a TBE operator, such as the operator name and supported input and output types. It is the basis for the backend to select and map operators. -This section takes a Square operator as an example to describe how to customize an operator. For details, see cases in [tests/st/ops/custom_ops_tbe](https://gitee.com/mindspore/mindspore/tree/r0.7/tests/st/ops/custom_ops_tbe) in the MindSpore source code. +This section takes a Square operator as an example to describe how to customize an operator. + +> For details, see cases in [tests/st/ops/custom_ops_tbe](https://gitee.com/mindspore/mindspore/tree/r0.7/tests/st/ops/custom_ops_tbe) in the MindSpore source code. ## Registering the Operator Primitive @@ -79,7 +81,7 @@ The entry function of an operator describes the internal process of compiling th 4. Call `cce_build_code` to compile and generate an operator binary file. > The input parameters of the entry function require the input information of each operator, output information of each operator, operator attributes (optional), and `kernel_name` (name of the generated operator binary file). The input and output information is encapsulated in dictionaries, including the input and output shape and dtype when the operator is called on the network. -For details about TBE operator development, visit the [TBE website](https://www.huaweicloud.com/ascend/dev/operator). For details about how to debug and optimize the TBE operator, visit the [Mind Studio website](https://www.huaweicloud.com/intl/en-us/ascend/mindstudio). +For details about TBE operator development, visit the [TBE website]((https://support.huaweicloud.com/odevg-A800_3000_3010/atlaste_10_0063.html)). For details about how to debug and optimize the TBE operator, visit the [Mind Studio website](https://support.huaweicloud.com/usermanual-mindstudioc73/atlasmindstudio_02_0043.html). ### Registering the Operator Information @@ -91,7 +93,7 @@ The operator information is key for the backend to select the operator implement ### Example -The following takes the TBE implementation `square_impl.py` of the `Square` operator as an example. `square_compute` is a computable function of the operator implementation. It describes the computation logic of `x * x` by calling the API provided by `te.lang.cce`. `cus_square_op_info ` is the operator information, which is defined by `TBERegOp`. +The following takes the TBE implementation `square_impl.py` of the `Square` operator as an example. `square_compute` is a computable function of the operator implementation. It describes the computation logic of `x * x` by calling the API provided by `te.lang.cce`. `cus_square_op_info ` is the operator information, which is defined by `TBERegOp`. The specific field meaning of the operator information visit the [TBE website](https://support.huaweicloud.com/odevg-A800_3000_3010/atlaste_10_0096.html). Note the following parameters when setting `TBERegOp`: @@ -247,4 +249,4 @@ The execution result is as follows: ``` x: [1. 4. 9.] dx: [2. 8. 18.] -``` \ No newline at end of file +``` diff --git a/tutorials/source_zh_cn/use/custom_operator.md b/tutorials/source_zh_cn/use/custom_operator.md index 6ae951de..440efc42 100644 --- a/tutorials/source_zh_cn/use/custom_operator.md +++ b/tutorials/source_zh_cn/use/custom_operator.md @@ -29,7 +29,9 @@ - 算子实现：通过TBE（Tensor Boost Engine）提供的特性语言接口，描述算子内部计算逻辑的实现。TBE提供了开发昇腾AI芯片自定义算子的能力。你可以在页面申请公测。 - 算子信息：描述TBE算子的基本信息，如算子名称、支持的输入输出类型等。它是后端做算子选择和映射时的依据。 -本文将以自定义Square算子为例，介绍自定义算子的步骤。更多详细内容可参考MindSpore源码中[tests/st/ops/custom_ops_tbe](https://gitee.com/mindspore/mindspore/tree/r0.7/tests/st/ops/custom_ops_tbe)下的用例。 +本文将以自定义Square算子为例，介绍自定义算子的步骤。 + +> 更多详细内容可参考MindSpore源码中[tests/st/ops/custom_ops_tbe](https://gitee.com/mindspore/mindspore/tree/r0.7/tests/st/ops/custom_ops_tbe)下的用例。 ## 注册算子原语 @@ -79,11 +81,11 @@ class CusSquare(PrimitiveWithInfer): 4. 调用`cce_build_code`编译生成算子二进制。 > 入口函数的输入参数有特殊要求，需要依次为：算子每个输入的信息、算子每个输出的信息、算子属性（可选）和`kernel_name`（生成算子二进制的名称）。输入和输出的信息用字典封装传入，其中包含该算子在网络中被调用时传入的实际输入和输出的shape和dtype。 -更多关于使用TBE开发算子的内容请参考[TBE文档](https://www.huaweicloud.com/ascend/dev/operator)，关于TBE算子的调试和性能优化请参考[MindStudio文档](https://www.huaweicloud.com/ascend/mindstudio)。 +更多关于使用TBE开发算子的内容请参考[TBE文档](https://support.huaweicloud.com/odevg-A800_3000_3010/atlaste_10_0063.html)，关于TBE算子的调试和性能优化请参考[MindStudio文档](https://support.huaweicloud.com/usermanual-mindstudioc73/atlasmindstudio_02_0043.html)。 ### 注册算子信息 -算子信息是指导后端选择算子实现的关键信息，同时也指导后端为算子插入合适的类型和格式转换。它通过`TBERegOp`接口定义，通过`op_info_register`装饰器将算子信息与算子实现入口函数绑定。当算子实现py文件被导入时，`op_info_register`装饰器会将算子信息注册到后端的算子信息库中。更多关于算子信息的使用方法请参考`TBERegOp`的成员方法的注释说明。 +算子信息是指导后端选择算子实现的关键信息，同时也指导后端为算子插入合适的类型和格式转换。它通过`TBERegOp`接口定义，通过`op_info_register`装饰器将算子信息与算子实现入口函数绑定。当算子实现py文件被导入时，`op_info_register`装饰器会将算子信息注册到后端的算子信息库中。更多关于算子信息的使用方法请参考`TBERegOp`的成员方法的注释说明，算子信息的字段含义可以参考[TBE文档](https://support.huaweicloud.com/odevg-A800_3000_3010/atlaste_10_0096.html)。 > - 算子信息中定义输入输出信息的个数和顺序、算子实现入口函数的参数中的输入输出信息的个数和顺序、算子原语中输入输出名称列表的个数和顺序，三者要完全一致。 > - 算子如果带属性，在算子信息中需要用`attr`描述属性信息，属性的名称与算子原语定义中的属性名称要一致。 @@ -247,4 +249,4 @@ pytest -s tests/st/ops/custom_ops_tbe/test_square.py::test_grad_net ``` x: [1. 4. 9.] dx: [2. 8. 18.] -``` \ No newline at end of file +``` -- GitLab