update readme to include Intel GPU (#440)

PaddlePaddle · Mar 9, 2023 · cf66c06 · cf66c06
1 parent d2ce4a1
commit cf66c06
Show file tree

Hide file tree

Showing 6 changed files with 34 additions and 42 deletions.
diff --git a/Paddle b/Paddle
diff --git a/README.md b/README.md
@@ -1,25 +1,21 @@
 # PaddleCustomDevice
 
-English | [简体中文](./README_cn.md)
+简体中文 | [English](./README_en.md)
 
-PaddlePaddle custom device implementaion.
+『飞桨』自定义硬件接入实现。
 
-## User Guides
+## 使用指南
 
-To follow up on latest features of custom device in PaddlePaddle, please refer to [Custom Device Support](https://www.paddlepaddle.org.cn/documentation/docs/en/develop/dev_guides/custom_device_docs/index_en.html)
+方案设计参考[Custom Device 接入方案介绍](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/dev_guides/custom_device_docs/custom_device_overview_cn.html)，开发指南请参考[新硬件接入示例](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/dev_guides/custom_device_docs/custom_device_example_cn.html)且示例代码位于 [CustomCPU](backends/custom_cpu/README_cn.md)。
 
-## Hardware Backends
+## 硬件后端
 
-PaddleCustomDevice has supported the following backends:
+飞桨自定义硬件接入支持如下硬件后端：
 
-- [PaddlePaddle Custom Device Implementaion for Ascend NPU](backends/npu/README.md)
-- [PaddlePaddle Custom Device Implementaion for Custom CPU](backends/custom_cpu/README.md)
-- [PaddlePaddle Custom Device Implementaion for Cambricon MLU](backends/mlu/README.md)
+- [飞桨自定义接入硬件后端(昇腾NPU)](backends/npu/README_cn.md)
+- [飞桨自定义接入硬件后端(寒武纪MLU)](backends/mlu/README_cn.md)
+- [飞桨自定义接入硬件后端(英特尔GPU)](backends/intel_gpu/README.md)
 
-## Environment Variables
-| Subject     | Variable Name       | Type   | Description    | Default Value |
-| -------- | -------------------------------- | ------ | --------------------------------- | ------------------------------------------------------------ |
-| Debug | CUSTOM_DEVICE_BLACK_LIST| String |   Ops in black_list will be fallbacked to CPU  |                                            |
+## 版权和许可证
 
-## Copyright and License
-PaddleCustomDevice is provided under the [Apache-2.0 license](LICENSE).
+PaddleCustomDevice由[Apache-2.0 license](LICENSE)提供。
diff --git a/README_cn.md b/README_cn.md
diff --git a/README_en.md b/README_en.md
@@ -0,0 +1,20 @@
+# PaddleCustomDevice
+
+English | [简体中文](./README.md)
+
+PaddlePaddle custom device implementaion.
+
+## User Guides
+
+Design doc refer to [Custom Device Overview](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/dev_guides/custom_device_docs/custom_device_overview_cn.html), and user guidance refer to [Custom Device Support](https://www.paddlepaddle.org.cn/documentation/docs/en/develop/dev_guides/custom_device_docs/index_en.html) with demo code in [CustomCPU](backends/custom_cpu/README_cn.md)。
+
+## Hardware Backends
+
+PaddleCustomDevice has supported the following backends:
+
+- [PaddlePaddle Custom Device Implementaion for Ascend NPU](backends/npu/README.md)
+- [PaddlePaddle Custom Device Implementaion for Cambricon MLU](backends/mlu/README.md)
+- [PaddlePaddle Custom Device Implementaion for Intel GPU](backends/intel_gpu/README.md)
+
+## Copyright and License
+PaddleCustomDevice is provided under the [Apache-2.0 license](LICENSE).
diff --git a/backends/npu/README.md b/backends/npu/README.md
@@ -171,6 +171,7 @@ WITH_ARM=OFF # Turn ON if aarch64
 
 | Subject     | Variable Name       | Type   | Description    | Default Value |
 | -------- | -------------------------------- | ------ | --------------------------------- | ------------------------------------------------------------ |
+| Debug     | CUSTOM_DEVICE_BLACK_LIST| String | Ops in back list will fallbacks to CPU  |  ""  |
 | Debug     | FLAGS_npu_check_nan_inf | Bool   | check nan or inf of all npu kernels | False                                                       |
 | Debug     | FLAGS_npu_blocking_run | Bool   | enable sync for all npu kernels | False                                                     |
 | Profiling | FLAGS_npu_profiling_dir | String |   ACL profiling output dir     | "ascend_profiling"                                           |

diff --git a/backends/npu/README_cn.md b/backends/npu/README_cn.md
@@ -170,6 +170,7 @@ WITH_ARM=OFF # 如果是 Aarch 环境，请设置为 ON
 
 | 主题   | 变量名称                         | 类型   | 说明                              | 默认值                                                       |
 | -------- | -------------------------------- | ------ | --------------------------------- | ------------------------------------------------------------ |
+| 调试     | CUSTOM_DEVICE_BLACK_LIST  | String   | 在黑名单内的算子会异构到CPU上运行 | "" |
 | 调试     | FLAGS_npu_check_nan_inf | Bool   | 是否开启所有NPU算子输入输出检查   | False                                                        |
 | 调试     | FLAGS_npu_blocking_run | Bool   | 是否开启强制同步执行所有 NPU 算子 | False                                                        |
 | 性能分析 | FLAGS_npu_profiling_dir | String | 设置 Profiling 数据保存目录       | "ascend_profiling"                                           |