Star1193
1200
Code
Code
Issues140
Pull Requests415
Actions
Discussion
Wiki
Analysis
Star1193
1200
Merged
docs: 按产品名称映射表统一产品名称（样例） #6571
munanhwAdd on Oct 8
docs: 按产品名称映射表统一产品名称（样例） #6571
Merged
munanhwAdd on Oct 8
Merged frommunanhw/asc-devkitdocs/product-name-examplesintocann/asc-devkitmaster
Total: 856 files changed+2944-2944
Mexamples/01_simd_cpp_api/00_introduction/00_quickstart/README.md+1-1
@@ -8,4 +8,4 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-| [hello_world](./hello_world) | 基于Ascend C的HelloWorld样例核函数直调方法 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |11+| [hello_world](./hello_world) | 基于Ascend C的HelloWorld样例核函数直调方法 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/00_introduction/00_quickstart/README_en.md+1-1
@@ -8,4 +8,4 @@ Introduces the kernel function direct invocation method based on Ascend C, demon
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [hello_world](./hello_world) | HelloWorld sample for direct kernel invocation based on Ascend C | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [hello_world](./hello_world) | HelloWorld sample for direct kernel invocation based on Ascend C | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/00_introduction/00_quickstart/hello_world/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -58,7 +58,7 @@
58 | 选项 | 可选值 | 说明 |58 | 选项 | 可选值 | 说明 |
59 |------|--------|------|59 |------|--------|------|
60 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |60 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |
61- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |61+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
62 62 
63- 执行结果 63- 执行结果
64 执行结果如下，说明执行成功。64 执行结果如下，说明执行成功。
Mexamples/01_simd_cpp_api/00_introduction/00_quickstart/hello_world/README_en.md+4-4
@@ -8,9 +8,9 @@ This sample demonstrates the basic process of running and verifying a kernel fun
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -60,7 +60,7 @@ Execute the following steps in the root directory of this sample to build and ru
60 | Option | Available Values | Description |60 | Option | Available Values | Description |
61 |--------|------------------|-------------|61 |--------|------------------|-------------|
62 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |62 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
63- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |63+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
64 64 
65- Execution Results65- Execution Results
66 66 
Mexamples/01_simd_cpp_api/00_introduction/01_add/README.md+2-2
@@ -8,5 +8,5 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-| [add](./add) | 基于静态Tensor编程实现Add向量加法 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |11+| [add](./add) | 基于静态Tensor编程实现Add向量加法 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
12-| [add_tpipe_tque](./add_tpipe_tque) | 基于TQue和TPipe编程实现Add向量加法 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |12+| [add_tpipe_tque](./add_tpipe_tque) | 基于TQue和TPipe编程实现Add向量加法 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/00_introduction/01_add/README_en.md+2-2
@@ -8,5 +8,5 @@ Vector computation samples based on Ascend C, introducing two different memory a
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [add](./add) | Static Tensor programming implementation | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [add](./add) | Static Tensor programming implementation | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [add_tpipe_tque](./add_tpipe_tque) | TQue and TPipe programming implementation | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |12+| [add_tpipe_tque](./add_tpipe_tque) | TQue and TPipe programming implementation | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/00_introduction/01_add/add/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -218,7 +218,7 @@ cat ./OPPROF_*/PipeUtilization.csv
218 | 选项 | 可选值 | 说明 |218 | 选项 | 可选值 | 说明 |
219 |------|--------|------|219 |------|--------|------|
220 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |220 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |
221- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |221+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
222 222 
223- 执行结果223- 执行结果
224 执行结果如下，说明精度对比成功。224 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/00_introduction/01_add/add/README_en.md+4-4
@@ -8,9 +8,9 @@ This sample demonstrates the basic usage of Ascend C vector addition.
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -219,7 +219,7 @@ Execute the following steps in the root directory of this sample to compile and
219 | Option | Available Values | Description |219 | Option | Available Values | Description |
220 |------|--------|------|220 |------|--------|------|
221 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |221 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
222- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |222+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
223 223 
224- Execution Results224- Execution Results
225 The execution result is as follows, indicating that the accuracy comparison succeeds.225 The execution result is as follows, indicating that the accuracy comparison succeeds.
Mexamples/01_simd_cpp_api/00_introduction/01_add/add_tpipe_tque/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -98,7 +98,7 @@
98 | 选项 | 可选值 | 说明 |98 | 选项 | 可选值 | 说明 |
99 |------|--------|------|99 |------|--------|------|
100 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |100 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |
101- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |101+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
102 102 
103- 执行结果 103- 执行结果
104 执行结果如下，说明精度对比成功。104 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/00_introduction/01_add/add_tpipe_tque/README_en.md+4-4
@@ -8,9 +8,9 @@ This sample implements the Add vector addition operation based on the memory and
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure Introduction15## Directory Structure Introduction
16 16 
@@ -98,7 +98,7 @@ Execute the following steps in the root directory of this sample to compile and
98 | Option | Available Values | Description |98 | Option | Available Values | Description |
99 |------|--------|------|99 |------|--------|------|
100 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |100 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
101- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |101+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
102 102 
103- Execution Result 103- Execution Result
104 The execution result is as follows, indicating that the precision comparison succeeded.104 The execution result is as follows, indicating that the precision comparison succeeded.
Mexamples/01_simd_cpp_api/00_introduction/02_matrix/README.md+2-2
@@ -8,5 +8,5 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-| [matmul_basic_api](./matmul_basic_api) | 基于静态Tensor编程范式实现矩阵乘计算 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |11+| [matmul_basic_api](./matmul_basic_api) | 基于静态Tensor编程范式实现矩阵乘计算 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
12-| [matmul_advanced_api](./matmul_advanced_api) | 基于Matmul高阶API实现矩阵乘计算 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |12+| [matmul_advanced_api](./matmul_advanced_api) | 基于Matmul高阶API实现矩阵乘计算 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/00_introduction/02_matrix/README_en.md+2-2
@@ -8,5 +8,5 @@ This section introduces matrix multiplication computation based on Ascend C.
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [matmul_basic_api](./matmul_basic_api) | Matrix multiplication implementation based on static Tensor programming paradigm | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [matmul_basic_api](./matmul_basic_api) | Matrix multiplication implementation based on static Tensor programming paradigm | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [matmul_advanced_api](./matmul_advanced_api) | Matrix multiplication implementation based on Matmul high-level API | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |12+| [matmul_advanced_api](./matmul_advanced_api) | Matrix multiplication implementation based on Matmul high-level API | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/00_introduction/02_matrix/matmul_advanced_api/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -119,7 +119,7 @@
119 | 选项 | 可选值 | 说明 |119 | 选项 | 可选值 | 说明 |
120 |------|--------|------|120 |------|--------|------|
121 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU 运行、NPU仿真 |121 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU 运行、NPU仿真 |
122- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |122+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
123 123 
124- 执行结果124- 执行结果
125 执行结果如下，说明精度对比成功。125 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/00_introduction/02_matrix/matmul_advanced_api/README_en.md+4-4
@@ -8,9 +8,9 @@ This sample uses the Matmul high-level API to implement matrix multiplication.
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -119,7 +119,7 @@ Execute the following steps in the root directory of this sample to build and ru
119 | Option | Available Values | Description |119 | Option | Available Values | Description |
120 |--------|------------------|-------------|120 |--------|------------------|-------------|
121 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU execution, NPU simulation |121 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU execution, NPU simulation |
122- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |122+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
123 123 
124- Execution Result124- Execution Result
125 The following execution result indicates successful precision comparison.125 The following execution result indicates successful precision comparison.
Mexamples/01_simd_cpp_api/00_introduction/02_matrix/matmul_basic_api/README.md+8-8
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -99,7 +99,7 @@
99 ```99 ```
100 例如搬运A矩阵时`{1, baseM, baseK, 0, K, baseM, 1, 0}`，将baseM×baseK的ND数据转为Nz格式。100 例如搬运A矩阵时`{1, baseM, baseK, 0, K, baseM, 1, 0}`，将baseM×baseK的ND数据转为Nz格式。
101 101 
102- **[`AscendC::LoadData2DParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)接口使用，描述Atlas A2 训练系列产品/Atlas A2 推理系列产品、Atlas A3 训练系列产品/Atlas A3 推理系列产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：102+ **[`AscendC::LoadData2DParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)接口使用，描述Atlas A2系列产品、Atlas A3系列产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：
103 103 
104 ```cpp104 ```cpp
105 struct LoadData2DParams {105 struct LoadData2DParams {
@@ -112,10 +112,10 @@
112 bool addrMode; // 地址更新方式，false=递增，true=递减，默认false112 bool addrMode; // 地址更新方式，false=递增，true=递减，默认false
113 };113 };
114 ```114 ```
115- 例如：Atlas A2 训练系列产品/Atlas A2 推理系列产品、Atlas A3 训练系列产品/Atlas A3 推理系列产品中，L0A上的排布格式为Zz，搬运A矩阵时`{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`；<br>115+ 例如：Atlas A2系列产品、Atlas A3系列产品中，L0A上的排布格式为Zz，搬运A矩阵时`{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`；<br>
116 搬运B矩阵时`ifTranspose=true`，完成Nz到Zn的转置搬运。116 搬运B矩阵时`ifTranspose=true`，完成Nz到Zn的转置搬运。
117 117 
118- **[`AscendC::LoadData2DParamsV2`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)接口使用，描述Ascend 950PR/Ascend 950DT产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：118+ **[`AscendC::LoadData2DParamsV2`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)接口使用，描述Ascend 950PR&950DT系列产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：
119 ```cpp119 ```cpp
120 struct LoadData2DParamsV2 {120 struct LoadData2DParamsV2 {
121 uint32_t mStartPosition; // M方向起始位置，单位：512B121 uint32_t mStartPosition; // M方向起始位置，单位：512B
@@ -128,7 +128,7 @@
128 uint8_t sid; // 预留，配置为0128 uint8_t sid; // 预留，配置为0
129 };129 };
130 ```130 ```
131- Ascend 950PR/Ascend 950DT产品中，L0A上的排布格式为Nz，搬运A矩阵时使用`{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}`，一次完成A矩阵Nz到Nz搬运；搬运B矩阵时使用`{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}`，一次完成B矩阵Nz到Zn搬运。131+ Ascend 950PR&950DT系列产品中，L0A上的排布格式为Nz，搬运A矩阵时使用`{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}`，一次完成A矩阵Nz到Nz搬运；搬运B矩阵时使用`{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}`，一次完成B矩阵Nz到Zn搬运。
132 132 
133 **[`AscendC::MmadParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — [`Mmad`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)接口使用，描述矩阵乘参数：133 **[`AscendC::MmadParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — [`Mmad`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)接口使用，描述矩阵乘参数：
134 134 
@@ -198,7 +198,7 @@
198 | 选项 | 可选值 | 说明 |198 | 选项 | 可选值 | 说明 |
199 |------|--------|------|199 |------|--------|------|
200 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |200 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |
201- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |201+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
202 202 
203- 执行结果 203- 执行结果
204 执行结果如下，说明精度对比成功。204 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/00_introduction/02_matrix/matmul_basic_api/README_en.md+8-8
@@ -8,9 +8,9 @@ This sample implements multi-core matrix multiplication computation based on the
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|-----------------------|10|---------|-----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -99,7 +99,7 @@ This sample implements multi-core matrix multiplication computation based on the
99 ```99 ```
100 For example, when transferring matrix A, `{1, baseM, baseK, 0, K, baseM, 1, 0}` converts baseM×baseK ND data to Nz format.100 For example, when transferring matrix A, `{1, baseM, baseK, 0, K, baseM, 1, 0}` converts baseM×baseK ND data to Nz format.
101 101 
102- **[`AscendC::LoadData2DParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — Used by [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md) interface, describes data transfer parameters for matrix A L1 to L0A and matrix B L1 to L0B in Atlas A2 Training Series Products/Atlas A2 Inference Series Products, Atlas A3 Training Series Products/Atlas A3 Inference Series Products:102+ **[`AscendC::LoadData2DParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — Used by [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md) interface, describes data transfer parameters for matrix A L1 to L0A and matrix B L1 to L0B in Atlas A2 products, Atlas A3 products:
103 103 
104 ```cpp104 ```cpp
105 struct LoadData2DParams {105 struct LoadData2DParams {
@@ -112,10 +112,10 @@ This sample implements multi-core matrix multiplication computation based on the
112 bool addrMode; // Address update mode, false=increment, true=decrement, default false112 bool addrMode; // Address update mode, false=increment, true=decrement, default false
113 };113 };
114 ```114 ```
115- For example: In Atlas A2 Training Series Products/Atlas A2 Inference Series Products, Atlas A3 Training Series Products/Atlas A3 Inference Series Products, the layout format on L0A is Zz. When transferring matrix A, use `{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`; <br>115+ For example: In Atlas A2 products, Atlas A3 products, the layout format on L0A is Zz. When transferring matrix A, use `{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`; <br>
116 When transferring matrix B, set `ifTranspose=true` to complete Nz to Zn transpose transfer.116 When transferring matrix B, set `ifTranspose=true` to complete Nz to Zn transpose transfer.
117 117 
118- **[`AscendC::LoadData2DParamsV2`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — Used by [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md) interface, describes data transfer parameters for matrix A L1 to L0A and matrix B L1 to L0B in Ascend 950PR/Ascend 950DT products:118+ **[`AscendC::LoadData2DParamsV2`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — Used by [`LoadData`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md) interface, describes data transfer parameters for matrix A L1 to L0A and matrix B L1 to L0B in Ascend 950PR&950DT products:
119 ```cpp119 ```cpp
120 struct LoadData2DParamsV2 {120 struct LoadData2DParamsV2 {
121 uint32_t mStartPosition; // M direction start position, unit: 512B121 uint32_t mStartPosition; // M direction start position, unit: 512B
@@ -128,7 +128,7 @@ This sample implements multi-core matrix multiplication computation based on the
128 uint8_t sid; // Reserved, set to 0128 uint8_t sid; // Reserved, set to 0
129 };129 };
130 ```130 ```
131- In Ascend 950PR/Ascend 950DT products, the layout format on L0A is Nz. When transferring matrix A, use `{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}` to complete A matrix Nz to Nz transfer in one operation; when transferring matrix B, use `{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}` to complete B matrix Nz to Zn transfer in one operation.131+ In Ascend 950PR&950DT products, the layout format on L0A is Nz. When transferring matrix A, use `{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}` to complete A matrix Nz to Nz transfer in one operation; when transferring matrix B, use `{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}` to complete B matrix Nz to Zn transfer in one operation.
132 132 
133 **[`AscendC::MmadParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — Used by [`Mmad`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md) interface, describes matrix multiplication parameters:133 **[`AscendC::MmadParams`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — Used by [`Mmad`](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md) interface, describes matrix multiplication parameters:
134 134 
@@ -198,7 +198,7 @@ Execute the following steps in the root directory of this sample to compile and
198 | Option | Available Values | Description |198 | Option | Available Values | Description |
199 |--------|------------------|-------------|199 |--------|------------------|-------------|
200 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |200 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
201- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |201+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
202 202 
203- Execution Result 203- Execution Result
204 The execution result is as follows, indicating successful precision comparison.204 The execution result is as follows, indicating successful precision comparison.
Mexamples/01_simd_cpp_api/00_introduction/03_fusion_operation/README.md+2-2
@@ -8,5 +8,5 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-| [matmul_leakyrelu_advanced_api](./matmul_leakyrelu_advanced_api) | 基于高阶API实现Matmul矩阵乘与LeakyRelu激活函数融合计算 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |11+| [matmul_leakyrelu_advanced_api](./matmul_leakyrelu_advanced_api) | 基于高阶API实现Matmul矩阵乘与LeakyRelu激活函数融合计算 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
12-| [matmul_leakyrelu_basic_api](./matmul_leakyrelu_basic_api) | 基于基础API实现Matmul矩阵乘与LeakyRelu激活函数融合计算 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |12+| [matmul_leakyrelu_basic_api](./matmul_leakyrelu_basic_api) | 基于基础API实现Matmul矩阵乘与LeakyRelu激活函数融合计算 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/00_introduction/03_fusion_operation/README_en.md+2-2
@@ -8,5 +8,5 @@ This example introduces custom fusion computation, including Cube and Vector fus
8 8 
9| Directory Name | Function Description | Supported Products |9| Directory Name | Function Description | Supported Products |
10| -------------- | -------------------- | --- |10| -------------- | -------------------- | --- |
11-| [matmul_leakyrelu_advanced_api](./matmul_leakyrelu_advanced_api) | Fused computation of Matmul matrix multiplication and LeakyRelu activation function based on high-level API | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [matmul_leakyrelu_advanced_api](./matmul_leakyrelu_advanced_api) | Fused computation of Matmul matrix multiplication and LeakyRelu activation function based on high-level API | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [matmul_leakyrelu_basic_api](./matmul_leakyrelu_basic_api) | Fused computation of Matmul matrix multiplication and LeakyRelu activation function based on basic API | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |12+| [matmul_leakyrelu_basic_api](./matmul_leakyrelu_basic_api) | Fused computation of Matmul matrix multiplication and LeakyRelu activation function based on basic API | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/00_introduction/03_fusion_operation/matmul_leakyrelu_advanced_api/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -154,7 +154,7 @@
154 | 选项 | 可选值 | 说明 |154 | 选项 | 可选值 | 说明 |
155 |------|--------|------|155 |------|--------|------|
156 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU 运行、NPU仿真 |156 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU 运行、NPU仿真 |
157- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |157+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
158 158 
159 > **说明：** 当 `CMAKE_ASC_ARCHITECTURES=dav-3510` 时，编译工程会自动为ASC编译添加 `-DENABLE_CV_COMM_VIA_SSBUF=true`，开启SSBUF通信；`dav-2201` 场景不会添加该编译宏。159 > **说明：** 当 `CMAKE_ASC_ARCHITECTURES=dav-3510` 时，编译工程会自动为ASC编译添加 `-DENABLE_CV_COMM_VIA_SSBUF=true`，开启SSBUF通信；`dav-2201` 场景不会添加该编译宏。
160 160 
Mexamples/01_simd_cpp_api/00_introduction/03_fusion_operation/matmul_leakyrelu_advanced_api/README_en.md+4-4
@@ -8,9 +8,9 @@ This example uses high-level APIs to implement fused computation of Matmul and L
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -154,7 +154,7 @@ Execute the following steps in the root directory of this example to build and r
154 | Option | Available Values | Description |154 | Option | Available Values | Description |
155 |--------|------------------|-------------|155 |--------|------------------|-------------|
156 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU run, NPU simulation |156 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU run, NPU simulation |
157- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |157+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
158 158 
159 > **Note:** When `CMAKE_ASC_ARCHITECTURES=dav-3510`, the build project automatically adds `-DENABLE_CV_COMM_VIA_SSBUF=true` for ASC compilation to enable SSBUF communication. The `dav-2201` scenario does not add this compilation macro.159 > **Note:** When `CMAKE_ASC_ARCHITECTURES=dav-3510`, the build project automatically adds `-DENABLE_CV_COMM_VIA_SSBUF=true` for ASC compilation to enable SSBUF communication. The `dav-2201` scenario does not add this compilation macro.
160 160 
Mexamples/01_simd_cpp_api/00_introduction/03_fusion_operation/matmul_leakyrelu_basic_api/README.md+9-9
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -119,7 +119,7 @@
119 ```119 ```
120 120 
121 > **注意**：L0A分形在不同产品上有差异：121 > **注意**：L0A分形在不同产品上有差异：
122- > - Ascend 950PR/Ascend 950DT产品：L0A的分形为Nz122+ > - Ascend 950PR&950DT系列产品：L0A的分形为Nz
123 > - Atlas A2/A3系列产品：L0A的分形为Zz123 > - Atlas A2/A3系列产品：L0A的分形为Zz
124 124 
125 核间同步与Vector侧数据流（1个Cube核的结果由2个Vector核处理）：125 核间同步与Vector侧数据流（1个Cube核的结果由2个Vector核处理）：
@@ -341,7 +341,7 @@
341 ```341 ```
342 例如搬运A矩阵时`{1, baseM, baseK, 0, K, baseM, 1, 0}`，将baseM×baseK的ND数据转为Nz格式。342 例如搬运A矩阵时`{1, baseM, baseK, 0, K, baseM, 1, 0}`，将baseM×baseK的ND数据转为Nz格式。
343 343 
344- **[AscendC::LoadData2DParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — `LoadData`接口使用，描述Atlas A2 训练系列产品/Atlas A2 推理系列产品、Atlas A3 训练系列产品/Atlas A3 推理系列产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：344+ **[AscendC::LoadData2DParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — `LoadData`接口使用，描述Atlas A2系列产品、Atlas A3系列产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：
345 ```cpp345 ```cpp
346 struct LoadData2DParams {346 struct LoadData2DParams {
347 int32_t startIndex; // 分形矩阵ID（0为第1个），单位：512B，[0, 65535]347 int32_t startIndex; // 分形矩阵ID（0为第1个），单位：512B，[0, 65535]
@@ -353,10 +353,10 @@
353 bool addrMode; // 地址更新方式，false=递增，true=递减，默认false353 bool addrMode; // 地址更新方式，false=递增，true=递减，默认false
354 };354 };
355 ```355 ```
356- 例如：Atlas A2 训练系列产品/Atlas A2 推理系列产品、Atlas A3 训练系列产品/Atlas A3 推理系列产品中，L0A上的排布格式为Zz，搬运A矩阵时`{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`；<br>356+ 例如：Atlas A2系列产品、Atlas A3系列产品中，L0A上的排布格式为Zz，搬运A矩阵时`{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`；<br>
357 搬运B矩阵时`ifTranspose=true`，完成Nz到Zn的转置搬运。357 搬运B矩阵时`ifTranspose=true`，完成Nz到Zn的转置搬运。
358 358 
359- **[AscendC::LoadData2DParamsV2](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D_V2.md)** — `LoadData`接口使用，描述Ascend 950PR/Ascend 950DT产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：359+ **[AscendC::LoadData2DParamsV2](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D_V2.md)** — `LoadData`接口使用，描述Ascend 950PR&950DT系列产品中A矩阵L1到L0A和B矩阵L1到L0B的数据搬运参数：
360 ```cpp360 ```cpp
361 struct LoadData2DParamsV2 {361 struct LoadData2DParamsV2 {
362 uint32_t mStartPosition; // M方向起始位置，单位：512B362 uint32_t mStartPosition; // M方向起始位置，单位：512B
@@ -369,7 +369,7 @@
369 uint8_t sid; // 预留，配置为0369 uint8_t sid; // 预留，配置为0
370 };370 };
371 ```371 ```
372- Ascend 950PR/Ascend 950DT产品中，L0A上的排布格式为Nz，搬运A矩阵时使用`{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}`，一次完成A矩阵Nz到Nz搬运；搬运B矩阵时使用`{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}`，一次完成B矩阵Nz到Zn搬运。372+ Ascend 950PR&950DT系列产品中，L0A上的排布格式为Nz，搬运A矩阵时使用`{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}`，一次完成A矩阵Nz到Nz搬运；搬运B矩阵时使用`{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}`，一次完成B矩阵Nz到Zn搬运。
373 373 
374 **[AscendC::MmadParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — `Mmad`接口使用，描述矩阵乘参数：374 **[AscendC::MmadParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — `Mmad`接口使用，描述矩阵乘参数：
375 ```cpp375 ```cpp
@@ -465,7 +465,7 @@
465 | 选项　　　　　 | 可选值　　　　　　　　　　　| 说明　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　 |465 | 选项　　　　　 | 可选值　　　　　　　　　　　| 说明　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　 |
466 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|466 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|
467 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真　　　　　　　　　　　　　　　　　　　　　　　　 |467 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真　　　　　　　　　　　　　　　　　　　　　　　　 |
468- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |468+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
469 469 
470- 执行结果470- 执行结果
471 471 
Mexamples/01_simd_cpp_api/00_introduction/03_fusion_operation/matmul_leakyrelu_basic_api/README_en.md+9-9
@@ -8,9 +8,9 @@ This example implements Matmul and LeakyRelu fusion computation based on the sta
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -119,7 +119,7 @@ This example implements Matmul and LeakyRelu fusion computation based on the sta
119 ```119 ```
120 120 
121 > **Note**: L0A fractal format differs across products:121 > **Note**: L0A fractal format differs across products:
122- > - Ascend 950PR/Ascend 950DT products: L0A fractal format is Nz122+ > - Ascend 950PR&950DT products: L0A fractal format is Nz
123 > - Atlas A2/A3 series products: L0A fractal format is Zz123 > - Atlas A2/A3 series products: L0A fractal format is Zz
124 124 
125 Inter-core synchronization and Vector side data flow (1 Cube core's result is processed by 2 Vector cores):125 Inter-core synchronization and Vector side data flow (1 Cube core's result is processed by 2 Vector cores):
@@ -341,7 +341,7 @@ This example implements Matmul and LeakyRelu fusion computation based on the sta
341 ```341 ```
342 For example, when transferring matrix A with `{1, baseM, baseK, 0, K, baseM, 1, 0}`, it converts baseM×baseK ND data to Nz format.342 For example, when transferring matrix A with `{1, baseM, baseK, 0, K, baseM, 1, 0}`, it converts baseM×baseK ND data to Nz format.
343 343 
344- **[AscendC::LoadData2DParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — Used by the `LoadData` interface to describe data transfer parameters for matrix A from L1 to L0A and matrix B from L1 to L0B in Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products:344+ **[AscendC::LoadData2DParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D.md)** — Used by the `LoadData` interface to describe data transfer parameters for matrix A from L1 to L0A and matrix B from L1 to L0B in Atlas A2 products and Atlas A3 products:
345 ```cpp345 ```cpp
346 struct LoadData2DParams {346 struct LoadData2DParams {
347 int32_t startIndex; // Fractal matrix ID (0 is the first), unit: 512B, [0, 65535]347 int32_t startIndex; // Fractal matrix ID (0 is the first), unit: 512B, [0, 65535]
@@ -353,10 +353,10 @@ This example implements Matmul and LeakyRelu fusion computation based on the sta
353 bool addrMode; // Address update mode, false=increment, true=decrement, default false353 bool addrMode; // Address update mode, false=increment, true=decrement, default false
354 };354 };
355 ```355 ```
356- For example: In Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, the layout format on L0A is Zz. When transferring matrix A, use `{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`;<br>356+ For example: In Atlas A2 products and Atlas A3 products, the layout format on L0A is Zz. When transferring matrix A, use `{0, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, 0, 0, false, 0}`;<br>
357 When transferring matrix B, set `ifTranspose=true` to complete Nz to Zn transpose transfer.357 When transferring matrix B, set `ifTranspose=true` to complete Nz to Zn transpose transfer.
358 358 
359- **[AscendC::LoadData2DParamsV2](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D_V2.md)** — Used by the `LoadData` interface to describe data transfer parameters for matrix A from L1 to L0A and matrix B from L1 to L0B in Ascend 950PR/Ascend 950DT products:359+ **[AscendC::LoadData2DParamsV2](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/cube_compute_load/LoadData_2D_V2.md)** — Used by the `LoadData` interface to describe data transfer parameters for matrix A from L1 to L0A and matrix B from L1 to L0B in Ascend 950PR&950DT products:
360 ```cpp360 ```cpp
361 struct LoadData2DParamsV2 {361 struct LoadData2DParamsV2 {
362 uint32_t mStartPosition; // Starting position in M direction, unit: 512B362 uint32_t mStartPosition; // Starting position in M direction, unit: 512B
@@ -369,7 +369,7 @@ This example implements Matmul and LeakyRelu fusion computation based on the sta
369 uint8_t sid; // Reserved, set to 0369 uint8_t sid; // Reserved, set to 0
370 };370 };
371 ```371 ```
372- In Ascend 950PR/Ascend 950DT products, the layout format on L0A is Nz. When transferring matrix A, use `{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}` to complete A matrix Nz to Nz transfer in one step; when transferring matrix B, use `{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}` to complete B matrix Nz to Zn transfer in one step.372+ In Ascend 950PR&950DT products, the layout format on L0A is Nz. When transferring matrix A, use `{0, 0, baseM / CUBE_BLOCK, baseK / CUBE_BLOCK, baseM / CUBE_BLOCK, baseM / CUBE_BLOCK, false, 0}` to complete A matrix Nz to Nz transfer in one step; when transferring matrix B, use `{0, 0, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, baseK / CUBE_BLOCK, baseN / CUBE_BLOCK, true, 0}` to complete B matrix Nz to Zn transfer in one step.
373 373 
374 **[AscendC::MmadParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — Used by the `Mmad` interface to describe matrix multiplication parameters:374 **[AscendC::MmadParams](../../../../../docs/zh/api/SIMD-API/basic_api/cube_compute_ISASI/mmad_compute/Mmad.md)** — Used by the `Mmad` interface to describe matrix multiplication parameters:
375 ```cpp375 ```cpp
@@ -465,7 +465,7 @@ Execute the following steps in the root directory of this example to compile and
465 | Option | Possible Values | Description |465 | Option | Possible Values | Description |
466 | -------| ----------------| --------------------------------------------------------------------------------------|466 | -------| ----------------| --------------------------------------------------------------------------------------|
467 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, NPU simulation |467 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, NPU simulation |
468- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |468+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
469 469 
470- Execution Result470- Execution Result
471 471 
Mexamples/01_simd_cpp_api/00_introduction/04_reg_compute/README.md+2-2
@@ -8,5 +8,5 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-| [add](./add) | 基于RegBase编程接口的Add样例 | Ascend 950PR/Ascend 950DT |11+| [add](./add) | 基于RegBase编程接口的Add样例 | Ascend 950PR&950DT系列产品 |
12-| [gelu](./gelu) | 基于 LocalMemAllocator 和 RegBase/VF 融合的 GELU 计算实现 | Ascend 950PR/Ascend 950DT |12+| [gelu](./gelu) | 基于 LocalMemAllocator 和 RegBase/VF 融合的 GELU 计算实现 | Ascend 950PR&950DT系列产品 |
Mexamples/01_simd_cpp_api/00_introduction/04_reg_compute/README_en.md+2-2
@@ -8,5 +8,5 @@ This example introduces the Vector computation functionality of RegBase programm
8 8 
9| Directory Name | Function Description | Supported Products |9| Directory Name | Function Description | Supported Products |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-| [add](./add) | Add example based on RegBase programming interface | Ascend 950PR/Ascend 950DT |11+| [add](./add) | Add example based on RegBase programming interface | Ascend 950PR&950DT products |
12-| [gelu](./gelu) | GELU computation example based on LocalMemAllocator and RegBase/VF fusion | Ascend 950PR/Ascend 950DT |12+| [gelu](./gelu) | GELU computation example based on LocalMemAllocator and RegBase/VF fusion | Ascend 950PR&950DT products |
Mexamples/01_simd_cpp_api/00_introduction/04_reg_compute/add/README.md+2-2
@@ -7,7 +7,7 @@
7 7 
8| 产品 | CANN软件版本 |8| 产品 | CANN软件版本 |
9|------|-------------|9|------|-------------|
10-| Ascend 950PR/Ascend 950DT | >= CANN 9.2.0 |10+| Ascend 950PR&950DT系列产品 | >= CANN 9.2.0 |
11 11 
12## 目录结构介绍12## 目录结构介绍
13```13```
@@ -84,7 +84,7 @@
84 | 选项 | 可选值 | 说明 |84 | 选项 | 可选值 | 说明 |
85 |------|--------|------|85 |------|--------|------|
86 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |86 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |
87- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU 架构：Ascend 950PR/Ascend 950DT |87+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU 架构：Ascend 950PR&950DT系列产品 |
88 88 
89- 执行结果 89- 执行结果
90 执行结果如下，说明精度对比成功。90 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/00_introduction/04_reg_compute/add/README_en.md+2-2
@@ -7,7 +7,7 @@ This example implements vector self-addition based on the RegBase programming pa
7 7 
8| Product | CANN Software Version |8| Product | CANN Software Version |
9|---------|----------------------|9|---------|----------------------|
10-| Ascend 950PR/Ascend 950DT | >= CANN 9.2.0 |10+| Ascend 950PR&950DT products | >= CANN 9.2.0 |
11 11 
12## Directory Structure12## Directory Structure
13```13```
@@ -84,7 +84,7 @@ Execute the following steps in the root directory of this example to build and r
84 | Option | Available Values | Description |84 | Option | Available Values | Description |
85 |--------|------------------|-------------|85 |--------|------------------|-------------|
86 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, NPU simulation |86 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, NPU simulation |
87- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU architecture: Ascend 950PR/Ascend 950DT |87+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU architecture: Ascend 950PR&950DT products |
88 88 
89- Execution Results 89- Execution Results
90 The execution results are as follows, indicating the precision comparison succeeded.90 The execution results are as follows, indicating the precision comparison succeeded.
Mexamples/01_simd_cpp_api/00_introduction/04_reg_compute/gelu/README.md+2-2
@@ -40,7 +40,7 @@ $$
40 40 
41| 产品 | CANN软件版本 |41| 产品 | CANN软件版本 |
42|------|-------------|42|------|-------------|
43-| Ascend 950PR/Ascend 950DT | >= CANN 9.2.0 |43+| Ascend 950PR&950DT系列产品 | >= CANN 9.2.0 |
44 44 
45## 算子实现45## 算子实现
46 46 
@@ -272,7 +272,7 @@ cat ./OPPROF_*/PipeUtilization.csv
272 | 选项 | 可选值 | 说明 |272 | 选项 | 可选值 | 说明 |
273 |------|--------|------|273 |------|--------|------|
274 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU运行、NPU仿真 |274 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU运行、NPU仿真 |
275- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510`（默认） | NPU架构：Ascend 950PR/Ascend 950DT |275+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510`（默认） | NPU架构：Ascend 950PR&950DT系列产品 |
276 | `CMAKE_VF_MODE` | `true`（默认）、`false` | VF融合模式：启用或禁用 `--cce-simd-vf-fusion` |276 | `CMAKE_VF_MODE` | `true`（默认）、`false` | VF融合模式：启用或禁用 `--cce-simd-vf-fusion` |
277 277 
278 278 
Mexamples/01_simd_cpp_api/00_introduction/04_reg_compute/gelu/README_en.md+2-2
@@ -40,7 +40,7 @@ The complete data path is: GM → UB → Register (step-by-step computation) →
40 40 
41| Product | CANN Software Version |41| Product | CANN Software Version |
42|------|-------------|42|------|-------------|
43-| Ascend 950PR/Ascend 950DT | >= CANN 9.2.0 |43+| Ascend 950PR&950DT products | >= CANN 9.2.0 |
44 44 
45## Operator Implementation45## Operator Implementation
46 46 
@@ -273,7 +273,7 @@ Execute the following steps in the root directory of this example to build and r
273 | Option | Available Values | Description |273 | Option | Available Values | Description |
274 |------|--------|------|274 |------|--------|------|
275 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU execution, NPU simulation |275 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU execution, NPU simulation |
276- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` (default) | NPU architecture: Ascend 950PR/Ascend 950DT |276+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` (default) | NPU architecture: Ascend 950PR&950DT products |
277 | `CMAKE_VF_MODE` | `true` (default), `false` | VF fusion mode: enable or disable `--cce-simd-vf-fusion` |277 | `CMAKE_VF_MODE` | `true` (default), `false` | VF fusion mode: enable or disable `--cce-simd-vf-fusion` |
278 278 
279 279 
Mexamples/01_simd_cpp_api/01_utilities/00_printf/README.md+2-2
@@ -5,5 +5,5 @@
5## 样例列表5## 样例列表
6| 目录名称 | 功能描述 | 支持的产品 |6| 目录名称 | 功能描述 | 支持的产品 |
7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
8-| [simple_printf](./simple_printf) | 使用静态Tensor编程模式实现矩阵乘法，展示AscendC::printf接口的基本使用方法 | Ascend 950PR/Ascend 950DT<br>Atlas A3训练系列产品/Atlas A3推理系列产品<br>Atlas A2训练系列产品/Atlas A2推理系列产品 |8+| [simple_printf](./simple_printf) | 使用静态Tensor编程模式实现矩阵乘法，展示AscendC::printf接口的基本使用方法 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
9-| [simd_vf_printf](./simd_vf_printf) | 使用vector编程模式，展示simd_vf侧printf接口的基本使用方法 | Ascend 950PR/Ascend 950DT |9+| [simd_vf_printf](./simd_vf_printf) | 使用vector编程模式，展示simd_vf侧printf接口的基本使用方法 | Ascend 950PR&950DT系列产品 |
Mexamples/01_simd_cpp_api/01_utilities/00_printf/README_en.md+2-2
@@ -8,5 +8,5 @@ This sample shows how to use the Ascend C printf interface and the simd_vf print
8 8 
9| Directory | Description | Supported Products |9| Directory | Description | Supported Products |
10| --- | --- | --- |10| --- | --- | --- |
11-| [simple_printf](./simple_printf) | Uses static Tensor programming mode to implement matrix multiplication, showing basic usage of the AscendC::printf interface | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series/Atlas A3 Inference Series<br>Atlas A2 Training Series/Atlas A2 Inference Series |11+| [simple_printf](./simple_printf) | Uses static Tensor programming mode to implement matrix multiplication, showing basic usage of the AscendC::printf interface | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [simd_vf_printf](./simd_vf_printf) | Uses vector programming mode, showing basic usage of the simd_vf printf interface | Ascend 950PR/Ascend 950DT |12+| [simd_vf_printf](./simd_vf_printf) | Uses vector programming mode, showing basic usage of the simd_vf printf interface | Ascend 950PR&950DT products |
Mexamples/01_simd_cpp_api/01_utilities/00_printf/simd_vf_printf/README.md+2-2
@@ -8,7 +8,7 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12 12 
13## 目录结构介绍13## 目录结构介绍
14 14 
@@ -70,7 +70,7 @@
70 70 
71 | 选项 | 可选值 | 说明 |71 | 选项 | 可选值 | 说明 |
72 |------|--------|------|72 |------|--------|------|
73- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510`（默认） | NPU架构：dav-3510对应Ascend 950PR/Ascend 950DT |73+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510`（默认） | NPU架构：dav-3510对应Ascend 950PR&950DT系列产品 |
74 74 
75- 执行结果75- 执行结果
76 执行日志中可以看到aicore侧和simd_vf侧printf输出。76 执行日志中可以看到aicore侧和simd_vf侧printf输出。
Mexamples/01_simd_cpp_api/01_utilities/00_printf/simd_vf_printf/README_en.md+2-2
@@ -8,7 +8,7 @@ This example introduces the usage of the printf API on the simd_vf side. The ker
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12 12 
13## Directory Structure13## Directory Structure
14 14 
@@ -72,7 +72,7 @@ Run the following steps in the root directory of this example to build and run t
72 72 
73 | Option | Values | Description |73 | Option | Values | Description |
74 |--------|--------|-------------|74 |--------|--------|-------------|
75- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` (default) | NPU architecture: dav-3510 corresponds to Ascend 950PR/Ascend 950DT |75+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` (default) | NPU architecture: dav-3510 corresponds to Ascend 950PR&950DT products |
76 76 
77- Execution result77- Execution result
78 The execution log shows printf output from both the AICore side and the simd_vf side.78 The execution log shows printf output from both the AICore side and the simd_vf side.
Mexamples/01_simd_cpp_api/01_utilities/00_printf/simple_printf/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -93,7 +93,7 @@
93 93 
94 | 选项 | 可选值 | 说明 |94 | 选项 | 可选值 | 说明 |
95 |------|--------|------|95 |------|--------|------|
96- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |96+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
97 97 
98- 执行结果98- 执行结果
99 最终执行结果如下，说明精度对比成功。99 最终执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/01_utilities/00_printf/simple_printf/README_en.md+4-4
@@ -8,9 +8,9 @@ This example introduces the usage of the printf API to print kernel function rel
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -92,7 +92,7 @@ Run the following steps in the root directory of this example to build and run t
92 92 
93 | Option | Values | Description |93 | Option | Values | Description |
94 |--------|--------|-------------|94 |--------|--------|-------------|
95- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |95+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
96 96 
97- Execution result97- Execution result
98 The final execution result is as follows, indicating successful accuracy comparison.98 The final execution result is as follows, indicating successful accuracy comparison.
Mexamples/01_simd_cpp_api/01_utilities/01_assert/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -91,7 +91,7 @@
91 91 
92 | 选项 | 可选值 | 说明 |92 | 选项 | 可选值 | 说明 |
93 |------|--------|------|93 |------|--------|------|
94- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |94+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
95 95 
96- 执行结果96- 执行结果
97 最终执行结果如下，说明精度对比成功。97 最终执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/01_utilities/01_assert/README_en.md+4-4
@@ -8,9 +8,9 @@ This example introduces the usage of the ascendc_assert assertion API. If the as
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -91,7 +91,7 @@ Run the following steps in the root directory of this example to build and run t
91 91 
92 | Option | Values | Description |92 | Option | Values | Description |
93 |--------|--------|-------------|93 |--------|--------|-------------|
94- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |94+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
95 95 
96- Execution result96- Execution result
97 The final execution result is as follows, indicating successful accuracy comparison.97 The final execution result is as follows, indicating successful accuracy comparison.
Mexamples/01_simd_cpp_api/01_utilities/02_dump/README.md+4-4
@@ -5,7 +5,7 @@
5## 样例列表5## 样例列表
6| 目录名称 | 功能描述 | 支持的产品 |6| 目录名称 | 功能描述 | 支持的产品 |
7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
8-| [simple_dump](./simple_dump) | 使用静态Tensor编程模式实现矩阵乘法，展示`asc_dump_gm`、`asc_dump_l1buf`、`asc_dump_cbuf`、`asc_dump_ubuf`接口的基本使用方法 | Ascend 950PR/Ascend 950DT<br>Atlas A3训练系列产品/Atlas A3推理系列产品<br>Atlas A2训练系列产品/Atlas A2推理系列产品 |8+| [simple_dump](./simple_dump) | 使用静态Tensor编程模式实现矩阵乘法，展示`asc_dump_gm`、`asc_dump_l1buf`、`asc_dump_cbuf`、`asc_dump_ubuf`接口的基本使用方法 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
9-| [simd_vf_dump](./simd_vf_dump) | 使用vector编程模式，展示simd_vf侧asc_dump_ubuf/asc_dump/asc_dump_reg接口的基本使用方法 | Ascend 950PR/Ascend 950DT |9+| [simd_vf_dump](./simd_vf_dump) | 使用vector编程模式，展示simd_vf侧asc_dump_ubuf/asc_dump/asc_dump_reg接口的基本使用方法 | Ascend 950PR&950DT系列产品 |
10-| [cube_buffer_dump](./cube_buffer_dump) | 以HiFloat8量化MatMul为背景，使用DumpTensor打印HiFloat8输入矩阵、Bias和per-channel scale在L1 Buffer、Bias Table Buffer和Fixpipe Buffer中的数据 | Ascend 950PR/Ascend 950DT |10+| [cube_buffer_dump](./cube_buffer_dump) | 以HiFloat8量化MatMul为背景，使用DumpTensor打印HiFloat8输入矩阵、Bias和per-channel scale在L1 Buffer、Bias Table Buffer和Fixpipe Buffer中的数据 | Ascend 950PR&950DT系列产品 |
11-| [python_exception_dump](./experimental/python_exception_dump) | NPU kernel运行时异常发生时，自动dump问题算子的异常上下文与kernel参数（args）用于离线问题定位，另提供纯Python的输入/输出tensor dump；外挂库方案，算子源码零改动 | Ascend 950PR/Ascend 950DT<br>Atlas A3训练系列产品/Atlas A3推理系列产品<br>Atlas A2训练系列产品/Atlas A2推理系列产品 |11+| [python_exception_dump](./experimental/python_exception_dump) | NPU kernel运行时异常发生时，自动dump问题算子的异常上下文与kernel参数（args）用于离线问题定位，另提供纯Python的输入/输出tensor dump；外挂库方案，算子源码零改动 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/01_utilities/02_dump/README_en.md+4-4
@@ -8,7 +8,7 @@ This sample shows how to use the Ascend C asc_dump interface series and the simd
8 8 
9| Directory | Description | Supported Products |9| Directory | Description | Supported Products |
10| --- | --- | --- |10| --- | --- | --- |
11-| [simple_dump](./simple_dump) | Uses static Tensor programming mode to implement matrix multiplication, showing basic usage of the `asc_dump_gm`, `asc_dump_l1buf`, `asc_dump_cbuf`, and `asc_dump_ubuf` interfaces | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series/Atlas A3 Inference Series<br>Atlas A2 Training Series/Atlas A2 Inference Series |11+| [simple_dump](./simple_dump) | Uses static Tensor programming mode to implement matrix multiplication, showing basic usage of the `asc_dump_gm`, `asc_dump_l1buf`, `asc_dump_cbuf`, and `asc_dump_ubuf` interfaces | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [simd_vf_dump](./simd_vf_dump) | Uses vector programming mode, showing basic usage of the simd_vf asc_dump_ubuf/asc_dump/asc_dump_reg interfaces | Ascend 950PR/Ascend 950DT |12+| [simd_vf_dump](./simd_vf_dump) | Uses vector programming mode, showing basic usage of the simd_vf asc_dump_ubuf/asc_dump/asc_dump_reg interfaces | Ascend 950PR&950DT products |
13-| [cube_buffer_dump](./cube_buffer_dump) | Uses HiFloat8 quantized MatMul as the context and DumpTensor to print the HiFloat8 input matrix, bias, and per-channel scale from L1 Buffer, Bias Table Buffer, and Fixpipe Buffer | Ascend 950PR/Ascend 950DT |13+| [cube_buffer_dump](./cube_buffer_dump) | Uses HiFloat8 quantized MatMul as the context and DumpTensor to print the HiFloat8 input matrix, bias, and per-channel scale from L1 Buffer, Bias Table Buffer, and Fixpipe Buffer | Ascend 950PR&950DT products |
14-| [python_exception_dump](./experimental/python_exception_dump) | When an NPU kernel hits a runtime exception, automatically dumps the exception context and kernel args of the failing operator for offline analysis, plus a pure-Python input/output tensor dump; plugin-library design with zero changes to operator source code | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series/Atlas A3 Inference Series<br>Atlas A2 Training Series/Atlas A2 Inference Series |14+| [python_exception_dump](./experimental/python_exception_dump) | When an NPU kernel hits a runtime exception, automatically dumps the exception context and kernel args of the failing operator for offline analysis, plus a pure-Python input/output tensor dump; plugin-library design with zero changes to operator source code | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/01_utilities/02_dump/cube_buffer_dump/README.md+1-1
@@ -9,7 +9,7 @@ scale在L1 Buffer、Bias Table Buffer和Fixpipe Buffer中的数据，展示dav-3
9 9 
10| 产品 | CANN软件版本 | HDK版本 |10| 产品 | CANN软件版本 | HDK版本 |
11|------|-------------|---------|11|------|-------------|---------|
12-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 | >= 25.7.RC1.6 |12+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 | >= 25.7.RC1.6 |
13 13 
14## 目录结构介绍14## 目录结构介绍
15 15 
Mexamples/01_simd_cpp_api/01_utilities/02_dump/cube_buffer_dump/README_en.md+1-1
@@ -10,7 +10,7 @@ Cube Buffer data printing on dav-3510.
10 10 
11| Product | CANN Software Version | HDK Version |11| Product | CANN Software Version | HDK Version |
12|---------|-----------------------|-------------|12|---------|-----------------------|-------------|
13-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 | >= 25.7.RC1.6 |13+| Ascend 950PR&950DT products | >= CANN 9.1.0 | >= 25.7.RC1.6 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
Mexamples/01_simd_cpp_api/01_utilities/02_dump/experimental/python_exception_dump/README.md+3-3
@@ -12,9 +12,9 @@
12 12 
13| 产品 | CANN软件版本 |13| 产品 | CANN软件版本 |
14|---|---|14|---|---|
15-| Ascend 950PR/Ascend 950DT | \>= CANN 9.1.0 |15+| Ascend 950PR&950DT系列产品 | \>= CANN 9.1.0 |
16-| Atlas A3训练系列产品/Atlas A3推理系列产品 | \>= CANN 9.1.0 |16+| Atlas A3系列产品 | \>= CANN 9.1.0 |
17-| Atlas A2训练系列产品/Atlas A2推理系列产品 | \>= CANN 9.1.0 |17+| Atlas A2系列产品 | \>= CANN 9.1.0 |
18 18 
19## 目录结构介绍19## 目录结构介绍
20 20 
Mexamples/01_simd_cpp_api/01_utilities/02_dump/experimental/python_exception_dump/README_en.md+3-3
@@ -12,9 +12,9 @@ This example demonstrates an NPU operator exception info auto-dump solution base
12 12 
13| Products | CANN Versions |13| Products | CANN Versions |
14|---|---|14|---|---|
15-| Ascend 950PR/Ascend 950DT | \>= CANN 9.1.0 |15+| Ascend 950PR&950DT products | \>= CANN 9.1.0 |
16-| Atlas A3 training series products/Atlas A3 inference series products | \>= CANN 9.1.0 |16+| Atlas A3 products | \>= CANN 9.1.0 |
17-| Atlas A2 training series products/Atlas A2 inference series products | \>= CANN 9.1.0 |17+| Atlas A2 products | \>= CANN 9.1.0 |
18 18 
19## Directory Structure19## Directory Structure
20 20 
Mexamples/01_simd_cpp_api/01_utilities/02_dump/simd_vf_dump/README.md+2-2
@@ -9,7 +9,7 @@
9 9 
10| 产品 | CANN软件版本 |10| 产品 | CANN软件版本 |
11|------|-------------|11|------|-------------|
12-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |12+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
13 13 
14## 目录结构介绍14## 目录结构介绍
15 15 
@@ -79,7 +79,7 @@
79 79 
80 | 选项 | 可选值 | 说明 |80 | 选项 | 可选值 | 说明 |
81 |------|--------|------|81 |------|--------|------|
82- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510`（默认） | NPU架构：dav-3510对应Ascend 950PR/Ascend 950DT |82+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510`（默认） | NPU架构：dav-3510对应Ascend 950PR&950DT系列产品 |
83 83 
84- 执行结果84- 执行结果
85 最终执行结果如下，说明精度对比成功。85 最终执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/01_utilities/02_dump/simd_vf_dump/README_en.md+2-2
@@ -8,7 +8,7 @@ This example introduces the usage of the asc_dump series APIs on the simd_vf sid
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12 12 
13## Directory Structure13## Directory Structure
14 14 
@@ -78,7 +78,7 @@ Run the following steps in the root directory of this example to build and run t
78 78 
79 | Option | Values | Description |79 | Option | Values | Description |
80 |--------|--------|-------------|80 |--------|--------|-------------|
81- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` (default) | NPU architecture: dav-3510 corresponds to Ascend 950PR/Ascend 950DT |81+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` (default) | NPU architecture: dav-3510 corresponds to Ascend 950PR&950DT products |
82 82 
83- Execution result83- Execution result
84 The final execution result is as follows, indicating successful accuracy comparison.84 The final execution result is as follows, indicating successful accuracy comparison.
Mexamples/01_simd_cpp_api/01_utilities/02_dump/simple_dump/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -94,7 +94,7 @@
94 94 
95 | 选项 | 可选值 | 说明 |95 | 选项 | 可选值 | 说明 |
96 |------|--------|------|96 |------|--------|------|
97- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |97+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
98 98 
99- 执行结果99- 执行结果
100 最终执行结果如下，说明精度对比成功。100 最终执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/01_utilities/02_dump/simple_dump/README_en.md+4-4
@@ -8,9 +8,9 @@ This example introduces the usage of the asc_dump_xxx series APIs to enable tens
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -94,7 +94,7 @@ Run the following steps in the root directory of this example to build and run t
94 94 
95 | Option | Values | Description |95 | Option | Values | Description |
96 |--------|--------|-------------|96 |--------|--------|-------------|
97- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |97+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
98 98 
99- Execution result99- Execution result
100 The final execution result is as follows, indicating successful accuracy comparison.100 The final execution result is as follows, indicating successful accuracy comparison.
Mexamples/01_simd_cpp_api/01_utilities/03_clock/README.md+2-2
@@ -8,7 +8,7 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.0.0-beta.2 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.0.0-beta.2 |
12 12 
13## 目录结构13## 目录结构
14 14 
@@ -108,7 +108,7 @@
108 | 选项 | 可选值 | 说明 |108 | 选项 | 可选值 | 说明 |
109 |------|--------|------|109 |------|--------|------|
110 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU 运行、NPU 仿真 |110 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`sim` | 运行模式：NPU 运行、NPU 仿真 |
111- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU 架构：`dav-3510` 对应 Ascend 950PR/Ascend 950DT |111+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU 架构：`dav-3510` 对应 Ascend 950PR&950DT系列产品 |
112 112 
113- 执行结果113- 执行结果
114 114 
Mexamples/01_simd_cpp_api/01_utilities/03_clock/README_en.md+2-2
@@ -8,7 +8,7 @@ This example demonstrates the usage of the `clock()` API for timestamp marking i
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.0.0-beta.2 |11+| Ascend 950PR&950DT products | >= CANN 9.0.0-beta.2 |
12 12 
13## Directory Structure13## Directory Structure
14 14 
@@ -108,7 +108,7 @@ Run the following steps in the root directory of this example to build and run t
108 | Option | Values | Description |108 | Option | Values | Description |
109 |--------|--------|-------------|109 |--------|--------|-------------|
110 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU execution, NPU simulation |110 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `sim` | Run mode: NPU execution, NPU simulation |
111- | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU architecture: `dav-3510` corresponds to Ascend 950PR/Ascend 950DT |111+ | `CMAKE_ASC_ARCHITECTURES` | `dav-3510` | NPU architecture: `dav-3510` corresponds to Ascend 950PR&950DT products |
112 112 
113- Execution result113- Execution result
114 114 
Mexamples/01_simd_cpp_api/01_utilities/04_profiling/README.md+2-2
@@ -5,5 +5,5 @@
5## 样例列表5## 样例列表
6| 目录名称 | 功能描述 | 支持的产品 |6| 目录名称 | 功能描述 | 支持的产品 |
7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
8-| [msprof](./msprof) | 本样例以Add计算为载体，展示如何通过`msopprof`采集上板性能数据。 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |8+| [msprof](./msprof) | 本样例以Add计算为载体，展示如何通过`msopprof`采集上板性能数据。 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
9-| [torch_library_report_tensor](./torch_library_report_tensor) | 本样例展示在`torch.library`自定义算子调用方式下，如何集成Profiling并在msprof结果中展示输入输出Tensor的Shape、数据类型和Format信息。 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |9+| [torch_library_report_tensor](./torch_library_report_tensor) | 本样例展示在`torch.library`自定义算子调用方式下，如何集成Profiling并在msprof结果中展示输入输出Tensor的Shape、数据类型和Format信息。 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/01_utilities/04_profiling/README_en.md+2-2
@@ -8,5 +8,5 @@ This sample demonstrates how to collect performance data using Profiling tools.
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [msprof](./msprof) | This sample uses Add as the workload to demonstrate how to collect on-device performance data with `msopprof`. | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [msprof](./msprof) | This sample uses Add as the workload to demonstrate how to collect on-device performance data with `msopprof`. | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [torch_library_report_tensor](./torch_library_report_tensor) | This sample demonstrates how to integrate Profiling in a `torch.library` custom operator call and display input/output Tensor shape, data type, and format information in msprof results. | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |12+| [torch_library_report_tensor](./torch_library_report_tensor) | This sample demonstrates how to integrate Profiling in a `torch.library` custom operator call and display input/output Tensor shape, data type, and format information in msprof results. | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/01_utilities/04_profiling/msprof/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -91,7 +91,7 @@ msOpProf工具是单算子性能分析工具，包含`msopprof`和`msopprof simu
91 91 
92 | 选项 | 可选值 | 说明 |92 | 选项 | 可选值 | 说明 |
93 |------|--------|------|93 |------|--------|------|
94- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：`dav-2201`对应Atlas A2训练系列产品/Atlas A2推理系列产品和Atlas A3训练系列产品/Atlas A3推理系列产品，`dav-3510`对应Ascend 950PR/Ascend 950DT |94+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：`dav-2201`对应Atlas A2系列产品和Atlas A3系列产品，`dav-3510`对应Ascend 950PR&950DT系列产品 |
95 95 
96- 执行结果96- 执行结果
97 97 
Mexamples/01_simd_cpp_api/01_utilities/04_profiling/msprof/README_en.md+4-4
@@ -8,9 +8,9 @@ This example uses Add computation as the carrier to demonstrate how to collect o
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -91,7 +91,7 @@ Run the following steps in the root directory of this example to build and run t
91 91 
92 | Option | Values | Description |92 | Option | Values | Description |
93 |--------|--------|-------------|93 |--------|--------|-------------|
94- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: `dav-2201` corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; `dav-3510` corresponds to Ascend 950PR/Ascend 950DT |94+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: `dav-2201` corresponds to Atlas A2 products and Atlas A3 products; `dav-3510` corresponds to Ascend 950PR&950DT products |
95 95 
96- Execution result96- Execution result
97 97 
Mexamples/01_simd_cpp_api/01_utilities/04_profiling/torch_library_report_tensor/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -126,7 +126,7 @@
126 126 
127 | 选项 | 可选值 | 说明 |127 | 选项 | 可选值 | 说明 |
128 |------|--------|------|128 |------|--------|------|
129- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |129+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
130 130 
131- 执行结果131- 执行结果
132 执行结果如下，说明Profiling数据采集和解析成功，其中DIR_NAME和PATH分别为落盘文件名和数据存放目录。132 执行结果如下，说明Profiling数据采集和解析成功，其中DIR_NAME和PATH分别为落盘文件名和数据存放目录。
Mexamples/01_simd_cpp_api/01_utilities/04_profiling/torch_library_report_tensor/README_en.md+4-4
@@ -8,9 +8,9 @@ This example demonstrates how to display operator input/output Shape, data type,
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -126,7 +126,7 @@ This example demonstrates how to display operator input/output Shape, data type,
126 126 
127 | Option | Values | Description |127 | Option | Values | Description |
128 |--------|--------|-------------|128 |--------|--------|-------------|
129- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |129+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
130 130 
131- Execution result131- Execution result
132 The execution result is as follows, indicating successful Profiling data collection and parsing, where DIR_NAME and PATH are the output file name and data storage directory respectively.132 The execution result is as follows, indicating successful Profiling data collection and parsing, where DIR_NAME and PATH are the output file name and data storage directory respectively.
Mexamples/01_simd_cpp_api/01_utilities/05_sanitizer/README.md+1-1
@@ -8,4 +8,4 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-| [ms_sanitizer](./ms_sanitizer) | 本样例通过Ascend C编程语言实现了Add样例，同时展示开发样例过程中，msSanitizer工具可以检测到哪些异常场景提前拦截 | Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |11+| [ms_sanitizer](./ms_sanitizer) | 本样例通过Ascend C编程语言实现了Add样例，同时展示开发样例过程中，msSanitizer工具可以检测到哪些异常场景提前拦截 | Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/01_utilities/05_sanitizer/README_en.md+1-1
@@ -8,4 +8,4 @@ This sample demonstrates the usage of the msSanitizer debugging tool based on Mi
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [ms_sanitizer](./ms_sanitizer) | This sample implements an Add example using the Ascend C programming language, demonstrating the abnormal scenarios that the msSanitizer tool can detect and intercept during sample development | Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [ms_sanitizer](./ms_sanitizer) | This sample implements an Add example using the Ascend C programming language, demonstrating the abnormal scenarios that the msSanitizer tool can detect and intercept during sample development | Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/01_utilities/05_sanitizer/ms_sanitizer/README.md+3-3
@@ -12,8 +12,8 @@ msSanitizer是基于昇腾AI处理器的异常检测工具，用于检测Ascend
12 12 
13| 产品 | CANN软件版本 |13| 产品 | CANN软件版本 |
14|------|-------------|14|------|-------------|
15-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |15+| Atlas A3系列产品 | >= CANN 9.0.0 |
16-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |16+| Atlas A2系列产品 | >= CANN 9.0.0 |
17 17 
18## 目录结构介绍18## 目录结构介绍
19 19 
@@ -182,7 +182,7 @@ z = x + y
182 182 
183 | 选项 | 可选值 | 说明 |183 | 选项 | 可选值 | 说明 |
184 |------|--------|------|184 |------|--------|------|
185- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` | NPU架构：`dav-2201`对应Atlas A2训练系列产品/Atlas A2推理系列产品和Atlas A3训练系列产品/Atlas A3推理系列产品 |185+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` | NPU架构：`dav-2201`对应Atlas A2系列产品和Atlas A3系列产品 |
186 186 
187- 执行结果187- 执行结果
188 188 
Mexamples/01_simd_cpp_api/01_utilities/05_sanitizer/ms_sanitizer/README_en.md+3-3
@@ -12,8 +12,8 @@ Refer to the "Environment Preparation" section in [Operator Development Tools](h
12 12 
13| Product | CANN Software Version |13| Product | CANN Software Version |
14|---------|----------------------|14|---------|----------------------|
15-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |15+| Atlas A3 products | >= CANN 9.0.0 |
16-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |16+| Atlas A2 products | >= CANN 9.0.0 |
17 17 
18## Directory Structure18## Directory Structure
19 19 
@@ -182,7 +182,7 @@ Run the following steps in the root directory of this example to build and run t
182 182 
183 | Option | Values | Description |183 | Option | Values | Description |
184 |--------|--------|-------------|184 |--------|--------|-------------|
185- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` | NPU architecture: `dav-2201` corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products |185+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` | NPU architecture: `dav-2201` corresponds to Atlas A2 products and Atlas A3 products |
186 186 
187- Execution result187- Execution result
188 188 
Mexamples/01_simd_cpp_api/01_utilities/06_cpu_debug/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -55,7 +55,7 @@
55 | 选项 | 可选值 | 说明 |55 | 选项 | 可选值 | 说明 |
56 |------|--------|------|56 |------|--------|------|
57 | `CMAKE_ASC_RUN_MODE` | `cpu` | 指定为`cpu`，开启CPU域编译 |57 | `CMAKE_ASC_RUN_MODE` | `cpu` | 指定为`cpu`，开启CPU域编译 |
58- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`、`dav-3510` | 指定NPU架构版本号，CMake会根据该值配置对应的CPU调试依赖库。`dav-2201`对应Atlas A2训练系列产品/Atlas A2推理系列产品和Atlas A3训练系列产品/Atlas A3推理系列产品，`dav-3510`对应Ascend 950PR/Ascend 950DT |58+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`、`dav-3510` | 指定NPU架构版本号，CMake会根据该值配置对应的CPU调试依赖库。`dav-2201`对应Atlas A2系列产品和Atlas A3系列产品，`dav-3510`对应Ascend 950PR&950DT系列产品 |
59 59 
60- 执行结果 60- 执行结果
61 执行结果如下，说明执行成功。61 执行结果如下，说明执行成功。
Mexamples/01_simd_cpp_api/01_utilities/06_cpu_debug/README_en.md+4-4
@@ -8,9 +8,9 @@ This example uses Add computation as the carrier to demonstrate the Ascend C CPU
8 8 
9| Product | CANN Software Version |9| Product | CANN Software Version |
10|---------|----------------------|10|---------|----------------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -56,7 +56,7 @@ Run the following steps in the root directory of this example to build and run t
56 | Option | Values | Description |56 | Option | Values | Description |
57 |--------|--------|-------------|57 |--------|--------|-------------|
58 | `CMAKE_ASC_RUN_MODE` | `cpu` | Set to `cpu` to enable CPU domain compilation |58 | `CMAKE_ASC_RUN_MODE` | `cpu` | Set to `cpu` to enable CPU domain compilation |
59- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`, `dav-3510` | Specify the NPU architecture version. CMake configures the corresponding CPU debug dependency libraries based on this value. `dav-2201` corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; `dav-3510` corresponds to Ascend 950PR/Ascend 950DT |59+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`, `dav-3510` | Specify the NPU architecture version. CMake configures the corresponding CPU debug dependency libraries based on this value. `dav-2201` corresponds to Atlas A2 products and Atlas A3 products; `dav-3510` corresponds to Ascend 950PR&950DT products |
60 60 
61- Execution result61- Execution result
62 The execution result is as follows, indicating successful execution.62 The execution result is as follows, indicating successful execution.
Mexamples/01_simd_cpp_api/01_utilities/08_simulator/README.md+7-7
@@ -9,9 +9,9 @@
9 9 
10| 产品 | CANN软件版本 |10| 产品 | CANN软件版本 |
11|------|-------------|11|------|-------------|
12-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |12+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
13-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A3系列产品 | >= CANN 9.0.0 |
14-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |14+| Atlas A2系列产品 | >= CANN 9.0.0 |
15 15 
16## 目录结构介绍16## 目录结构介绍
17 17 
@@ -86,7 +86,7 @@
86 | 选项 | 说明 |86 | 选项 | 说明 |
87 |------|------|87 |------|------|
88 | `CMAKE_ASC_RUN_MODE` | 指定为`sim`，开启NPU仿真模式 |88 | `CMAKE_ASC_RUN_MODE` | 指定为`sim`，开启NPU仿真模式 |
89- | `CMAKE_ASC_ARCHITECTURES` | 指定NPU架构版本号。`dav-2201`对应Atlas A2训练系列产品/Atlas A2推理系列产品和Atlas A3训练系列产品/Atlas A3推理系列产品，`dav-3510`对应Ascend 950PR/Ascend 950DT |89+ | `CMAKE_ASC_ARCHITECTURES` | 指定NPU架构版本号。`dav-2201`对应Atlas A2系列产品和Atlas A3系列产品，`dav-3510`对应Ascend 950PR&950DT系列产品 |
90 90 
91 精度对比成功时，输出如下：91 精度对比成功时，输出如下：
92 92 
@@ -104,12 +104,12 @@ msopprof simulator --soc-version=<soc_version> ./demo
104 104 
105> AI处理器的型号`<soc_version>`请通过如下方式获取：105> AI处理器的型号`<soc_version>`请通过如下方式获取：
106>106>
107->- Atlas A2 训练系列产品/Atlas A2 推理系列产品107+>- Atlas A2系列产品
108> - 针对以上产品型号：在安装昇腾AI处理器的服务器执行`npu-smi info`命令进行查询，获取**Name**信息。实际配置值为AscendName，例如**Name**取值为xxxyy，实际配置值为Ascendxxxyy。108> - 针对以上产品型号：在安装昇腾AI处理器的服务器执行`npu-smi info`命令进行查询，获取**Name**信息。实际配置值为AscendName，例如**Name**取值为xxxyy，实际配置值为Ascendxxxyy。
109>109>
110 110 
111-> - Ascend 950PR/Ascend 950DT111+> - Ascend 950PR&950DT系列产品
112-> - Atlas A3 训练系列产品/Atlas A3 推理系列产品112+> - Atlas A3系列产品
113> - 针对以上产品型号，在安装昇腾AI处理器的服务器执行`npu-smi info -t board -i <id> -c <chip_id>`命令进行查询，获取**Chip Name**和**NPU Name**信息，实际配置值为Chip Name_NPU Name。例如**Chip Name**取值为Ascendxxx，**NPU Name**取值为1234，实际配置值为Ascendxxx_1234。113> - 针对以上产品型号，在安装昇腾AI处理器的服务器执行`npu-smi info -t board -i <id> -c <chip_id>`命令进行查询，获取**Chip Name**和**NPU Name**信息，实际配置值为Chip Name_NPU Name。例如**Chip Name**取值为Ascendxxx，**NPU Name**取值为1234，实际配置值为Ascendxxx_1234。
114>114>
115> 其中，`id`为设备ID，通过`npu-smi info -l`命令查出的NPU ID即为设备ID；`chip_id`为芯片ID，通过`npu-smi info -m`命令查出的Chip ID即为芯片ID。115> 其中，`id`为设备ID，通过`npu-smi info -l`命令查出的NPU ID即为设备ID；`chip_id`为芯片ID，通过`npu-smi info -m`命令查出的Chip ID即为芯片ID。
Mexamples/01_simd_cpp_api/01_utilities/08_simulator/README_en.md+7-7
@@ -8,9 +8,9 @@ This sample uses MatmulLeakyRelu fusion computing as a carrier to demonstrate th
8 8 
9| Products | CANN Software Version |9| Products | CANN Software Version |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 training series products/Atlas A3 inference series products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 training series products/Atlas A2 inference series products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Introduction to directory structure15## Introduction to directory structure
16 16 
@@ -85,7 +85,7 @@ Perform the following steps in the root directory of this sample to compile and
85 | Options | Description |85 | Options | Description |
86 |------|------|86 |------|------|
87 | `CMAKE_ASC_RUN_MODE` | Specify `sim` to enable NPU simulation mode |87 | `CMAKE_ASC_RUN_MODE` | Specify `sim` to enable NPU simulation mode |
88- | `CMAKE_ASC_ARCHITECTURES` | Specifies the NPU architecture version number. `dav-2201` corresponds to Atlas A2 training series products/Atlas A2 reasoning series products and Atlas A3 training series products/Atlas A3 reasoning series products. `dav-3510` corresponds to Ascend 950PR/Ascend 950DT |88+ | `CMAKE_ASC_ARCHITECTURES` | Specifies the NPU architecture version number. `dav-2201` corresponds to Atlas A2 products and Atlas A3 products. `dav-3510` corresponds to Ascend 950PR&950DT products |
89 89 
90 When the accuracy comparison is successful, the output is as follows:90 When the accuracy comparison is successful, the output is as follows:
91 91 
@@ -103,12 +103,12 @@ msopprof simulator --soc-version=<soc_version> ./demo
103 103 
104> Please obtain the AI ​​processor model `<soc_version>` through the following methods:104> Please obtain the AI ​​processor model `<soc_version>` through the following methods:
105>105>
106->- Atlas A2 training series products/Atlas A2 inference series products106+>- Atlas A2 products
107> - For the above product models: Execute the `npu-smi info` command on the server where the Ascend AI processor is installed to query and obtain the **Name** information. The actual configuration value is AscendName. For example, the value of **Name** is xxxyy, and the actual configuration value is Ascendxxxyy.107> - For the above product models: Execute the `npu-smi info` command on the server where the Ascend AI processor is installed to query and obtain the **Name** information. The actual configuration value is AscendName. For example, the value of **Name** is xxxyy, and the actual configuration value is Ascendxxxyy.
108>108>
109 109 
110-> - Ascend 950PR/Ascend 950DT110+> - Ascend 950PR&950DT products
111-> - Atlas A3 training series products/Atlas A3 inference series products111+> - Atlas A3 products
112> - For the above product models, execute the `npu-smi info -t board -i <id> -c <chip_id>` command on the server where the Ascend AI processor is installed to query and obtain the **Chip Name** and **NPU Name** information. The actual configuration value is Chip Name_NPU Name. For example, the value of **Chip Name** is Ascendxxx, the value of **NPU Name** is 1234, and the actual configuration value is Ascendxxx_1234.112> - For the above product models, execute the `npu-smi info -t board -i <id> -c <chip_id>` command on the server where the Ascend AI processor is installed to query and obtain the **Chip Name** and **NPU Name** information. The actual configuration value is Chip Name_NPU Name. For example, the value of **Chip Name** is Ascendxxx, the value of **NPU Name** is 1234, and the actual configuration value is Ascendxxx_1234.
113>113>
114> Among them, `id` is the device ID, and the NPU ID found through the `npu-smi info -l` command is the device ID; `chip_id` is the chip ID, and the Chip ID found out through the `npu-smi info -m` command is the chip ID.114> Among them, `id` is the device ID, and the NPU ID found through the `npu-smi info -l` command is the device ID; `chip_id` is the chip ID, and the Chip ID found out through the `npu-smi info -m` command is the chip ID.
Mexamples/01_simd_cpp_api/01_utilities/09_log/README.md+4-4
@@ -10,9 +10,9 @@
10 10 
11| 产品 | CANN软件版本 |11| 产品 | CANN软件版本 |
12|------|-------------|12|------|-------------|
13-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |13+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
14-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |14+| Atlas A3系列产品 | >= CANN 9.0.0 |
15-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |15+| Atlas A2系列产品 | >= CANN 9.0.0 |
16 16 
17## 目录结构介绍17## 目录结构介绍
18 18 
@@ -122,7 +122,7 @@
122 | 选项 | 可选值 | 说明 |122 | 选项 | 可选值 | 说明 |
123 |------|--------|------|123 |------|--------|------|
124 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |124 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |
125- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：<br>dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>dav-3510 对应 Ascend 950PR/Ascend 950DT |125+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：<br>dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品<br>dav-3510 对应 Ascend 950PR&950DT系列产品 |
126 126 
127- 执行结果（NPU模式，以非打屏模式下为例，落盘文件详情可参考[Ascend 日志功能参考](https://hiascend.com/document/redirect/CannCommunitylogref)）127- 执行结果（NPU模式，以非打屏模式下为例，落盘文件详情可参考[Ascend 日志功能参考](https://hiascend.com/document/redirect/CannCommunitylogref)）
128 执行后会打印类似以下信息：128 执行后会打印类似以下信息：
Mexamples/01_simd_cpp_api/01_utilities/09_log/README_en.md+4-4
@@ -10,9 +10,9 @@ For more log function details, refer to: [Ascend Log Function Reference](https:/
10 10 
11| Product | CANN Software Version |11| Product | CANN Software Version |
12|---------|-----------------------|12|---------|-----------------------|
13-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |13+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
14-| Atlas A3 training series products/Atlas A3 inference series products | >= CANN 9.0.0 |14+| Atlas A3 products | >= CANN 9.0.0 |
15-| Atlas A2 training series products/Atlas A2 inference series products | >= CANN 9.0.0 |15+| Atlas A2 products | >= CANN 9.0.0 |
16 16 
17## Directory Structure17## Directory Structure
18 18 
@@ -122,7 +122,7 @@ Run the following steps in the root directory of this example to build and execu
122 | Option | Values | Description |122 | Option | Values | Description |
123 |------|--------|------|123 |------|--------|------|
124 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, NPU simulation |124 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, NPU simulation |
125- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture:<br>dav-2201 corresponds to Atlas A2 training series products/Atlas A2 inference series products and Atlas A3 training series products/Atlas A3 inference series products<br>dav-3510 corresponds to Ascend 950PR/Ascend 950DT |125+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture:<br>dav-2201 corresponds to Atlas A2 products and Atlas A3 products<br>dav-3510 corresponds to Ascend 950PR&950DT products |
126 126 
127- Execution Result (NPU mode, using non-screen-output mode as an example; for file storage details, refer to [Ascend Log Function Reference](https://hiascend.com/document/redirect/CannCommunitylogref))127- Execution Result (NPU mode, using non-screen-output mode as an example; for file storage details, refer to [Ascend Log Function Reference](https://hiascend.com/document/redirect/CannCommunitylogref))
128 After execution, the following information is printed:128 After execution, the following information is printed:
Mexamples/01_simd_cpp_api/01_utilities/README.md+9-9
@@ -5,12 +5,12 @@
5## 样例列表5## 样例列表
6| 目录名称 | 功能描述 | 支持的产品 |6| 目录名称 | 功能描述 | 支持的产品 |
7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |7| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
8-| [00_printf](./00_printf) | Printf接口打印功能 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |8+| [00_printf](./00_printf) | Printf接口打印功能 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
9-| [01_assert](./01_assert) | Assert接口异常检测功能 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |9+| [01_assert](./01_assert) | Assert接口异常检测功能 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
10-| [02_dump](./02_dump) | Dump数据调测功能 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |10+| [02_dump](./02_dump) | Dump数据调测功能 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
11-| [03_clock](./03_clock) | Kernel函数执行耗时打印 | Ascend 950PR/Ascend 950DT |11+| [03_clock](./03_clock) | Kernel函数执行耗时打印 | Ascend 950PR&950DT系列产品 |
12-| [04_profiling](./04_profiling) | 使用Profiling工具采集性能数据 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品|12+| [04_profiling](./04_profiling) | 使用Profiling工具采集性能数据 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品|
13-| [05_sanitizer](./05_sanitizer) | Sanitizer调测工具的使用方法 | Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |13+| [05_sanitizer](./05_sanitizer) | Sanitizer调测工具的使用方法 | Atlas A3系列产品<br>Atlas A2系列产品 |
14-| [06_cpu_debug](./06_cpu_debug) | CPU调测模式 | Ascend 950PR/Ascend 950DT<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品 |14+| [06_cpu_debug](./06_cpu_debug) | CPU调测模式 | Ascend 950PR&950DT系列产品<br>Atlas A2系列产品<br>Atlas A3系列产品 |
15-| [08_simulator](./08_simulator) | CAmodel仿真与问题分析 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |15+| [08_simulator](./08_simulator) | CAmodel仿真与问题分析 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
16-| [09_log](./09_log) | Kernel日志功能（日志打屏/落盘/级别控制） | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |16+| [09_log](./09_log) | Kernel日志功能（日志打屏/落盘/级别控制） | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/01_utilities/README_en.md+9-9
@@ -8,12 +8,12 @@ Introduction to common debugging interfaces and tools for Ascend C.
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [00_printf](./00_printf) | Printf interface printing function | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [00_printf](./00_printf) | Printf interface printing function | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [01_assert](./01_assert) | Assert interface exception detection function | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |12+| [01_assert](./01_assert) | Assert interface exception detection function | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
13-| [02_dump](./02_dump) | Dump data debugging function | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |13+| [02_dump](./02_dump) | Dump data debugging function | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
14-| [03_clock](./03_clock) | Kernel function execution time printing | Ascend 950PR/Ascend 950DT |14+| [03_clock](./03_clock) | Kernel function execution time printing | Ascend 950PR&950DT products |
15-| [04_profiling](./04_profiling) | Collecting performance data using Profiling tool | Ascend 950PR/Ascend 950DT<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products |15+| [04_profiling](./04_profiling) | Collecting performance data using Profiling tool | Ascend 950PR&950DT products<br>Atlas A2 products<br>Atlas A3 products |
16-| [05_sanitizer](./05_sanitizer) | Usage of Sanitizer debugging tool | Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |16+| [05_sanitizer](./05_sanitizer) | Usage of Sanitizer debugging tool | Atlas A3 products<br>Atlas A2 products |
17-| [06_cpu_debug](./06_cpu_debug) | CPU debugging mode | Ascend 950PR/Ascend 950DT<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products |17+| [06_cpu_debug](./06_cpu_debug) | CPU debugging mode | Ascend 950PR&950DT products<br>Atlas A2 products<br>Atlas A3 products |
18-| [08_simulator](./08_simulator) | CAmodel simulation and problem analysis | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |18+| [08_simulator](./08_simulator) | CAmodel simulation and problem analysis | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
19-| [09_log](./09_log) | Kernel log functions (log to screen/disk/level control) | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |19+| [09_log](./09_log) | Kernel log functions (log to screen/disk/level control) | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/README.md+3-3
@@ -10,6 +10,6 @@
10## 样例列表10## 样例列表
11| 目录名称 | 功能描述 | 支持的产品 |11| 目录名称 | 功能描述 | 支持的产品 |
12| ------------------------------------------------------------ | ---------------------------------------------------- | --- |12| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
13-| [pybind](./pybind) | 本样例展示如何使用pybind11注册自定义算子 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |13+| [pybind](./pybind) | 本样例展示如何使用pybind11注册自定义算子 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
14-| [torch_library](./torch_library) | 本样例展示如何使用PyTorch的torch.library机制注册自定义算子 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |14+| [torch_library](./torch_library) | 本样例展示如何使用PyTorch的torch.library机制注册自定义算子 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
15-| [ge_torchair](./ge_torchair) | 本样例展示Ascend C算子接入PyTorch图模式 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |15+| [ge_torchair](./ge_torchair) | 本样例展示Ascend C算子接入PyTorch图模式 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/README_en.md+3-3
@@ -14,6 +14,6 @@ This directory introduces different modes for integrating Ascend C custom operat
14 14 
15| Directory Name | Description | Supported Products |15| Directory Name | Description | Supported Products |
16| -------------- | ----------- | ------------------ |16| -------------- | ----------- | ------------------ |
17-| [pybind](./pybind) | This sample demonstrates how to register a custom operator using pybind11 | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |17+| [pybind](./pybind) | This sample demonstrates how to register a custom operator using pybind11 | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
18-| [torch_library](./torch_library) | This sample demonstrates how to register a custom operator using PyTorch's torch.library mechanism | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |18+| [torch_library](./torch_library) | This sample demonstrates how to register a custom operator using PyTorch's torch.library mechanism | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
19-| [ge_torchair](./ge_torchair) | This sample demonstrates how to integrate an Ascend C operator into PyTorch graph mode | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |19+| [ge_torchair](./ge_torchair) | This sample demonstrates how to integrate an Ascend C operator into PyTorch graph mode | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/ge_torchair/README.md+4-4
@@ -19,9 +19,9 @@
19 19 
20| 产品 | CANN软件版本 |20| 产品 | CANN软件版本 |
21|------|-------------|21|------|-------------|
22-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |22+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
23-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.1.0 |23+| Atlas A3系列产品 | >= CANN 9.1.0 |
24-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.1.0 |24+| Atlas A2系列产品 | >= CANN 9.1.0 |
25 25 
26## 目录结构介绍26## 目录结构介绍
27 27 
@@ -146,7 +146,7 @@ opt_model = torch.compile(model, fullgraph=True, backend=npu_backend, dynamic=Fa
146 146 
147 | 选项 | 可选值 | 说明 |147 | 选项 | 可选值 | 说明 |
148 |------|--------|------|148 |------|--------|------|
149- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：`dav-2201` 对应Atlas A2训练系列产品/Atlas A2推理系列产品与Atlas A3训练系列产品/Atlas A3推理系列产品，`dav-3510` 对应Ascend 950PR/Ascend 950DT |149+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：`dav-2201` 对应Atlas A2系列产品与Atlas A3系列产品，`dav-3510` 对应Ascend 950PR&950DT系列产品 |
150 150 
151- 执行结果151- 执行结果
152 152 
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/ge_torchair/README_en.md+4-4
@@ -19,9 +19,9 @@ This example uses the AddCustom operator to demonstrate how to integrate an Asce
19 19 
20| Product | CANN Version |20| Product | CANN Version |
21|---------|-------------|21|---------|-------------|
22-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |22+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
23-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.1.0 |23+| Atlas A3 products | >= CANN 9.1.0 |
24-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.1.0 |24+| Atlas A2 products | >= CANN 9.1.0 |
25 25 
26## Directory Structure26## Directory Structure
27 27 
@@ -146,7 +146,7 @@ The compiled model runs on the NPU side. When GE reaches the `AddCustom` node, i
146 146 
147 | Option | Values | Description |147 | Option | Values | Description |
148 |--------|--------|-------------|148 |--------|--------|-------------|
149- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: `dav-2201` corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; `dav-3510` corresponds to Ascend 950PR/Ascend 950DT |149+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: `dav-2201` corresponds to Atlas A2 products and Atlas A3 products; `dav-3510` corresponds to Ascend 950PR&950DT products |
150 150 
151- Execution result151- Execution result
152 152 
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/pybind/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.1.0 |12+| Atlas A3系列产品 | >= CANN 9.1.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.1.0 |13+| Atlas A2系列产品 | >= CANN 9.1.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -99,7 +99,7 @@
99 99 
100 | 选项　　　　　 | 可选值　　　　　　　　　　　| 说明　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　 |100 | 选项　　　　　 | 可选值　　　　　　　　　　　| 说明　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　 |
101 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|101 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|
102- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品 与 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |102+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品 与 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
103 103 
104- 执行结果104- 执行结果
105 105 
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/pybind/README_en.md+4-4
@@ -8,9 +8,9 @@ This example demonstrates how to register a custom operator using pybind11, base
8 8 
9| Product | CANN Version |9| Product | CANN Version |
10|---------|-------------|10|---------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.1.0 |12+| Atlas A3 products | >= CANN 9.1.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.1.0 |13+| Atlas A2 products | >= CANN 9.1.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -99,7 +99,7 @@ This example demonstrates how to register a custom operator using pybind11, base
99 99 
100 | Option | Values | Description |100 | Option | Values | Description |
101 | -------| -------| ------------|101 | -------| -------| ------------|
102- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |102+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
103 103 
104- Execution result104- Execution result
105 105 
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/torch_library/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.1.0 |12+| Atlas A3系列产品 | >= CANN 9.1.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.1.0 |13+| Atlas A2系列产品 | >= CANN 9.1.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -104,7 +104,7 @@
104 104 
105 | 选项 | 可选值 | 说明 |105 | 选项 | 可选值 | 说明 |
106 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|106 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|
107- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品 与 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |107+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品 与 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
108 108 
109- 执行结果109- 执行结果
110 110 
Mexamples/01_simd_cpp_api/02_features/00_framework/00_pytorch/torch_library/README_en.md+4-4
@@ -8,9 +8,9 @@ This example demonstrates how to register a custom operator using the torch.libr
8 8 
9| Product | CANN Version |9| Product | CANN Version |
10|---------|-------------|10|---------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.1.0 |12+| Atlas A3 products | >= CANN 9.1.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.1.0 |13+| Atlas A2 products | >= CANN 9.1.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -104,7 +104,7 @@ This example demonstrates how to register a custom operator using the torch.libr
104 104 
105 | Option | Values | Description |105 | Option | Values | Description |
106 | -------| -------| ------------|106 | -------| -------| ------------|
107- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |107+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
108 108 
109- Execution result109- Execution result
110 110 
Mexamples/01_simd_cpp_api/02_features/00_framework/01_tensorflow/README.md+2-2
@@ -8,5 +8,5 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| --------- | --------- | --- |10| --------- | --------- | --- |
11-| [tensorflow_builtin](./tensorflow_builtin) | 样例展示将Ascend C自定义算子映射到TensorFlow内置算子，并通过TensorFlow调用 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |11+| [tensorflow_builtin](./tensorflow_builtin) | 样例展示将Ascend C自定义算子映射到TensorFlow内置算子，并通过TensorFlow调用 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
12-| [tensorflow_custom](./tensorflow_custom) | 样例展示将Ascend C自定义算子映射到TensorFlow自定义算子，并通过TensorFlow调用 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |12+| [tensorflow_custom](./tensorflow_custom) | 样例展示将Ascend C自定义算子映射到TensorFlow自定义算子，并通过TensorFlow调用 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/02_features/00_framework/01_tensorflow/README_en.md+2-2
@@ -8,5 +8,5 @@ This sample demonstrates the implementation methods for custom operators in the
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [tensorflow_builtin](./tensorflow_builtin) | This sample demonstrates mapping Ascend C custom operators to TensorFlow built-in operators and invoking them through TensorFlow | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [tensorflow_builtin](./tensorflow_builtin) | This sample demonstrates mapping Ascend C custom operators to TensorFlow built-in operators and invoking them through TensorFlow | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-| [tensorflow_custom](./tensorflow_custom) | This sample demonstrates mapping Ascend C custom operators to TensorFlow custom operators and invoking them through TensorFlow | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |12+| [tensorflow_custom](./tensorflow_custom) | This sample demonstrates mapping Ascend C custom operators to TensorFlow custom operators and invoking them through TensorFlow | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/02_features/00_framework/01_tensorflow/tensorflow_builtin/README.md+3-3
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
Mexamples/01_simd_cpp_api/02_features/00_framework/01_tensorflow/tensorflow_builtin/README_en.md+3-3
@@ -8,9 +8,9 @@ This example demonstrates how to map an Ascend C custom operator to a TensorFlow
8 8 
9| Product | CANN Version |9| Product | CANN Version |
10|---------|-------------|10|---------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
Mexamples/01_simd_cpp_api/02_features/00_framework/01_tensorflow/tensorflow_custom/README.md+3-3
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
Mexamples/01_simd_cpp_api/02_features/00_framework/01_tensorflow/tensorflow_custom/README_en.md+3-3
@@ -8,9 +8,9 @@ This example demonstrates how to map the Ascend C custom operator Add computatio
8 8 
9| Product | CANN Version |9| Product | CANN Version |
10|---------|-------------|10|---------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
Mexamples/01_simd_cpp_api/02_features/00_framework/02_onnx/README.md+1-1
@@ -8,4 +8,4 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| --------- | --------- | --- |10| --------- | --------- | --- |
11-| [onnx_plugin](./onnx_plugin) | 样例介绍通过ONNX网络调用的方式调用自定义算子 | Ascend 950PR/Ascend 950DT<br>Atlas A3 训练系列产品/Atlas A3 推理系列产品<br>Atlas A2 训练系列产品/Atlas A2 推理系列产品 |11+| [onnx_plugin](./onnx_plugin) | 样例介绍通过ONNX网络调用的方式调用自定义算子 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/02_features/00_framework/02_onnx/README_en.md+1-1
@@ -8,4 +8,4 @@ This example introduces the implementation method of custom operators for the ON
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| -------------- | ----------- | --- |10| -------------- | ----------- | --- |
11-| [onnx_plugin](./onnx_plugin) | Example demonstrating how to invoke custom operators through ONNX network calls | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+| [onnx_plugin](./onnx_plugin) | Example demonstrating how to invoke custom operators through ONNX network calls | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/02_features/00_framework/02_onnx/onnx_plugin/README.md+9-9
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |12+| Atlas A3系列产品 | >= CANN 9.0.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |13+| Atlas A2系列产品 | >= CANN 9.0.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -67,18 +67,18 @@
67 67 
68 > AI处理器的型号<soc_version>请通过如下方式获取：68 > AI处理器的型号<soc_version>请通过如下方式获取：
69 > - 针对如下产品型号：在安装昇腾AI处理器的服务器执行`npu-smi info`命令进行查询，获取**Name**信息。实际配置值为AscendName，例如**Name**取值为xxxyy，实际配置值为Ascendxxxyy。69 > - 针对如下产品型号：在安装昇腾AI处理器的服务器执行`npu-smi info`命令进行查询，获取**Name**信息。实际配置值为AscendName，例如**Name**取值为xxxyy，实际配置值为Ascendxxxyy。
70- > - Atlas A2 训练系列产品 / Atlas A2 推理系列产品70+ > - Atlas A2系列产品
71- > - Atlas 200I/500 A2 推理产品71+ > - Atlas 200I/500 A2推理产品
72- > - Atlas 推理系列产品72+ > - Atlas推理系列产品
73- > - Atlas 训练系列产品73+ > - Atlas训练系列产品
74 >74 >
75 > - 针对如下产品型号，在安装昇腾AI处理器的服务器执行`npu-smi info -t board -i <id> -c <chip_id>`命令进行查询，获取**Chip Name**和**NPU Name**信息，实际配置值为Chip Name_NPU Name。例如**Chip Name**取值为Ascendxxx，**NPU Name**取值为1234，实际配置值为Ascendxxx_1234。其中：75 > - 针对如下产品型号，在安装昇腾AI处理器的服务器执行`npu-smi info -t board -i <id> -c <chip_id>`命令进行查询，获取**Chip Name**和**NPU Name**信息，实际配置值为Chip Name_NPU Name。例如**Chip Name**取值为Ascendxxx，**NPU Name**取值为1234，实际配置值为Ascendxxx_1234。其中：
76 >76 >
77 > id：设备id，通过`npu-smi info -l`命令查出的NPU ID即为设备id77 > id：设备id，通过`npu-smi info -l`命令查出的NPU ID即为设备id
78 >78 >
79 > chip_id：芯片id，通过`npu-smi info -m`命令查出的Chip ID即为芯片id79 > chip_id：芯片id，通过`npu-smi info -m`命令查出的Chip ID即为芯片id
80- > - Ascend 950PR/Ascend 950DT80+ > - Ascend 950PR&950DT系列产品
81- > - Atlas A3 训练系列产品 / Atlas A3 推理系列产品81+ > - Atlas A3系列产品
82 >82 >
83 > 基于同系列的AI处理器型号创建的样例工程，其基础功能（基于该工程进行样例开发、编译和部署）通用。83 > 基于同系列的AI处理器型号创建的样例工程，其基础功能（基于该工程进行样例开发、编译和部署）通用。
84 84 
Mexamples/01_simd_cpp_api/02_features/00_framework/02_onnx/onnx_plugin/README_en.md+6-6
@@ -8,9 +8,9 @@ This example uses the LeakyRelu computation to demonstrate how to invoke a custo
8 8 
9| Product | CANN Version |9| Product | CANN Version |
10|---------|-------------|10|---------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
12-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |12+| Atlas A3 products | >= CANN 9.0.0 |
13-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |13+| Atlas A2 products | >= CANN 9.0.0 |
14 14 
15## Directory Structure15## Directory Structure
16 16 
@@ -67,7 +67,7 @@ This example uses the `aclmdlExecute` interface to perform model inference. The
67 67 
68 > Obtain the AI processor model `<soc_version>` as follows:68 > Obtain the AI processor model `<soc_version>` as follows:
69 > - For the following product models: Run the `npu-smi info` command on the server with the Ascend AI processor installed to query the **Name** information. The actual configuration value is AscendName. For example, if the **Name** value is xxxyy, the actual configuration value is Ascendxxxyy.69 > - For the following product models: Run the `npu-smi info` command on the server with the Ascend AI processor installed to query the **Name** information. The actual configuration value is AscendName. For example, if the **Name** value is xxxyy, the actual configuration value is Ascendxxxyy.
70- > - Atlas A2 Training Series Products / Atlas A2 Inference Series Products70+ > - Atlas A2 products
71 > - Atlas 200I/500 A2 Inference Products71 > - Atlas 200I/500 A2 Inference Products
72 > - Atlas Inference Series Products72 > - Atlas Inference Series Products
73 > - Atlas Training Series Products73 > - Atlas Training Series Products
@@ -77,8 +77,8 @@ This example uses the `aclmdlExecute` interface to perform model inference. The
77 > id: device ID, obtained from the NPU ID queried by the `npu-smi info -l` command77 > id: device ID, obtained from the NPU ID queried by the `npu-smi info -l` command
78 >78 >
79 > chip_id: chip ID, obtained from the Chip ID queried by the `npu-smi info -m` command79 > chip_id: chip ID, obtained from the Chip ID queried by the `npu-smi info -m` command
80- > - Ascend 950PR/Ascend 950DT80+ > - Ascend 950PR&950DT products
81- > - Atlas A3 Training Series Products / Atlas A3 Inference Series Products81+ > - Atlas A3 products
82 >82 >
83 > Example projects created based on AI processor models of the same series share the same basic functionality (example development, compilation, and deployment based on the project).83 > Example projects created based on AI processor models of the same series share the same basic functionality (example development, compilation, and deployment based on the project).
84 84 
Mexamples/01_simd_cpp_api/02_features/00_framework/03_ge/triple_chevron_notation/README.md+4-4
@@ -20,9 +20,9 @@
20 20 
21| 产品 | CANN软件版本 |21| 产品 | CANN软件版本 |
22|------|-------------|22|------|-------------|
23-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |23+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
24-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.1.0 |24+| Atlas A3系列产品 | >= CANN 9.1.0 |
25-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.1.0 |25+| Atlas A2系列产品 | >= CANN 9.1.0 |
26 26 
27## 目录结构介绍27## 目录结构介绍
28 28 
@@ -109,7 +109,7 @@ const auto runRet = session.RunGraph(GRAPH_ID, inputs, outputs);
109 109 
110 | 选项　　　　　 | 可选值　　　　　　　　　　　| 说明　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　 |110 | 选项　　　　　 | 可选值　　　　　　　　　　　| 说明　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　 |
111 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|111 | ----------------| -----------------------------| --------------------------------------------------------------------------------------|
112- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品 与 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |112+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品 与 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
113 113 
114- 执行结果114- 执行结果
115 115 
Mexamples/01_simd_cpp_api/02_features/00_framework/03_ge/triple_chevron_notation/README_en.md+4-4
@@ -20,9 +20,9 @@ The figure above shows the complete integration pipeline from left to right. The
20 20 
21| Product | CANN Version |21| Product | CANN Version |
22|---------|-------------|22|---------|-------------|
23-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |23+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
24-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.1.0 |24+| Atlas A3 products | >= CANN 9.1.0 |
25-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.1.0 |25+| Atlas A2 products | >= CANN 9.1.0 |
26 26 
27## Directory Structure27## Directory Structure
28 28 
@@ -109,7 +109,7 @@ const auto runRet = session.RunGraph(GRAPH_ID, inputs, outputs);
109 109 
110 | Option | Values | Description |110 | Option | Values | Description |
111 | -------| -------| ------------|111 | -------| -------| ------------|
112- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |112+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
113 113 
114- Execution result114- Execution result
115 115 
Mexamples/01_simd_cpp_api/02_features/00_framework/04_aclgraph/README.md+4-4
@@ -12,9 +12,9 @@ ACLGraph 可以先捕获指定流上的一组运行时任务，再将这些任
12 12 
13| 产品 | CANN软件版本 |13| 产品 | CANN软件版本 |
14|------|-------------|14|------|-------------|
15-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |15+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
16-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.1.0 |16+| Atlas A3系列产品 | >= CANN 9.1.0 |
17-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.1.0 |17+| Atlas A2系列产品 | >= CANN 9.1.0 |
18 18 
19## 目录结构介绍19## 目录结构介绍
20 20 
@@ -136,7 +136,7 @@ for (int i = 0; i < 5; ++i) {
136 136 
137 | 选项 | 可选值 | 说明 |137 | 选项 | 可选值 | 说明 |
138 |------|--------|------|138 |------|--------|------|
139- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品 与 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |139+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品 与 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
140 | `SCENARIO_NUM` | `1`（默认）、`2` | 捕获场景编号：1表示单流捕获，2表示双流捕获 |140 | `SCENARIO_NUM` | `1`（默认）、`2` | 捕获场景编号：1表示单流捕获，2表示双流捕获 |
141 141 
142- 执行结果142- 执行结果
Mexamples/01_simd_cpp_api/02_features/00_framework/04_aclgraph/README_en.md+4-4
@@ -12,9 +12,9 @@ This example uses the Add operator to show how to capture Ascend C `<<<>>>` kern
12 12 
13| Product | CANN Version |13| Product | CANN Version |
14|---------|-------------|14|---------|-------------|
15-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |15+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
16-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.1.0 |16+| Atlas A3 products | >= CANN 9.1.0 |
17-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.1.0 |17+| Atlas A2 products | >= CANN 9.1.0 |
18 18 
19## Directory Structure19## Directory Structure
20 20 
@@ -137,7 +137,7 @@ The example also calls `aclmdlRIDebugJsonPrint` to output the captured task info
137 137 
138 | Option | Values | Description |138 | Option | Values | Description |
139 |--------|--------|-------------|139 |--------|--------|-------------|
140- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products and Atlas A3 Training Series Products/Atlas A3 Inference Series Products; dav-3510 corresponds to Ascend 950PR/Ascend 950DT |140+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products and Atlas A3 products; dav-3510 corresponds to Ascend 950PR&950DT products |
141 | `SCENARIO_NUM` | `1` (default), `2` | Capture scenario number: 1 indicates single-stream capture, and 2 indicates double-stream capture |141 | `SCENARIO_NUM` | `1` (default), `2` | Capture scenario number: 1 indicates single-stream capture, and 2 indicates double-stream capture |
142 142 
143- Execution result143- Execution result
Mexamples/01_simd_cpp_api/02_features/02_tiling_selector/vector_tiling_strategy/README.md+4-4
@@ -10,9 +10,9 @@
10 10 
11| 产品 | CANN软件版本 |11| 产品 | CANN软件版本 |
12|------|--------------|12|------|--------------|
13-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |13+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
14-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |14+| Atlas A3系列产品 | >= CANN 9.0.0 |
15-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |15+| Atlas A2系列产品 | >= CANN 9.0.0 |
16 16 
17## 目录结构介绍17## 目录结构介绍
18 18 
@@ -216,7 +216,7 @@
216 | 选项 | 可选值 | 说明 |216 | 选项 | 可选值 | 说明 |
217 |------|--------|------|217 |------|--------|------|
218 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |218 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU 运行、CPU调试、NPU仿真 |
219- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |219+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
220 | `SCENARIO_NUM` | `0`（默认）、`1`、`2`、`3` | 场景编号：0（主块均分）、1（尾块均分）、2（尾核切分）、3（尾核尾块切分） |220 | `SCENARIO_NUM` | `0`（默认）、`1`、`2`、`3` | 场景编号：0（主块均分）、1（尾块均分）、2（尾核切分）、3（尾核尾块切分） |
221 221 
222- 执行结果222- 执行结果
Mexamples/01_simd_cpp_api/02_features/02_tiling_selector/vector_tiling_strategy/README_en.md+4-4
@@ -10,9 +10,9 @@ The sample uses 8 Cores. The main tile data volume in each core is 3200 `half` e
10 10 
11| Product | CANN software version |11| Product | CANN software version |
12|------|--------------|12|------|--------------|
13-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |13+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
14-| Atlas A3 training series products/Atlas A3 inference series products | >= CANN 9.0.0 |14+| Atlas A3 products | >= CANN 9.0.0 |
15-| Atlas A2 training series products/Atlas A2 inference series products | >= CANN 9.0.0 |15+| Atlas A2 products | >= CANN 9.0.0 |
16 16 
17## Directory Structure17## Directory Structure
18 18 
@@ -208,7 +208,7 @@ In the sample root directory, perform the following steps to build and run the s
208 | Option | Optional value | Description |208 | Option | Optional value | Description |
209 |------|--------|------|209 |------|--------|------|
210 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, and NPU simulation |210 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debugging, and NPU simulation |
211- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: `dav-2201` maps to Atlas A2 training series products, Atlas A2 inference series products, Atlas A3 training series products, and Atlas A3 inference series products. `dav-3510` maps to Ascend 950PR and Ascend 950DT |211+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: `dav-2201` maps to Atlas A2 products and Atlas A3 products. `dav-3510` maps to Ascend 950PR and Ascend 950DT |
212 | `SCENARIO_NUM` | `0` (default), `1`, `2`, `3` | Scenario number: 0 (main tile equal split), 1 (tail block equal split), 2 (tail core split), and 3 (tail core split with tail blocks) |212 | `SCENARIO_NUM` | `0` (default), `1`, `2`, `3` | Scenario number: 0 (main tile equal split), 1 (tail block equal split), 2 (tail core split), and 3 (tail core split with tail blocks) |
213 213 
214- Execution result.214- Execution result.
Mexamples/01_simd_cpp_api/02_features/04_compile/00_basic_compile/README.md+4-4
@@ -15,9 +15,9 @@
15 15 
16| 产品 | CANN软件版本 |16| 产品 | CANN软件版本 |
17|------|-------------|17|------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
19-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |19+| Atlas A3系列产品 | >= CANN 9.0.0 |
20-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |20+| Atlas A2系列产品 | >= CANN 9.0.0 |
21 21 
22## 目录结构介绍22## 目录结构介绍
23 23 
@@ -119,7 +119,7 @@ target_compile_options(demo PRIVATE
119 | 选项 | 可选值 | 说明 |119 | 选项 | 可选值 | 说明 |
120 |------|--------|------|120 |------|--------|------|
121 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |121 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |
122- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |122+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
123 123 
124- 执行结果124- 执行结果
125 执行结果如下，说明精度对比成功。125 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/02_features/04_compile/00_basic_compile/README_en.md+4-4
@@ -15,9 +15,9 @@ Figure 1: Code structure and call relationship
15 15 
16| Product | CANN Version |16| Product | CANN Version |
17|---------|-------------|17|---------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
19-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |19+| Atlas A3 products | >= CANN 9.0.0 |
20-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |20+| Atlas A2 products | >= CANN 9.0.0 |
21 21 
22## Directory Structure22## Directory Structure
23 23 
@@ -122,7 +122,7 @@ Run the following steps in the root directory of this example to build and run t
122 | Option | Values | Description |122 | Option | Values | Description |
123 |--------|--------|-------------|123 |--------|--------|-------------|
124 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |124 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
125- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products/Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |125+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products/Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
126 126 
127- Execution result127- Execution result
128 128 
Mexamples/01_simd_cpp_api/02_features/04_compile/01_separate_compile/README.md+4-4
@@ -15,9 +15,9 @@
15 15 
16| 产品 | CANN软件版本 |16| 产品 | CANN软件版本 |
17|------|-------------|17|------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
19-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |19+| Atlas A3系列产品 | >= CANN 9.0.0 |
20-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |20+| Atlas A2系列产品 | >= CANN 9.0.0 |
21 21 
22## 目录结构介绍22## 目录结构介绍
23 23 
@@ -124,7 +124,7 @@ target_compile_options(demo PRIVATE
124 | 选项 | 可选值 | 说明 |124 | 选项 | 可选值 | 说明 |
125 |------|--------|------|125 |------|--------|------|
126 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |126 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |
127- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |127+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
128 128 
129- 执行结果129- 执行结果
130 执行结果如下，说明精度对比成功。130 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/02_features/04_compile/01_separate_compile/README_en.md+4-4
@@ -15,9 +15,9 @@ Figure 1: Code structure and call relationship
15 15 
16| Product | CANN Version |16| Product | CANN Version |
17|---------|-------------|17|---------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
19-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |19+| Atlas A3 products | >= CANN 9.0.0 |
20-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |20+| Atlas A2 products | >= CANN 9.0.0 |
21 21 
22## Directory Structure22## Directory Structure
23 23 
@@ -126,7 +126,7 @@ Run the following steps in the root directory of this example to build and run t
126 | Option | Values | Description |126 | Option | Values | Description |
127 |--------|--------|-------------|127 |--------|--------|-------------|
128 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |128 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
129- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products/Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |129+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products/Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
130 130 
131- Execution result131- Execution result
132 132 
Mexamples/01_simd_cpp_api/02_features/04_compile/02_dynamic_library_compile/README.md+4-4
@@ -15,9 +15,9 @@
15 15 
16| 产品 | CANN软件版本 |16| 产品 | CANN软件版本 |
17|------|-------------|17|------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
19-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |19+| Atlas A3系列产品 | >= CANN 9.0.0 |
20-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |20+| Atlas A2系列产品 | >= CANN 9.0.0 |
21 21 
22## 目录结构介绍22## 目录结构介绍
23 23 
@@ -95,7 +95,7 @@ add_library(add_kernel SHARED
95 | 选项 | 可选值 | 说明 |95 | 选项 | 可选值 | 说明 |
96 |------|--------|------|96 |------|--------|------|
97 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |97 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |
98- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |98+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
99 99 
100- 执行结果100- 执行结果
101 执行结果如下，说明精度对比成功。101 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/02_features/04_compile/02_dynamic_library_compile/README_en.md+4-4
@@ -15,9 +15,9 @@ Figure 1: Code structure and call relationship
15 15 
16| Product | CANN Version |16| Product | CANN Version |
17|---------|-------------|17|---------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
19-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |19+| Atlas A3 products | >= CANN 9.0.0 |
20-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |20+| Atlas A2 products | >= CANN 9.0.0 |
21 21 
22## Directory Structure22## Directory Structure
23 23 
@@ -97,7 +97,7 @@ Run the following steps in the root directory of this example to build and run t
97 | Option | Values | Description |97 | Option | Values | Description |
98 |--------|--------|-------------|98 |--------|--------|-------------|
99 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |99 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
100- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products/Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |100+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products/Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
101 101 
102- Execution result102- Execution result
103 103 
Mexamples/01_simd_cpp_api/02_features/04_compile/03_static_library_compile/README.md+4-4
@@ -15,9 +15,9 @@
15 15 
16| 产品 | CANN软件版本 |16| 产品 | CANN软件版本 |
17|------|-------------|17|------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
19-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.0.0 |19+| Atlas A3系列产品 | >= CANN 9.0.0 |
20-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.0.0 |20+| Atlas A2系列产品 | >= CANN 9.0.0 |
21 21 
22## 目录结构介绍22## 目录结构介绍
23 23 
@@ -95,7 +95,7 @@ add_library(add_kernel STATIC
95 | 选项 | 可选值 | 说明 |95 | 选项 | 可选值 | 说明 |
96 |------|--------|------|96 |------|--------|------|
97 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |97 | `CMAKE_ASC_RUN_MODE` | `npu`（默认）、`cpu`、`sim` | 运行模式：NPU运行、CPU调试、NPU仿真 |
98- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |98+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
99 99 
100- 执行结果100- 执行结果
101 执行结果如下，说明精度对比成功。101 执行结果如下，说明精度对比成功。
Mexamples/01_simd_cpp_api/02_features/04_compile/03_static_library_compile/README_en.md+4-4
@@ -15,9 +15,9 @@ Figure 1: Code structure and call relationship
15 15 
16| Product | CANN Version |16| Product | CANN Version |
17|---------|-------------|17|---------|-------------|
18-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |18+| Ascend 950PR&950DT products | >= CANN 9.1.0 |
19-| Atlas A3 Training Series Products/Atlas A3 Inference Series Products | >= CANN 9.0.0 |19+| Atlas A3 products | >= CANN 9.0.0 |
20-| Atlas A2 Training Series Products/Atlas A2 Inference Series Products | >= CANN 9.0.0 |20+| Atlas A2 products | >= CANN 9.0.0 |
21 21 
22## Directory Structure22## Directory Structure
23 23 
@@ -97,7 +97,7 @@ Run the following steps in the root directory of this example to build and run t
97 | Option | Values | Description |97 | Option | Values | Description |
98 |--------|--------|-------------|98 |--------|--------|-------------|
99 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |99 | `CMAKE_ASC_RUN_MODE` | `npu` (default), `cpu`, `sim` | Run mode: NPU execution, CPU debug, NPU simulation |
100- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 Training Series Products/Atlas A2 Inference Series Products/Atlas A3 Training Series Products/Atlas A3 Inference Series Products, dav-3510 corresponds to Ascend 950PR/Ascend 950DT |100+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201` (default), `dav-3510` | NPU architecture: dav-2201 corresponds to Atlas A2 products/Atlas A3 products, dav-3510 corresponds to Ascend 950PR&950DT products |
101 101 
102- Execution result102- Execution result
103 103 
Mexamples/01_simd_cpp_api/02_features/05_aclrtc/README.md+2-2
@@ -8,5 +8,5 @@
8 8 
9| 目录名称 | 功能描述 | 支持的产品 |9| 目录名称 | 功能描述 | 支持的产品 |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-|[rtc_hello_world](./rtc_hello_world) | 本样例介绍了使用aclrtc接口实现HelloWorld核函数的调用。 | Ascend 950PR/Ascend 950DT<br>Atlas A3训练系列产品/Atlas A3推理系列产品<br>Atlas A2训练系列产品/Atlas A2推理系列产品 |11+|[rtc_hello_world](./rtc_hello_world) | 本样例介绍了使用aclrtc接口实现HelloWorld核函数的调用。 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
12-|[rtc_template_add](./rtc_template_add) | 基于aclrtc运行时编译 + LocalMemAllocator（静态Tensor）实现Add。 | Ascend 950PR/Ascend 950DT<br>Atlas A3训练系列产品/Atlas A3推理系列产品<br>Atlas A2训练系列产品/Atlas A2推理系列产品 |12+|[rtc_template_add](./rtc_template_add) | 基于aclrtc运行时编译 + LocalMemAllocator（静态Tensor）实现Add。 | Ascend 950PR&950DT系列产品<br>Atlas A3系列产品<br>Atlas A2系列产品 |
Mexamples/01_simd_cpp_api/02_features/05_aclrtc/README_en.md+2-2
@@ -8,5 +8,5 @@ This example demonstrates using aclrtc (runtime compilation) interfaces to compi
8 8 
9| Directory Name | Description | Supported Products |9| Directory Name | Description | Supported Products |
10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |10| ------------------------------------------------------------ | ---------------------------------------------------- | --- |
11-|[rtc_hello_world](./rtc_hello_world) | This example demonstrates calling HelloWorld kernel function using aclrtc interfaces. | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |11+|[rtc_hello_world](./rtc_hello_world) | This example demonstrates calling HelloWorld kernel function using aclrtc interfaces. | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
12-|[rtc_template_add](./rtc_template_add) | Implements Add using aclrtc runtime compilation + LocalMemAllocator (static Tensor) | Ascend 950PR/Ascend 950DT<br>Atlas A3 Training Series Products/Atlas A3 Inference Series Products<br>Atlas A2 Training Series Products/Atlas A2 Inference Series Products |12+|[rtc_template_add](./rtc_template_add) | Implements Add using aclrtc runtime compilation + LocalMemAllocator (static Tensor) | Ascend 950PR&950DT products<br>Atlas A3 products<br>Atlas A2 products |
Mexamples/01_simd_cpp_api/02_features/05_aclrtc/rtc_hello_world/README.md+4-4
@@ -8,9 +8,9 @@
8 8 
9| 产品 | CANN软件版本 |9| 产品 | CANN软件版本 |
10|------|-------------|10|------|-------------|
11-| Ascend 950PR/Ascend 950DT | >= CANN 9.1.0 |11+| Ascend 950PR&950DT系列产品 | >= CANN 9.1.0 |
12-| Atlas A3 训练系列产品/Atlas A3 推理系列产品 | >= CANN 9.1.0 |12+| Atlas A3系列产品 | >= CANN 9.1.0 |
13-| Atlas A2 训练系列产品/Atlas A2 推理系列产品 | >= CANN 9.1.0 |13+| Atlas A2系列产品 | >= CANN 9.1.0 |
14 14 
15## 目录结构介绍15## 目录结构介绍
16 16 
@@ -71,7 +71,7 @@
71 71 
72 | 选项 | 可选值 | 说明 |72 | 选项 | 可选值 | 说明 |
73 |------|--------|------|73 |------|--------|------|
74- | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2 训练系列产品/Atlas A2 推理系列产品和 Atlas A3 训练系列产品/Atlas A3 推理系列产品，dav-3510 对应 Ascend 950PR/Ascend 950DT |74+ | `CMAKE_ASC_ARCHITECTURES` | `dav-2201`（默认）、`dav-3510` | NPU 架构：dav-2201 对应 Atlas A2系列产品和 Atlas A3系列产品，dav-3510 对应 Ascend 950PR&950DT系列产品 |
75 75 
76 > [!WARNING] 注意76 > [!WARNING] 注意
77 Aclrtc仅支持NPU运行模式，不支持cpu调试或sim仿真模式。77 Aclrtc仅支持NPU运行模式，不支持cpu调试或sim仿真模式。
Mexamples/04_aicpu/README.md+2-2
Mexamples/README.md+3-3
Mexamples/README_en.md+2-2