revert add_example-ops-transformer-AtomGit

Mexamples/add_example/README.md+1-1

	@@ -66,4 +66,4 @@ $$
66		66
67	\| 调用方式 \| 调用样例 \| 说明 \|	67	\| 调用方式 \| 调用样例 \| 说明 \|
68	\|--------------\|------------------------------------------------------------------------\|--------------------------------------------------------------\|	68	\|--------------\|------------------------------------------------------------------------\|--------------------------------------------------------------\|
69	-\| 图模式调用 \| [test_geir_add_example](./examples/test_geir_add_example.cpp) \| 通过[算子IR](./op_graph/add_example_proto.h)构图方式调用AddExample算子。 \|	69	+\| aclnn调用 \| [test_aclnn_add_example](./examples/test_aclnn_add_example.cpp) \| 通过直调的方式调用AddExample算子。 \|

Mexamples/add_example/examples/test_aclnn_add_example.cpp+4-4

Dexamples/add_example/examples/test_geir_add_example.cpp+0-300

	@@ -1,300 +0,0 @@
1	-/**
2	- * Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	- * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	- * CANN Open Software License Agreement Version 2.0 (the "License").
5	- * Please refer to the License for details. You may not use this file except in compliance with the License.
6	- * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	- * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	- * See LICENSE in the root of the software repository for the full text of the License.
9	- */
10	-
11	-#include <iostream>
12	-#include <fstream>
13	-#include <string.h>
14	-#include <stdint.h>
15	-#include <vector>
16	-#include <string>
17	-#include <map>
18	-#include "assert.h"
19	-
20	-#include "graph.h"
21	-#include "types.h"
22	-#include "tensor.h"
23	-#include "ge_error_codes.h"
24	-#include "ge_api_types.h"
25	-#include "ge_api.h"
26	-#include "array_ops.h"
27	-#include "ge_ir_build.h"
28	-
29	-#include "experiment_ops.h"
30	-#include "nn_other.h"
31	-#include "../op_graph/add_example_proto.h"
32	-
33	-#define FAILED -1
34	-#define SUCCESS 0
35	-
36	-using namespace ge;
37	-using std::map;
38	-using std::string;
39	-using std::vector;
40	-#define ADD_INPUT(intputIndex, intputName, intputDtype, inputShape) \
41	- vector<int64_t> placeholder##intputIndex##_shape = inputShape; \
42	- auto placeholder##intputIndex = op::Data("placeholder" + intputIndex).set_attr_index(0); \
43	- TensorDesc placeholder##intputIndex##_desc = \
44	- TensorDesc(ge::Shape(placeholder##intputIndex##_shape), FORMAT_ND, intputDtype); \
45	- placeholder##intputIndex##_desc.SetPlacement(ge::kPlacementHost); \
46	- placeholder##intputIndex##_desc.SetFormat(FORMAT_ND); \
47	- Tensor tensor_placeholder##intputIndex; \
48	- ret = GenOnesData(placeholder##intputIndex##_shape, \
49	- tensor_placeholder##intputIndex, \
50	- placeholder##intputIndex##_desc, \
51	- intputDtype, \
52	- 2); \
53	- if (ret != SUCCESS) { \
54	- printf("%s - ERROR - [XIR]: Generate input data failed\n", GetTime().c_str()); \
55	- return FAILED; \
56	- } \
57	- placeholder##intputIndex.update_input_desc_x(placeholder##intputIndex##_desc); \
58	- input.push_back(tensor_placeholder##intputIndex); \
59	- graph.AddOp(placeholder##intputIndex); \
60	- add1.set_input_##intputName(placeholder##intputIndex); \
61	- inputs.push_back(placeholder##intputIndex);
62	-
63	-#define ADD_CONST_INPUT(intputIndex, intputName, intputDtype, inputShape) \
64	- vector<int64_t> placeholder##intputIndex##_shape = inputShape; \
65	- auto placeholder##intputIndex = op::Const("placeholder" + intputIndex); \
66	- TensorDesc placeholder##intputIndex##_desc = \
67	- TensorDesc(ge::Shape(placeholder##intputIndex##_shape), FORMAT_ND, intputDtype); \
68	- placeholder##intputIndex##_desc.SetPlacement(ge::kPlacementHost); \
69	- placeholder##intputIndex##_desc.SetFormat(FORMAT_ND); \
70	- Tensor tensor_placeholder##intputIndex; \
71	- ret = GenOnesData(placeholder##intputIndex##_shape, \
72	- tensor_placeholder##intputIndex, \
73	- placeholder##intputIndex##_desc, \
74	- intputDtype, \
75	- 2); \
76	- if (ret != SUCCESS) { \
77	- printf("%s - ERROR - [XIR]: Generate input data failed\n", GetTime().c_str()); \
78	- return FAILED; \
79	- } \
80	- placeholder##intputIndex.SetAttr("value", tensor_placeholder##intputIndex); \
81	- placeholder##intputIndex.update_output_desc_y(placeholder##intputIndex##_desc); \
82	- graph.AddOp(placeholder##intputIndex); \
83	- add1.set_input_##intputName(placeholder##intputIndex); \
84	- add1.update_input_desc_##intputName(placeholder##intputIndex##_desc); \
85	- inputs.push_back(placeholder##intputIndex);
86	-
87	-#define ADD_OUTPUT(outputIndex, outputName, outputDtype, outputShape) \
88	- TensorDesc outputName##outputIndex##_desc = \
89	- TensorDesc(ge::Shape(outputShape), FORMAT_ND, outputDtype); \
90	- add1.update_output_desc_##outputName(outputName##outputIndex##_desc);
91	-
92	-string GetTime()
93	-{
94	- time_t timep;
95	- time(&timep);
96	- char tmp[64];
97	- strftime(tmp, sizeof(tmp), "%Y-%m-%d %H:%M:%S,000", localtime(&timep));
98	- return tmp;
99	-}
100	-
101	-uint32_t GetDataTypeSize(DataType dt)
102	-{
103	- uint32_t dilation = 1;
104	- uint32_t oneByte = 1;
105	- uint32_t twoByte = 2;
106	- uint32_t fourByte = 4;
107	- uint32_t eightByte = 8;
108	-
109	- if (dt == ge::DT_FLOAT) {
110	- dilation = fourByte;
111	- } else if (dt == ge::DT_FLOAT16) {
112	- dilation = twoByte;
113	- } else if (dt == ge::DT_BF16) {
114	- dilation = twoByte;
115	- } else if (dt == ge::DT_INT16) {
116	- dilation = twoByte;
117	- } else if (dt == ge::DT_UINT16) {
118	- dilation = twoByte;
119	- } else if (dt == ge::DT_INT32) {
120	- dilation = fourByte;
121	- } else if (dt == ge::DT_UINT32) {
122	- dilation = fourByte;
123	- } else if (dt == ge::DT_INT64) {
124	- dilation = eightByte;
125	- } else if (dt == ge::DT_UINT64) {
126	- dilation = eightByte;
127	- } else if (dt == ge::DT_INT8) {
128	- dilation = oneByte;
129	- }
130	- return dilation;
131	-}
132	-
133	-int32_t GenOnesDataFloat32(vector<int64_t> shapes, Tensor &input_tensor, TensorDesc &input_tensor_desc, float value)
134	-{
135	- input_tensor_desc.SetRealDimCnt(shapes.size());
136	- size_t size = 1;
137	- for (uint32_t i = 0; i < shapes.size(); i++) {
138	- size *= shapes[i];
139	- }
140	- uint32_t byteSizeFloat32 = 4;
141	- uint32_t data_len = size * byteSizeFloat32;
142	- float *pData = new (std::nothrow) float[size];
143	-
144	- for (size_t i = 0; i < size; ++i) {
145	- *(pData + i) = value;
146	- }
147	- input_tensor = Tensor(input_tensor_desc, (uint8_t *)pData, data_len);
148	- return SUCCESS;
149	-}
150	-
151	-int32_t GenOnesData(
152	- vector<int64_t> shapes, Tensor &input_tensor, TensorDesc &input_tensor_desc, DataType data_type, int value)
153	-{
154	- input_tensor_desc.SetRealDimCnt(shapes.size());
155	- size_t size = 1;
156	- for (uint32_t i = 0; i < shapes.size(); i++) {
157	- size *= shapes[i];
158	- }
159	- uint32_t data_len = size * GetDataTypeSize(data_type);
160	- int32_t *pData = new (std::nothrow) int32_t[data_len];
161	- for (uint32_t i = 0; i < size; ++i) {
162	- *(pData + i) = value;
163	- }
164	- input_tensor = Tensor(input_tensor_desc, reinterpret_cast<uint8_t *>(pData), data_len);
165	- return SUCCESS;
166	-}
167	-
168	-int32_t WriteDataToFile(string bin_file, uint64_t data_size, uint8_t *inputData)
169	-{
170	- FILE *fp;
171	- fp = fopen(bin_file.c_str(), "w");
172	- fwrite(inputData, sizeof(uint8_t), data_size, fp);
173	- fclose(fp);
174	- return SUCCESS;
175	-}
176	-
177	-int CreateOppInGraph(DataType inDtype, std::vector<ge::Tensor> &input, std::vector<Operator> &inputs,
178	- std::vector<Operator> &outputs, Graph &graph)
179	-{
180	- Status ret = SUCCESS;
181	- // 自定义代码：添加单算子定义到图中
182	- auto add1 = op::AddExample("add1");
183	- std::vector<int64_t> xShape = {32,4,4,4};
184	- ADD_INPUT(1, x1, inDtype, xShape);
185	- ADD_INPUT(2, x2, inDtype, xShape);
186	-
187	- ADD_OUTPUT(1, y, inDtype, xShape);
188	-
189	- outputs.push_back(add1);
190	- // 添加完毕
191	- return SUCCESS;
192	-}
193	-
194	-int main(int argc, char *argv[])
195	-{
196	- const char *graph_name = "tc_ge_irrun_test";
197	- Graph graph(graph_name);
198	- std::vector<ge::Tensor> input;
199	-
200	- printf("%s - INFO - [XIR]: Start to initialize ge using ge global options\n", GetTime().c_str());
201	- std::map<AscendString, AscendString> global_options = {{"ge.exec.deviceId", "0"}, {"ge.graphRunMode", "1"}};
202	- Status ret = ge::GEInitialize(global_options);
203	- if (ret != SUCCESS) {
204	- printf("%s - INFO - [XIR]: Initialize ge using ge global options failed\n", GetTime().c_str());
205	- return FAILED;
206	- }
207	- printf("%s - INFO - [XIR]: Initialize ge using ge global options success\n", GetTime().c_str());
208	-
209	- std::vector<Operator> inputs{};
210	- std::vector<Operator> outputs{};
211	-
212	- std::cout << argv[1] << std::endl;
213	- char *endptr;
214	-
215	- DataType inDtype = DT_FLOAT;
216	-
217	- std::cout << inDtype << std::endl;
218	-
219	- ret = CreateOppInGraph(inDtype, input, inputs, outputs, graph);
220	- if (ret != SUCCESS) {
221	- printf("%s - ERROR - [XIR]: Create ir session using build options failed\n", GetTime().c_str());
222	- return FAILED;
223	- }
224	-
225	- if (!inputs.empty() && !outputs.empty()) {
226	- graph.SetInputs(inputs).SetOutputs(outputs);
227	- }
228	-
229	- std::map<AscendString, AscendString> build_options = {
230	-
231	- };
232	- printf("%s - INFO - [XIR]: Start to create ir session using build options\n", GetTime().c_str());
233	- ge::Session *session = new Session(build_options);
234	-
235	- if (session == nullptr) {
236	- printf("%s - ERROR - [XIR]: Create ir session using build options failed\n", GetTime().c_str());
237	- return FAILED;
238	- }
239	- printf("%s - INFO - [XIR]: Create ir session using build options success\n", GetTime().c_str());
240	- printf("%s - INFO - [XIR]: Start to add compute graph to ir session\n", GetTime().c_str());
241	-
242	- std::map<AscendString, AscendString> graph_options = {
243	-
244	- };
245	- uint32_t graph_id = 0;
246	- ret = session->AddGraph(graph_id, graph, graph_options);
247	-
248	- printf("%s - INFO - [XIR]: Session add ir compute graph to ir session success\n", GetTime().c_str());
249	- printf("%s - INFO - [XIR]: dump graph to txt\n", GetTime().c_str());
250	- std::string file_path = "./dump";
251	- aclgrphDumpGraph(graph, file_path.c_str(), file_path.length());
252	- printf("%s - INFO - [XIR]: Start to run ir compute graph\n", GetTime().c_str());
253	- std::vector<ge::Tensor> output;
254	- ret = session->RunGraph(graph_id, input, output);
255	- if (ret != SUCCESS) {
256	- printf("%s - INFO - [XIR]: Run graph failed\n", GetTime().c_str());
257	- delete session;
258	- GEFinalize();
259	- return FAILED;
260	- }
261	- printf("%s - INFO - [XIR]: Session run ir compute graph success\n", GetTime().c_str());
262	-
263	- int input_num = input.size();
264	- for (int i = 0; i < input_num; i++) {
265	- std::cout << "input " << i << " dtype : " << input[i].GetTensorDesc().GetDataType() << std::endl;
266	- string input_file = "./tc_ge_irrun_test_0008_npu_input_" + std::to_string(i) + ".bin";
267	- uint8_t *input_data_i = input[i].GetData();
268	- int64_t input_shape = input[i].GetTensorDesc().GetShape().GetShapeSize();
269	- std::cout << "this is " << i << "th input, input shape size =" << input_shape << std::endl;
270	- uint32_t data_size = input_shape * GetDataTypeSize(input[i].GetTensorDesc().GetDataType());
271	- WriteDataToFile((const char *)input_file.c_str(), data_size, input_data_i);
272	- }
273	-
274	- int output_num = output.size();
275	- for (int i = 0; i < output_num; i++) {
276	- std::cout << "output " << i << " dtype : " << output[i].GetTensorDesc().GetDataType() << std::endl;
277	- string output_file = "./tc_ge_irrun_test_0008_npu_output_" + std::to_string(i) + ".bin";
278	- uint8_t *output_data_i = output[i].GetData();
279	- int64_t output_shape = output[i].GetTensorDesc().GetShape().GetShapeSize();
280	- std::cout << "this is " << i << "th output, output shape size =" << output_shape << std::endl;
281	- uint32_t data_size = output_shape * GetDataTypeSize(output[i].GetTensorDesc().GetDataType());
282	- WriteDataToFile((const char *)output_file.c_str(), data_size, output_data_i);
283	- }
284	-
285	- ge::AscendString error_msg = ge::GEGetErrorMsgV2();
286	- std::string error_str(error_msg.GetString());
287	- std::cout << "Error message: " << error_str << std::endl;
288	- ge::AscendString warning_msg = ge::GEGetWarningMsgV2();
289	- std::string warning_str(warning_msg.GetString());
290	- std::cout << "Warning message: " << warning_str << std::endl;
291	- printf("%s - INFO - [XIR]: Precision is ok\n", GetTime().c_str());
292	- printf("%s - INFO - [XIR]: Start to finalize ir graph session\n", GetTime().c_str());
293	- ret = ge::GEFinalize();
294	- if (ret != SUCCESS) {
295	- printf("%s - INFO - [XIR]: Finalize ir graph session failed\n", GetTime().c_str());
296	- return FAILED;
297	- }
298	- printf("%s - INFO - [XIR]: Finalize ir graph session success\n", GetTime().c_str());
299	- return SUCCESS;
300	-}

Dexamples/add_example/op_graph/add_example_graph_infer.cpp+0-38

	@@ -1,38 +0,0 @@
1	-/**
2	- * Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	- * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	- * CANN Open Software License Agreement Version 2.0 (the "License").
5	- * Please refer to the License for details. You may not use this file except in compliance with the License.
6	- * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	- * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	- * See LICENSE in the root of the software repository for the full text of the License.
9	- */
10	-
11	-/*!
12	- * \file add_example_graph_infer.cpp
13	- * \brief add_example operater graph infer resource
14	- */
15	-
16	-#include "register/op_impl_registry.h"
17	-#include "log/log.h"
18	-
19	-namespace ops {
20	-using namespace ge;
21	-
22	-static constexpr int64_t IDX_0 = 0;
23	-
24	-static ge::graphStatus InferDataTypeAddExample(gert::InferDataTypeContext* context)
25	-{
26	- OP_LOGD(context->GetNodeName(), "Begin to do InferDataTypeAddExample");
27	-
28	- // 设置输出的dtype
29	- ge::DataType sizeDtype = context->GetInputDataType(IDX_0);
30	- context->SetOutputDataType(IDX_0, sizeDtype);
31	-
32	- OP_LOGD(context->GetNodeName(), "End to do InferDataTypeAddExample");
33	- return GRAPH_SUCCESS;
34	-}
35	-
36	-IMPL_OP(AddExample).InferDataType(InferDataTypeAddExample);
37	-
38	-}; // namespace ops

Dexamples/add_example/op_graph/fusion_pass/.gitkeep+0-0

The file is empty

Mexamples/add_example/op_host/CMakeLists.txt+9-2

	@@ -6,6 +6,13 @@
6	# THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,	6	# THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	# INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.	7	# INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	# See LICENSE in the root of the software repository for the full text of the License.	8	# See LICENSE in the root of the software repository for the full text of the License.
9	-# -----------------------------------------------------------------------------------------------------------	9	+# ----------------------------------------------------------------------------
10		10
11	-add_modules_sources(OPTYPE add_example ACLNNTYPE aclnn)	11	+if (BUILD_OPEN_PROJECT)
		12	+ target_sources(op_host_aclnn PRIVATE
		13	+ add_example_def.cpp
		14	+ )
		15	+ add_modules_sources(OPTYPE add_example ACLNNTYPE aclnn)
		16	+else()
		17	+ add_modules_sources(OPTYPE add_example ACLNNTYPE aclnn)
		18	+endif()

Mexamples/add_example/op_host/add_example_def.cpp+4-2

	@@ -9,13 +9,14 @@
9	*/	9	*/
10		10
11	/*!	11	/*!
12	- * \file add_example.cpp	12	+ * \file add_example_def.cpp
13	* \brief	13	* \brief
14	*/	14	*/
15	#include "register/op_def_registry.h"	15	#include "register/op_def_registry.h"
16		16
17	namespace ops {	17	namespace ops {
18	-class AddExample : public OpDef {	18	+class AddExample : public OpDef
		19	+{
19	public:	20	public:
20	explicit AddExample(const char* name) : OpDef(name)	21	explicit AddExample(const char* name) : OpDef(name)
21	{	22	{
	@@ -47,6 +48,7 @@ public:
47	.PrecisionReduceFlag(true)	48	.PrecisionReduceFlag(true)
48	.ExtendCfgInfo("opFile.value", "add_example"); // 这里制定的值会对应到kernel入口文件名.cpp	49	.ExtendCfgInfo("opFile.value", "add_example"); // 这里制定的值会对应到kernel入口文件名.cpp
49	this->AICore().AddConfig("ascend910b", aicoreConfig); // 其他的soc版本补充部分配置项	50	this->AICore().AddConfig("ascend910b", aicoreConfig); // 其他的soc版本补充部分配置项
		51	+ this->AICore().AddConfig("ascend910_93", aicoreConfig);
50	}	52	}
51	};	53	};
52	OP_ADD(AddExample); // 添加算子信息库	54	OP_ADD(AddExample); // 添加算子信息库

Mexamples/add_example/op_host/add_example_tiling.cpp+5-4

	@@ -17,12 +17,12 @@
17	#include "util/math_util.h"	17	#include "util/math_util.h"
18	#include "tiling_base/tiling_util.h"	18	#include "tiling_base/tiling_util.h"
19	#include "tiling_base/tiling_templates_registry.h"	19	#include "tiling_base/tiling_templates_registry.h"
20	-#include "../op_kernel/add_example_tiling_data.h"	20	+#include "examples/add_example/op_kernel/add_example_tiling_data.h"
21	-#include "../op_kernel/add_example_tiling_key.h"	21	+#include "examples/add_example/op_kernel/add_example_tiling_key.h"
22		22
23	namespace optiling {	23	namespace optiling {
24		24
25	-using namespace Ops::Math::OpTiling;	25	+using namespace Ops::Transformer::OpTiling;
26		26
27	const uint32_t BLOCK_DIM = 8;	27	const uint32_t BLOCK_DIM = 8;
28	const int64_t TILE_NUM = 8;	28	const int64_t TILE_NUM = 8;
	@@ -34,7 +34,8 @@ constexpr uint32_t INDEXONE = 1;
34	constexpr uint32_t INDEXTWO = 2;	34	constexpr uint32_t INDEXTWO = 2;
35	constexpr uint32_t INDEXTHREE = 3;	35	constexpr uint32_t INDEXTHREE = 3;
36		36
37	-struct AddExampleCompileInfo {};	37	+struct AddExampleCompileInfo {
		38	+};
38		39
39	// 获取平台信息如ubSize, coreNum	40	// 获取平台信息如ubSize, coreNum
40	static ge::graphStatus GetPlatformInfo(gert::TilingContext* context, uint64_t& ubSize, int64_t& coreNum)	41	static ge::graphStatus GetPlatformInfo(gert::TilingContext* context, uint64_t& ubSize, int64_t& coreNum)

Mexamples/add_example/op_kernel/add_example.h+5-16

	@@ -58,10 +58,9 @@ private:
58	template <typename T>	58	template <typename T>
59	__aicore__ inline void AddExample<T>::Init(GM_ADDR x, GM_ADDR y, GM_ADDR z, const AddExampleTilingData* tilingData)	59	__aicore__ inline void AddExample<T>::Init(GM_ADDR x, GM_ADDR y, GM_ADDR z, const AddExampleTilingData* tilingData)
60	{	60	{
61	- blockLength_ = (tilingData->totalLength + AscendC::GetBlockNum() - 1) / AscendC::GetBlockNum();	61	+ blockLength_ = tilingData->totalLength / AscendC::GetBlockNum();
62	tileNum_ = tilingData->tileNum;	62	tileNum_ = tilingData->tileNum;
63	- tileLength_ = ((blockLength_ + tileNum_ - 1) / tileNum_ / BUFFER_NUM) ?	63	+ tileLength_ = blockLength_ / tileNum_ / BUFFER_NUM;
64	- ((blockLength_ + tileNum_ - 1) / tileNum_ / BUFFER_NUM) : 1;
65		64
66	inputGMX.SetGlobalBuffer((__gm__ T)x + blockLength_ AscendC::GetBlockIdx(), blockLength_);	65	inputGMX.SetGlobalBuffer((__gm__ T)x + blockLength_ AscendC::GetBlockIdx(), blockLength_);
67	inputGMY.SetGlobalBuffer((__gm__ T)y + blockLength_ AscendC::GetBlockIdx(), blockLength_);	66	inputGMY.SetGlobalBuffer((__gm__ T)y + blockLength_ AscendC::GetBlockIdx(), blockLength_);
	@@ -77,13 +76,8 @@ __aicore__ inline void AddExample<T>::CopyIn(int32_t progress)
77	{	76	{
78	AscendC::LocalTensor<T> xLocal = inputQueueX.AllocTensor<T>();	77	AscendC::LocalTensor<T> xLocal = inputQueueX.AllocTensor<T>();
79	AscendC::LocalTensor<T> yLocal = inputQueueY.AllocTensor<T>();	78	AscendC::LocalTensor<T> yLocal = inputQueueY.AllocTensor<T>();
80	- AscendC::DataCopyParams copyParams;	79	+ AscendC::DataCopy(xLocal, inputGMX[progress * tileLength_], tileLength_);
81	- copyParams.blockCount = 1;	80	+ AscendC::DataCopy(yLocal, inputGMY[progress * tileLength_], tileLength_);
82	- copyParams.blockLen = tileLength_ * sizeof(T);
83	- copyParams.srcStride = 0;
84	- copyParams.dstStride = 0;
85	- AscendC::DataCopyPad(xLocal, inputGMX[progress * tileLength_], copyParams, {false, 0, 0, 0});
86	- AscendC::DataCopyPad(yLocal, inputGMY[progress * tileLength_], copyParams, {false, 0, 0, 0});
87	inputQueueX.EnQue(xLocal);	81	inputQueueX.EnQue(xLocal);
88	inputQueueY.EnQue(yLocal);	82	inputQueueY.EnQue(yLocal);
89	}	83	}
	@@ -92,12 +86,7 @@ template <typename T>
92	__aicore__ inline void AddExample<T>::CopyOut(int32_t progress)	86	__aicore__ inline void AddExample<T>::CopyOut(int32_t progress)
93	{	87	{
94	AscendC::LocalTensor<T> zLocal = outputQueueZ.DeQue<T>();	88	AscendC::LocalTensor<T> zLocal = outputQueueZ.DeQue<T>();
95	- AscendC::DataCopyParams copyParams;	89	+ AscendC::DataCopy(outputGMZ[progress * tileLength_], zLocal, tileLength_);
96	- copyParams.blockCount = 1;
97	- copyParams.blockLen = tileLength_ * sizeof(T);
98	- copyParams.srcStride = 0;
99	- copyParams.dstStride = 0;
100	- AscendC::DataCopyPad(outputGMZ[progress * tileLength_], zLocal, copyParams);
101	outputQueueZ.FreeTensor(zLocal);	90	outputQueueZ.FreeTensor(zLocal);
102	}	91	}
103		92

Dexamples/add_example/tests/.gitkeep+0-0

The file is empty

Dexamples/add_example/tests/CMakeLists.txt+0-16

	@@ -1,16 +0,0 @@
1	-# -----------------------------------------------------------------------------------------------------------
2	-# Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	-# This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	-# CANN Open Software License Agreement Version 2.0 (the "License").
5	-# Please refer to the License for details. You may not use this file except in compliance with the License.
6	-# THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	-# INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	-# See LICENSE in the root of the software repository for the full text of the License.
9	-# -----------------------------------------------------------------------------------------------------------
10	-
11	-file(GLOB CURRENT_DIRS RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}/*)
12	-foreach(SUB_DIR ${CURRENT_DIRS})
13	- if(EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/${SUB_DIR}/CMakeLists.txt")
14	- add_subdirectory(${SUB_DIR})
15	- endif()
16	-endforeach()

Dexamples/add_example/tests/ut/CMakeLists.txt+0-16

	@@ -1,16 +0,0 @@
1	-# -----------------------------------------------------------------------------------------------------------
2	-# Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	-# This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	-# CANN Open Software License Agreement Version 2.0 (the "License").
5	-# Please refer to the License for details. You may not use this file except in compliance with the License.
6	-# THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	-# INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	-# See LICENSE in the root of the software repository for the full text of the License.
9	-# -----------------------------------------------------------------------------------------------------------
10	-
11	-file(GLOB CURRENT_DIRS RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}/*)
12	-foreach(SUB_DIR ${CURRENT_DIRS})
13	- if(EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/${SUB_DIR}/CMakeLists.txt")
14	- add_subdirectory(${SUB_DIR})
15	- endif()
16	-endforeach()

Dexamples/add_example/tests/ut/op_host/CMakeLists.txt+0-20

	@@ -1,20 +0,0 @@
1	-# -----------------------------------------------------------------------------------------------------------
2	-# Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	-# This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	-# CANN Open Software License Agreement Version 2.0 (the "License").
5	-# Please refer to the License for details. You may not use this file except in compliance with the License.
6	-# THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	-# INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	-# See LICENSE in the root of the software repository for the full text of the License.
9	-# -----------------------------------------------------------------------------------------------------------
10	-
11	-if(UT_TEST_ALL OR OP_HOST_UT)
12	- add_modules_ut_sources(UT_NAME ${OP_TILING_MODULE_NAME} MODE PRIVATE DIR ${CMAKE_CURRENT_SOURCE_DIR})
13	-endif()
14	-
15	-file(GLOB CURRENT_DIRS RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}/*)
16	-foreach(SUB_DIR ${CURRENT_DIRS})
17	- if(EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/${SUB_DIR}/CMakeLists.txt")
18	- add_subdirectory(${SUB_DIR})
19	- endif()
20	-endforeach()

Dexamples/add_example/tests/ut/op_host/test_add_example_tiling.cpp+0-73

	@@ -1,73 +0,0 @@
1	-/**
2	- * Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	- * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	- * CANN Open Software License Agreement Version 2.0 (the "License").
5	- * Please refer to the License for details. You may not use this file except in compliance with the License.
6	- * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	- * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	- * See LICENSE in the root of the software repository for the full text of the License.
9	- */
10	-
11	-#include <iostream>
12	-#include <gtest/gtest.h>
13	-#include "tiling_context_faker.h"
14	-#include "tiling_case_executor.h"
15	-#include "../../../op_kernel/add_example_tiling_data.h"
16	-
17	-using namespace std;
18	-using namespace ge;
19	-
20	-class AddExampleTiling : public testing::Test {
21	-protected:
22	- static void SetUpTestCase()
23	- {
24	- std::cout << "AddExampleTiling SetUp" << std::endl;
25	- }
26	-
27	- static void TearDownTestCase()
28	- {
29	- std::cout << "AddExampleTiling TearDown" << std::endl;
30	- }
31	-};
32	-
33	-std::map<std::string, std::string> soc_version_infos = {{"Short_SoC_version", "Ascend910B"}};
34	-
35	-TEST_F(AddExampleTiling, add_example_0) {
36	- struct AddExampleCompileInfo {} compileInfo;
37	- gert::TilingContextPara tilingContextPara("AddExample",
38	- {
39	- {{{1, 2, 8, 16}, {1, 2, 8, 16}}, ge::DT_FLOAT, ge::FORMAT_ND}, // input tensor1
40	- {{{1, 2, 8, 16}, {1, 2, 8, 16}}, ge::DT_FLOAT, ge::FORMAT_ND}, // input tensor2
41	- },
42	- {
43	- {{{1, 2, 8, 16}, {1, 2, 8, 16}}, ge::DT_FLOAT, ge::FORMAT_ND}, // output tensor
44	- },
45	- {
46	- /* attrs */
47	- },
48	- &compileInfo);
49	- uint64_t expectTilingKey = 0;
50	- string expectTilingData = "256 8 ";
51	- std::vector<size_t> expectWorkspaces = {1024 * 1024 * 16};
52	- ExecuteTestCase(tilingContextPara, ge::GRAPH_SUCCESS, expectTilingKey, expectTilingData, expectWorkspaces);
53	-}
54	-
55	-TEST_F(AddExampleTiling, add_example_1) {
56	- struct AddExampleCompileInfo {} compileInfo;
57	- gert::TilingContextPara tilingContextPara("AddExample",
58	- {
59	- {{{1, 2, 8, 16}, {1, 2, 8, 16}}, ge::DT_INT32, ge::FORMAT_ND}, // input tensor1
60	- {{{1, 2, 8, 16}, {1, 2, 8, 16}}, ge::DT_INT32, ge::FORMAT_ND}, // input tensor2
61	- },
62	- {
63	- {{{1, 2, 8, 16}, {1, 2, 8, 16}}, ge::DT_INT32, ge::FORMAT_ND}, // output tensor
64	- },
65	- {
66	- /* attrs */
67	- },
68	- &compileInfo);
69	- uint64_t expectTilingKey = 1;
70	- string expectTilingData = "256 8 ";
71	- std::vector<size_t> expectWorkspaces = {1024 * 1024 * 16};
72	- ExecuteTestCase(tilingContextPara, ge::GRAPH_SUCCESS, expectTilingKey, expectTilingData, expectWorkspaces);
73	-}

Dexamples/add_example/tests/ut/op_kernel/CMakeLists.txt+0-26

	@@ -1,26 +0,0 @@
1	-# -----------------------------------------------------------------------------------------------------------
2	-# Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	-# This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	-# CANN Open Software License Agreement Version 2.0 (the "License").
5	-# Please refer to the License for details. You may not use this file except in compliance with the License.
6	-# THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	-# INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	-# See LICENSE in the root of the software repository for the full text of the License.
9	-# -----------------------------------------------------------------------------------------------------------
10	-
11	-if (UT_TEST_ALL OR OP_KERNEL_UT)
12	- # 需要将Tiling依赖的文件添加到CMakeLists.txt中
13	- # set(elewise_common_tiling_files
14	- # ${CANN_ROOT}/ops/built-in/op_tiling/runtime/elewise_tiling.cc
15	- # )
16	- # 算子自己的tiling文件路径
17	- set(add_example_tiling_files
18	- ${CMAKE_CURRENT_SOURCE_DIR}/../../../op_host/add_example_tiling.cpp
19	- )
20	- # 使用AddOpTestCase
21	- # param1：算子名称，以kernel方式命名
22	- # param2：soc版本，多个以分号分隔，例如："ascend910_9599;AscendB1"
23	- # param3：自定义编译选项，一般填写测试的一种典型数据类型组合，不需要则传入空字符串，例如："-DDTYPE_X=float"，多个使用空格分隔，例如："-DDTYPE_X=float -DDTYPE_Y=float"
24	- # param4：该算子依赖的所有tiling源码文件
25	- AddOpTestCase(add_example "ascend910b" "" "${add_example_tiling_files}")
26	-endif()

Dexamples/add_example/tests/ut/op_kernel/add_example_tiling.h+0-58

	@@ -1,58 +0,0 @@
1	-/**
2	- * Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	- * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	- * CANN Open Software License Agreement Version 2.0 (the "License").
5	- * Please refer to the License for details. You may not use this file except in compliance with the License.
6	- * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	- * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	- * See LICENSE in the root of the software repository for the full text of the License.
9	- */
10	-
11	-/*!
12	- * \file is_finite_tiling.h
13	- * \brief
14	- */
15	-
16	-#ifndef _I_ADD_EXAMPLE_TILING_H_
17	-#define _I_ADD_EXAMPLE_TILING_H_
18	-
19	-#include <cstdint>
20	-
21	-#include "../../../op_kernel/add_example_tiling_data.h"
22	-#include "kernel_tiling/kernel_tiling.h"
23	-
24	-// using namespace AddExampleNs;
25	-
26	-#define __aicore__
27	-#ifdef __NPU_TILING__
28	-inline[aicore] void InitTilingData(const __gm__ uint8_t tiling, AddExampleTilingData constData)
29	-{
30	- const __gm__ uint32_t src = (const __gm__ uint32_t )tiling;
31	- uint32_t dst = (uint32_t )constData;
32	- for (size_t i = 0; i < sizeof(AddExampleTilingData) / 4; i++) {
33	- (dst + i) = (src + i);
34	- }
35	-}
36	-#else
37	-inline void InitTilingData(uint8_t tiling, AddExampleTilingData constData)
38	-{
39	- memcpy(constData, tiling, sizeof(AddExampleTilingData));
40	-}
41	-#endif // __NPU_TILING__
42	-
43	-#define CONVERT_TILING_DATA(tilingStruct, tilingDataPointer, tilingPointer) \
44	- __ubuf__ tilingStruct *tilingDataPointer = \
45	- reinterpret_cast<__ubuf__ tilingStruct >((__ubuf__ uint8_t )(tilingPointer));
46	-
47	-#define INIT_TILING_DATA(tilingStruct, tilingDataPointer, tilingPointer) \
48	- CONVERT_TILING_DATA(tilingStruct, tilingDataPointer, tilingPointer);
49	-
50	-#define GET_TILING_DATA_WITH_STRUCT(tilingStruct, tilingData, tilingArg) \
51	- tilingStruct tilingData; \
52	- InitTilingData(tilingArg, &tilingData)
53	-
54	-#define GET_TILING_DATA(tilingData, tilingArg) \
55	- AddExampleTilingData tilingData; \
56	- InitTilingData(tilingArg, &tilingData)
57	-
58	-#endif // _I_ADD_EXAMPLE_TILING_H_

Dexamples/add_example/tests/ut/op_kernel/test_add_example.cpp+0-82

	@@ -1,82 +0,0 @@
1	-/**
2	- * Copyright (c) 2025 Huawei Technologies Co., Ltd.
3	- * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
4	- * CANN Open Software License Agreement Version 2.0 (the "License").
5	- * Please refer to the License for details. You may not use this file except in compliance with the License.
6	- * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
7	- * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
8	- * See LICENSE in the root of the software repository for the full text of the License.
9	- */
10	-
11	-/*!
12	- * \file test_add_example.cpp
13	- * \brief
14	- */
15	-
16	-#include "../../../op_kernel/add_example.cpp"
17	-#include "add_example_tiling.h"
18	-#include <array>
19	-#include <vector>
20	-#include <iostream>
21	-#include <string>
22	-#include <cstdint>
23	-#include <cstdlib>
24	-#include "gtest/gtest.h"
25	-#include "tikicpulib.h"
26	-#include "data_utils.h"
27	-
28	-using namespace std;
29	-
30	-class add_example_test : public testing::Test {
31	-protected:
32	- static void SetUpTestCase()
33	- {
34	- cout << "add_example_test SetUp\n" << endl;
35	- }
36	- static void TearDownTestCase()
37	- {
38	- cout << "add_example_test TearDown\n" << endl;
39	- }
40	-};
41	-
42	-TEST_F(add_example_test, test_case_0)
43	-{
44	- size_t xByteSize = 32 * 4 * 4 * 4 * sizeof(float);
45	- size_t yByteSize = 32 * 4 * 4 * 4 * sizeof(float);
46	- size_t zByteSize = 32 * 4 * 4 * 4 * sizeof(float);
47	- size_t tiling_data_size = sizeof(AddExampleTilingData);
48	- uint32_t blockDim = 8;
49	-
50	- // 分配全局内存（AscendC::GmAlloc 是 Ascend 高层 API，正确）
51	- uint8_t* x = (uint8_t*)AscendC::GmAlloc(xByteSize);
52	- uint8_t* y = (uint8_t*)AscendC::GmAlloc(yByteSize);
53	- uint8_t* z = (uint8_t*)AscendC::GmAlloc(zByteSize);
54	- uint8_t* workspace = (uint8_t)AscendC::GmAlloc(1024 1024 * 16);
55	- uint8_t* tiling = (uint8_t*)AscendC::GmAlloc(tiling_data_size);
56	-
57	- char* path_ = get_current_dir_name();
58	- string path(path_);
59	-
60	- AddExampleTilingData* tilingDatafromBin = reinterpret_cast<AddExampleTilingData*>(tiling);
61	- tilingDatafromBin->totalLength = 32 * 4 * 4 * 4;
62	- tilingDatafromBin->tileNum = 8;
63	-
64	- ICPU_SET_TILING_KEY(0);
65	- AscendC::SetKernelMode(KernelMode::AIV_MODE);
66	-
67	- ICPU_RUN_KF(add_example<0>,
68	- blockDim,
69	- x,
70	- y,
71	- z,
72	- workspace,
73	- (uint8_t *)(tilingDatafromBin));
74	-
75	- // 释放资源
76	- AscendC::GmFree(x);
77	- AscendC::GmFree(y);
78	- AscendC::GmFree(z);
79	- AscendC::GmFree(workspace);
80	- AscendC::GmFree(tiling);
81	- free(path_);
82	-}

	@@ -11,7 +11,7 @@
11	#include <iostream>	11	#include <iostream>
12	#include <vector>	12	#include <vector>
13	#include "acl/acl.h"	13	#include "acl/acl.h"
14	-#include "aclnn_add_example.h"	14	+#include "aclnnop/aclnn_add_example.h"
15		15
16	#define CHECK_RET(cond, return_expr) \	16	#define CHECK_RET(cond, return_expr) \
17	do { \	17	do { \
	@@ -97,21 +97,21 @@ int main()
97	aclTensor* selfX = nullptr;	97	aclTensor* selfX = nullptr;
98	void* selfXDeviceAddr = nullptr;	98	void* selfXDeviceAddr = nullptr;
99	std::vector<int64_t> selfXShape = {32, 4, 4, 4};	99	std::vector<int64_t> selfXShape = {32, 4, 4, 4};
100	- std::vector<float> selfXHostData(2048, 1);	100	+ std::vector<float> selfXHostData(2048, 1); // 2048：创建包含3244*4=2048个元素的向量
101	ret = CreateAclTensor(selfXHostData, selfXShape, &selfXDeviceAddr, aclDataType::ACL_FLOAT, &selfX);	101	ret = CreateAclTensor(selfXHostData, selfXShape, &selfXDeviceAddr, aclDataType::ACL_FLOAT, &selfX);
102	CHECK_RET(ret == ACL_SUCCESS, return ret);	102	CHECK_RET(ret == ACL_SUCCESS, return ret);
103		103
104	aclTensor* selfY = nullptr;	104	aclTensor* selfY = nullptr;
105	void* selfYDeviceAddr = nullptr;	105	void* selfYDeviceAddr = nullptr;
106	std::vector<int64_t> selfYShape = {32, 4, 4, 4};	106	std::vector<int64_t> selfYShape = {32, 4, 4, 4};
107	- std::vector<float> selfYHostData(2048, 1);	107	+ std::vector<float> selfYHostData(2048, 1); // 2048：创建包含3244*4=2048个元素的向量
108	ret = CreateAclTensor(selfYHostData, selfYShape, &selfYDeviceAddr, aclDataType::ACL_FLOAT, &selfY);	108	ret = CreateAclTensor(selfYHostData, selfYShape, &selfYDeviceAddr, aclDataType::ACL_FLOAT, &selfY);
109	CHECK_RET(ret == ACL_SUCCESS, return ret);	109	CHECK_RET(ret == ACL_SUCCESS, return ret);
110		110
111	aclTensor* out = nullptr;	111	aclTensor* out = nullptr;
112	void* outDeviceAddr = nullptr;	112	void* outDeviceAddr = nullptr;
113	std::vector<int64_t> outShape = {32, 4, 4, 4};	113	std::vector<int64_t> outShape = {32, 4, 4, 4};
114	- std::vector<float> outHostData(2048, 1);	114	+ std::vector<float> outHostData(2048, 1); // 2048：创建包含3244*4=2048个元素的向量
115	ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);	115	ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
116	CHECK_RET(ret == ACL_SUCCESS, return ret);	116	CHECK_RET(ret == ACL_SUCCESS, return ret);
117		117