* Copyright (c) 2026 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/
* \file test_geir_expand.cpp
* \brief Test Expand via GE IR graph mode
*/
#include <cstdio>
#include <cstring>
#include <ctime>
#include <map>
#include <memory>
#include <string>
#include <vector>
#include "ge_api.h"
#include "ge_api_types.h"
#include "ge_error_codes.h"
#include "ge_ir_build.h"
#include "graph.h"
#include "tensor.h"
#include "types.h"
#include "../op_graph/expand_proto.h"
#define FAILED -1
#define SUCCESS 0
using namespace ge;
using std::map;
using std::string;
using std::vector;
#define LOG_PRINT(message, ...) \
do { \
printf(message, ##__VA_ARGS__); \
} while (0)
namespace ge {
REG_OP(Data).INPUT(x, TensorType::ALL()).OUTPUT(y, TensorType::ALL()).ATTR(index, Int, 0).OP_END_FACTORY_REG(Data)
}
namespace {
string GetTime()
{
time_t timep;
time(&timep);
char tmp[64] = {0};
struct tm tm_info;
strftime(tmp, sizeof(tmp), "%Y-%m-%d %H:%M:%S,000", localtime_r(&timep, &tm_info));
return tmp;
}
uint32_t GetDataTypeSize(DataType data_type)
{
if (data_type == ge::DT_BOOL || data_type == ge::DT_INT8 || data_type == ge::DT_UINT8) {
return 1;
}
if (data_type == ge::DT_INT16 || data_type == ge::DT_UINT16 || data_type == ge::DT_FLOAT16 ||
data_type == ge::DT_BF16) {
return 2;
}
if (data_type == ge::DT_INT32 || data_type == ge::DT_UINT32 || data_type == ge::DT_FLOAT) {
return 4;
}
if (data_type == ge::DT_INT64 || data_type == ge::DT_UINT64 || data_type == ge::DT_DOUBLE) {
return 8;
}
return 0;
}
uint16_t FloatToBf16Bits(float value)
{
uint32_t bits = 0;
std::memcpy(&bits, &value, sizeof(bits));
return static_cast<uint16_t>(bits >> 16);
}
int32_t WriteDataToFile(const string& bin_file, uint64_t data_size, const uint8_t* input_data)
{
FILE* fp = fopen(bin_file.c_str(), "wb");
if (fp == nullptr) {
printf("Failed to open file: %s\n", bin_file.c_str());
return FAILED;
}
size_t written = fwrite(input_data, sizeof(uint8_t), data_size, fp);
fclose(fp);
if (written != data_size) {
printf("Failed to write file: %s\n", bin_file.c_str());
return FAILED;
}
return SUCCESS;
}
int32_t GenBf16Data(
const vector<int64_t>& shapes, Tensor& input_tensor, TensorDesc& input_tensor_desc, const vector<uint16_t>& values)
{
input_tensor_desc.SetRealDimCnt(shapes.size());
size_t size = 1;
for (size_t i = 0; i < shapes.size(); ++i) {
size *= static_cast<size_t>(shapes[i]);
}
if (values.size() != size) {
return FAILED;
}
input_tensor =
Tensor(input_tensor_desc, reinterpret_cast<const uint8_t*>(values.data()), values.size() * sizeof(uint16_t));
return SUCCESS;
}
int32_t GenInt64Data(
const vector<int64_t>& shapes, Tensor& input_tensor, TensorDesc& input_tensor_desc, const vector<int64_t>& values)
{
input_tensor_desc.SetRealDimCnt(shapes.size());
size_t size = 1;
for (size_t i = 0; i < shapes.size(); ++i) {
size *= static_cast<size_t>(shapes[i]);
}
if (values.size() != size) {
return FAILED;
}
input_tensor =
Tensor(input_tensor_desc, reinterpret_cast<const uint8_t*>(values.data()), values.size() * sizeof(int64_t));
return SUCCESS;
}
int CreateOppInGraph(vector<ge::Tensor>& input, vector<Operator>& inputs, vector<Operator>& outputs, Graph& graph)
{
auto expand = op::Expand("expand_graph");
vector<int64_t> x_shape = {1, 2};
vector<int64_t> shape_shape = {2};
vector<int64_t> y_shape = {3, 2};
auto x = op::Data("placeholder0").set_attr_index(0);
TensorDesc x_desc(ge::Shape(x_shape), FORMAT_ND, DT_BF16);
x_desc.SetPlacement(ge::kPlacementHost);
Tensor x_tensor;
int ret = GenBf16Data(x_shape, x_tensor, x_desc, {FloatToBf16Bits(1.0F), FloatToBf16Bits(2.0F)});
if (ret != SUCCESS) {
printf("%s - ERROR - [XIR]: Generate input x data failed\n", GetTime().c_str());
return FAILED;
}
x.update_input_desc_x(x_desc);
x.update_output_desc_y(x_desc);
input.push_back(x_tensor);
graph.AddOp(x);
expand.set_input_x(x);
inputs.push_back(x);
auto shape = op::Data("placeholder1").set_attr_index(1);
TensorDesc shape_desc(ge::Shape(shape_shape), FORMAT_ND, DT_INT64);
shape_desc.SetPlacement(ge::kPlacementHost);
Tensor shape_tensor;
ret = GenInt64Data(shape_shape, shape_tensor, shape_desc, {3, 2});
if (ret != SUCCESS) {
printf("%s - ERROR - [XIR]: Generate input shape data failed\n", GetTime().c_str());
return FAILED;
}
shape.update_input_desc_x(shape_desc);
shape.update_output_desc_y(shape_desc);
input.push_back(shape_tensor);
graph.AddOp(shape);
expand.set_input_shape(shape);
expand.update_input_desc_shape(shape_desc);
inputs.push_back(shape);
TensorDesc y_desc(ge::Shape(y_shape), FORMAT_ND, DT_BF16);
expand.update_output_desc_y(y_desc);
outputs.push_back(expand);
return SUCCESS;
}
int SaveInputOutput(const vector<ge::Tensor>& input, const vector<ge::Tensor>& output)
{
for (size_t i = 0; i < input.size(); ++i) {
string input_file = "./tc_ge_irrun_expand_npu_input_" + std::to_string(i) + ".bin";
const uint8_t* input_data = input[i].GetData();
int64_t shape_size = input[i].GetTensorDesc().GetShape().GetShapeSize();
uint32_t data_size =
static_cast<uint32_t>(shape_size) * GetDataTypeSize(input[i].GetTensorDesc().GetDataType());
if (WriteDataToFile(input_file, data_size, input_data) != SUCCESS) {
return FAILED;
}
}
const uint16_t expected[] = {FloatToBf16Bits(1.0F), FloatToBf16Bits(2.0F), FloatToBf16Bits(1.0F),
FloatToBf16Bits(2.0F), FloatToBf16Bits(1.0F), FloatToBf16Bits(2.0F)};
for (size_t i = 0; i < output.size(); ++i) {
string output_file = "./tc_ge_irrun_expand_npu_output_" + std::to_string(i) + ".bin";
const uint8_t* output_data = output[i].GetData();
int64_t shape_size = output[i].GetTensorDesc().GetShape().GetShapeSize();
uint32_t data_size =
static_cast<uint32_t>(shape_size) * GetDataTypeSize(output[i].GetTensorDesc().GetDataType());
if (WriteDataToFile(output_file, data_size, output_data) != SUCCESS) {
return FAILED;
}
const uint16_t* result = reinterpret_cast<const uint16_t*>(output_data);
for (int64_t j = 0; j < shape_size; ++j) {
LOG_PRINT("result[%ld] bf16 bits: 0x%04x\n", j, static_cast<unsigned int>(result[j]));
if (result[j] != expected[j]) {
printf("Expand output mismatch at index %ld\n", j);
return FAILED;
}
}
}
return SUCCESS;
}
}
int main(int argc, char* argv[])
{
(void)argc;
(void)argv;
const char* graph_name = "tc_ge_irrun_expand";
Graph graph(graph_name);
vector<ge::Tensor> input;
printf("%s - INFO - [XIR]: Start to initialize ge using ge global options\n", GetTime().c_str());
map<AscendString, AscendString> global_options = {{"ge.exec.deviceId", "0"}, {"ge.graphRunMode", "1"}};
Status ret = ge::GEInitialize(global_options);
if (ret != SUCCESS) {
printf("%s - ERROR - [XIR]: Initialize ge using ge global options failed\n", GetTime().c_str());
return FAILED;
}
vector<Operator> inputs;
vector<Operator> outputs;
ret = CreateOppInGraph(input, inputs, outputs, graph);
if (ret != SUCCESS) {
ge::GEFinalize();
return FAILED;
}
if (!inputs.empty() && !outputs.empty()) {
graph.SetInputs(inputs).SetOutputs(outputs);
}
map<AscendString, AscendString> build_options;
std::unique_ptr<ge::Session> session(new (std::nothrow) Session(build_options));
if (session == nullptr) {
printf("%s - ERROR - [XIR]: Create ir session failed\n", GetTime().c_str());
ge::GEFinalize();
return FAILED;
}
map<AscendString, AscendString> graph_options = {{"ge.exec.precision_mode", "allow_mix_precision"}};
uint32_t graph_id = 0;
ret = session->AddGraph(graph_id, graph, graph_options);
if (ret != SUCCESS) {
printf("%s - ERROR - [XIR]: Add graph failed\n", GetTime().c_str());
ge::GEFinalize();
return FAILED;
}
vector<ge::Tensor> output;
ret = session->RunGraph(graph_id, input, output);
if (ret != SUCCESS) {
printf("%s - ERROR - [XIR]: Run graph failed\n", GetTime().c_str());
ge::GEFinalize();
return FAILED;
}
printf("Input dtype: %d\n", DT_BF16);
printf(
"Hint: if your active OPP also enables AI Core Expand for BF16, temporarily mask the AI Core Expand entry to "
"probe AICPU dispatch.\n");
ret = SaveInputOutput(input, output);
if (ret != SUCCESS) {
ge::GEFinalize();
return FAILED;
}
ret = ge::GEFinalize();
if (ret != SUCCESS) {
printf("%s - ERROR - [XIR]: GE Finalize failed\n", GetTime().c_str());
return FAILED;
}
return SUCCESS;
}
