#undef __FPU_PRESENT
#include "stm32h7xx.h"
#include "main.h"
#include "crc.h"
#include "memorymap.h"
#include "gpio.h"
#include "app_x-cube-ai.h"
#include "network.h"
#include "ai_platform.h"
#include "network_data.h"
#include <stdlib.h>
#include <stdio.h>
#include "network_data.h" // 确保包含权重数据声明
/* 手动定义缺失的宏（必须与network.c中的实际值一致） */
#ifndef AI_NETWORK_WEIGHTS_SIZE
#define AI_NETWORK_WEIGHTS_SIZE 794136 // 根据network.c中的值填写
#endif
#ifndef AI_NETWORK_ACTIVATIONS_SIZE
#define AI_NETWORK_ACTIVATIONS_SIZE 45572 // 根据network.c中的值填写
#endif
/* 输入输出缓冲区定义 */
AI_ALIGNED(32) static float input_data[90 * 3 * 1 * 1] = {0}; // 严格匹配network.c输入形状
AI_ALIGNED(32) static float output_data[1 * 1 * 6 * 1] = {0}; // 匹配输出形状
static ai_handle network_handle = AI_HANDLE_NULL;
/* 函数原型声明 -------------------------------------------------------------*/
void SystemClock_Config(void);
static void MPU_Config(void);
static int argmax(const float *arr, int size);
void Error_Handler(void);
int main(void)
{
MPU_Config();
HAL_Init();
SystemClock_Config();
MX_GPIO_Init();
MX_CRC_Init();
MX_Core_Init();
printf("AI Model Initialized\n");
/* 1. 创建网络实例 */
ai_error err = ai_mnetwork_create(
AI_NETWORK_MODEL_NAME, // 使用network.h中定义的模型名称
&network_handle,
NULL // 使用默认配置
);
if (err.type != AI_ERROR_NONE) {
printf("Network create error: 0x%08X\n", err.code);
Error_Handler();
}
printf("1\n");
/* 2. 初始化网络参数 */
ai_network_params params = {
.params = {
.format = AI_BUFFER_FORMAT_U8,
.n_batches = 1,
.height = 1,
.width = AI_NETWORK_WEIGHTS_SIZE,
.channels = 1,
.data = AI_HANDLE_PTR(ai_network_data_weights_get())
},
.activations = {
.format = AI_BUFFER_FORMAT_U8,
.n_batches = 1,
.height = 1,
.width = AI_NETWORK_ACTIVATIONS_SIZE,
.channels = 1,
.data = AI_HANDLE_PTR(input_data)
}
};
if (!ai_mnetwork_init(network_handle, ¶ms)) {
printf("Network init failed\n");
ai_mnetwork_destroy(network_handle);
Error_Handler();
}
printf("2\n");
/* 3. 配置输入输出缓冲区 */
ai_buffer input_buf = AI_BUFFER_OBJ_INIT(
AI_BUFFER_FORMAT_FLOAT, // format_
90, // height (根据network.c中的AI_SHAPE_INIT(90,3,1,1))
3, // width
1, // channels
1, // n_batches
input_data // data
);
ai_buffer output_buf = AI_BUFFER_OBJ_INIT(
AI_BUFFER_FORMAT_FLOAT,
1, // height (根据network.c中的AI_SHAPE_INIT(1,1,6,1))
1,
6,
1,
output_data
);
/* 主循环 */
srand(HAL_GetTick());
printf("3\n");
while (1)
{
/* 生成输入数据 */
for (int i = 0; i < 90 * 3; i++) {
input_data[i] = (2.0f * rand() / RAND_MAX) - 1.0f;
}
printf("4\n");
/* 缓存维护 */
SCB_CleanDCache_by_Addr(
(uint32_t*)((uintptr_t)input_data & ~0x1F),
sizeof(input_data) + 32
);
// 在调用前添加地址验证
printf("ai_network_run实际地址: 0x%08X\n", (uint32_t)ai_network_run);
/* 执行推理 */
int result = ai_mnetwork_run(network_handle, &input_buf, &output_buf);
if (result != AI_ERROR_NONE) {
printf("Inference error: %d\n", result);
Error_Handler();
}
/* 处理输出 */
SCB_InvalidateDCache_by_Addr(
(uint32_t*)((uintptr_t)output_data & ~0x1F),
sizeof(output_data) + 32
);
int class_id = argmax(output_data, 6);
printf("Predicted class: %d\n", class_id);
HAL_Delay(1000);
}
}
/* 实现必须严格匹配声明 */
static int argmax(const float *arr, int size)
{
int max_idx = 0;
for (int i = 1; i < size; i++) {
if (arr[i] > arr[max_idx]) max_idx = i;
}
return max_idx;
}
/**
* @brief System Clock Configuration
* @retval None
*/
void SystemClock_Config(void)
{
RCC_OscInitTypeDef RCC_OscInitStruct = {0};
RCC_ClkInitTypeDef RCC_ClkInitStruct = {0};
/** Supply configuration update enable
*/
HAL_PWREx_ConfigSupply(PWR_LDO_SUPPLY);
/** Configure the main internal regulator output voltage
*/
__HAL_PWR_VOLTAGESCALING_CONFIG(PWR_REGULATOR_VOLTAGE_SCALE3);
while(!__HAL_PWR_GET_FLAG(PWR_FLAG_VOSRDY)) {}
/** Initializes the RCC Oscillators according to the specified parameters
* in the RCC_OscInitTypeDef structure.
*/
RCC_OscInitStruct.OscillatorType = RCC_OSCILLATORTYPE_HSI;
RCC_OscInitStruct.HSIState = RCC_HSI_DIV1;
RCC_OscInitStruct.HSICalibrationValue = RCC_HSICALIBRATION_DEFAULT;
RCC_OscInitStruct.PLL.PLLState = RCC_PLL_NONE;
if (HAL_RCC_OscConfig(&RCC_OscInitStruct) != HAL_OK)
{
Error_Handler();
}
/** Initializes the CPU, AHB and APB buses clocks
*/
RCC_ClkInitStruct.ClockType = RCC_CLOCKTYPE_HCLK|RCC_CLOCKTYPE_SYSCLK
|RCC_CLOCKTYPE_PCLK1|RCC_CLOCKTYPE_PCLK2
|RCC_CLOCKTYPE_D3PCLK1|RCC_CLOCKTYPE_D1PCLK1;
RCC_ClkInitStruct.SYSCLKSource = RCC_SYSCLKSOURCE_HSI;
RCC_ClkInitStruct.SYSCLKDivider = RCC_SYSCLK_DIV1;
RCC_ClkInitStruct.AHBCLKDivider = RCC_HCLK_DIV1;
RCC_ClkInitStruct.APB3CLKDivider = RCC_APB3_DIV1;
RCC_ClkInitStruct.APB1CLKDivider = RCC_APB1_DIV1;
RCC_ClkInitStruct.APB2CLKDivider = RCC_APB2_DIV2;
RCC_ClkInitStruct.APB4CLKDivider = RCC_APB4_DIV1;
if (HAL_RCC_ClockConfig(&RCC_ClkInitStruct, FLASH_LATENCY_1) != HAL_OK)
{
Error_Handler();
}
}
/* USER CODE BEGIN 4 */
/* USER CODE END 4 */
/* MPU Configuration */
void MPU_Config(void)
{
MPU_Region_InitTypeDef MPU_InitStruct = {0};
/* Disables the MPU */
HAL_MPU_Disable();
/** Initializes and configures the Region and the memory to be protected
*/
MPU_InitStruct.Enable = MPU_REGION_ENABLE;
MPU_InitStruct.Number = MPU_REGION_NUMBER0;
MPU_InitStruct.BaseAddress = 0x24000000;
MPU_InitStruct.Size = MPU_REGION_SIZE_512KB;
MPU_InitStruct.SubRegionDisable = 0x87;
MPU_InitStruct.TypeExtField = MPU_TEX_LEVEL0;
MPU_InitStruct.AccessPermission = MPU_REGION_FULL_ACCESS;
MPU_InitStruct.DisableExec = MPU_INSTRUCTION_ACCESS_DISABLE;
MPU_InitStruct.IsShareable = MPU_ACCESS_NOT_SHAREABLE;
MPU_InitStruct.IsCacheable = MPU_ACCESS_CACHEABLE;
MPU_InitStruct.IsBufferable = MPU_ACCESS_NOT_BUFFERABLE;
HAL_MPU_ConfigRegion(&MPU_InitStruct);
/* Enables the MPU */
HAL_MPU_Enable(MPU_PRIVILEGED_DEFAULT);
}
/**
* @brief This function is executed in case of error occurrence.
* @retval None
*/
void Error_Handler(void)
{
/* USER CODE BEGIN Error_Handler_Debug */
/* User can add his own implementation to report the HAL error return state */
__disable_irq();
while (1)
{
}
/* USER CODE END Error_Handler_Debug */
}
#ifdef USE_FULL_ASSERT
/**
* @brief Reports the name of the source file and the source line number
* where the assert_param error has occurred.
* @PAram file: pointer to the source file name
* @PAram line: assert_param error line source number
* @retval None
*/
void assert_failed(uint8_t *file, uint32_t line)
{
/* USER CODE BEGIN 6 */
/* User can add his own implementation to report the file name and line number,
ex: printf("Wrong parameters value: file %s on line %d\r\n", file, line) */
/* USER CODE END 6 */
}
#endif /* USE_FULL_ASSERT */
This is my main.c file main loop. I call cubeai to generate code file app-cube-ai.c file inside the function. The model initialization was successful, and the model parameters were successfully matched. However, there was a delay in the model inference. The code for Caton is: int result = ai_mnetwork_run(network_handle, &input_buf, &output_buf); . I went inside the model inference function and found that Caton's code was if (inn)
return inn->entry->ai_run(inn->handle, input, output);
else
return 0; Please help me
