请教关于STM32F4的浮点单元问题

只看该作者 · 2011-11-24 13:30

本帖最后由 sonicll 于 2011-11-24 13:55 编辑

使用环境为 STM32F4DISCOVERY + MDK4.22

因为看到MDK的options for target选项里，可以选择使用或不使用FPU，想看看使用FPU和不使用FPU在浮点计算上有多大差距，就写了个简单的程序，加减乘除计算，看看每秒钟能计算多少次，把次数和计算结果从串口输出来，代码如下

#include "stm32f4xx.h"
#include <stdio.h>
#include <string.h>

void InitTIM(void);
void InitUart(void);
void UartSendStr(char buf[],int length);

char TIM2_Flag = 0;

int main(void)
{
      unsigned long count = 0;
      float i = 0;
      float j = 0.134245223423424556;
      float k = 0.253252534536366666;

      char buffer[128];

      SystemInit();
      InitTIM();
      InitUart();

      while(1)
      {
            i = (j+5123)*821/(k-456);

            count++;

            if (TIM2_Flag) //TIM2的中断处理函数里把TIM2_Flag置1，周期1秒
            {
                     TIM2_Flag = 0;

                     sprintf(buffer, "%d,%f\n",count,i);

                     UartSendStr(buffer,strlen(buffer));

                     count = 0;
            }
      }
}

void UartSendStr(char buf[],int length)
{
      int i;

      for(i=0;i<length;i++)
      {
            USART_SendData(USART3, buf);
            while (USART_GetFlagStatus(USART3, USART_FLAG_TC) == RESET);
      }
}

void InitUart(void)
{
      GPIO_InitTypeDef  GPIO_InitStructure;
      USART_InitTypeDef USART_InitStructure;

      RCC_AHB1PeriphClockCmd(RCC_AHB1Periph_GPIOD, ENABLE);
      RCC_APB1PeriphClockCmd(RCC_APB1Periph_USART3, ENABLE);

      GPIO_InitStructure.GPIO_Speed = GPIO_Speed_50MHz;
      GPIO_InitStructure.GPIO_Mode = GPIO_Mode_AF;
      GPIO_InitStructure.GPIO_OType = GPIO_OType_PP;
      GPIO_InitStructure.GPIO_PuPd = GPIO_PuPd_UP;

      GPIO_InitStructure.GPIO_Pin = GPIO_Pin_8 | GPIO_Pin_9;
      GPIO_Init(GPIOD, &GPIO_InitStructure);

      GPIO_PinAFConfig(GPIOD, GPIO_PinSource8, GPIO_AF_USART3);
      GPIO_PinAFConfig(GPIOD, GPIO_PinSource9, GPIO_AF_USART3);

      USART_InitStructure.USART_BaudRate = 115200;
      USART_InitStructure.USART_WordLength = USART_WordLength_8b;
      USART_InitStructure.USART_StopBits = USART_StopBits_1;
      USART_InitStructure.USART_Parity = USART_Parity_No;
      USART_InitStructure.USART_HardwareFlowControl = USART_HardwareFlowControl_None;
      USART_InitStructure.USART_Mode = USART_Mode_Rx | USART_Mode_Tx;

      USART_Init(USART3, &USART_InitStructure);
      USART_Cmd(USART3, ENABLE);

}

void InitTIM(void)
{

      TIM_TimeBaseInitTypeDef  TIM_TimeBaseStructure;
      NVIC_InitTypeDef NVIC_InitStructure;

      RCC_APB1PeriphClockCmd(RCC_APB1Periph_TIM2, ENABLE);

      NVIC_InitStructure.NVIC_IRQChannel = TIM2_IRQn;
      NVIC_InitStructure.NVIC_IRQChannelPreemptionPriority = 0;
      NVIC_InitStructure.NVIC_IRQChannelSubPriority = 1;
      NVIC_InitStructure.NVIC_IRQChannelCmd = ENABLE;
      NVIC_Init(&NVIC_InitStructure);

      TIM_TimeBaseStructure.TIM_Period = 41999;
      TIM_TimeBaseStructure.TIM_Prescaler = 1999;
      TIM_TimeBaseStructure.TIM_ClockDivision = TIM_CKD_DIV1;
      TIM_TimeBaseStructure.TIM_CounterMode = TIM_CounterMode_Down;
      TIM_TimeBaseInit(TIM2, &TIM_TimeBaseStructure);

      TIM_Cmd(TIM2, ENABLE);
      TIM_PrescalerConfig(TIM2, TIM_TimeBaseStructure.TIM_Prescaler, TIM_PSCReloadMode_Immediate);
      TIM_ClearFlag(TIM2, TIM_FLAG_Update);
      TIM_ITConfig(TIM2, TIM_IT_Update, ENABLE);

}

MDK的选项里，target选项卡里只选择是否使用FPU，其他选项不选；C/C++选项卡里，优化等级选为0，其他选项都不选

观察结果为：
使用FPU，count为4657245
不使用FPU，count为916266
区别还是比较明显；

接下来是问题，当我把i、j、k三个变量的类型设成double后，结果变成了这样：
使用FPU，count为284680
不使用FPU，count为530620

我知道F4的FPU是单精度的，但是也不应该出现double类型使用FPU，速度反而下降的现象啊，请大家帮忙看看，是程序问题还是MDK有什么设置不对，工程在附件里