STM32 runın at 16 mhz

[FaintSmile]

Posted on December 14, 2013 at 09:32

I wrote 1 sn delay code for stm32 and ı used 168 mhz system clock.My delay cycle ıs 16000 for each ms and ıt work led blınk 1 sn.It dıdn't work when ıt was 168000 shouldn't it have worked at 168000 cycle.my code ıs

****  File        : main.c****  Abstract    : main function.****  Functions   : main****  Environment : Atollic TrueSTUDIO(R)**                STMicroelectronics STM32F4xx Standard Peripherals Library****  Distribution: The file is distributed “as is,�? without any warranty**                of any kind.****  (c)Copyright Atollic AB.**  You may use this file as-is or modify it according to the needs of your**  project. This file may only be built (assembled or compiled and linked)**  using the Atollic TrueSTUDIO(R) product. The use of this file together**  with other tools than Atollic TrueSTUDIO(R) is not permitted.********************************************************************************//* Includes */#include ''stm32f4xx.h''/* Private macro *//* Private variables *//* Private function prototypes *//* Private functions *//****===========================================================================****  Abstract: main program****===========================================================================*/void delay_ms(int delay1){ delay1 = delay1*16000; while (--delay1) { }}int main(void){ GPIO_InitTypeDef GPIO_InitStruct; RCC_AHB1PeriphClockCmd(RCC_AHB1Periph_GPIOD, ENABLE); GPIO_InitStruct.GPIO_Pin = GPIO_Pin_15 | GPIO_Pin_14 | GPIO_Pin_13 | GPIO_Pin_12; // we want to configure all LED GPIO pins GPIO_InitStruct.GPIO_Mode = GPIO_Mode_OUT; // we want the pins to be an output GPIO_InitStruct.GPIO_Speed = GPIO_Speed_100MHz; // this sets the GPIO modules clock speed GPIO_InitStruct.GPIO_OType = GPIO_OType_PP; // this sets the pin type to push / pull (as opposed to open drain) GPIO_InitStruct.GPIO_PuPd = GPIO_PuPd_NOPULL; // this sets the pullup / pulldown resistors to be inactive GPIO_Init(GPIOD, &GPIO_InitStruct);  // this passes the configuration to the Init function which takes care of the low level stuff  /* Infinite loop */  while (1)  { // if(1 == GPIO_ReadInputDataBit( GPIOA,GPIO_Pin_0)) // { // while(1 == GPIO_ReadInputDataBit( GPIOA,GPIO_Pin_0));  GPIO_SetBits ( GPIOD, GPIO_Pin_14) ;  delay_ms(1000);  GPIO_ResetBits (GPIOD, GPIO_Pin_14) ;  delay_ms(1000); // }  }}/* * Callback used by stm32f4_discovery_audio_codec.c. * Refer to stm32f4_discovery_audio_codec.h for more info. */void EVAL_AUDIO_TransferComplete_CallBack(uint32_t pBuffer, uint32_t Size){  /* TODO, implement your code here */  return;}/* * Callback used by stm324xg_eval_audio_codec.c. * Refer to stm324xg_eval_audio_codec.h for more info. */uint16_t EVAL_AUDIO_GetSampleCallBack(void){  /* TODO, implement your code here */  return -1;}

[Tesla DeLorean]

Posted on December 14, 2013 at 13:16

If you are concerned about the processor speed, use the MCOx pins and output internal clocks, and measure them on a scope.

Observe that the processor in fact starts for a 16 MHz HSI clock, and has to be switched to something else, typically via SystemInit() in system_stm32f4xx.cFinally look at the assembler code generated by the compiler, and the optimization level selected. Unless you're writing code in assembler you probably have little control of cycle level performance.

[FaintSmile]

Posted on December 15, 2013 at 09:11Clive you are right but why ıs ıt happening for  exapmle when ı use systemclick for delays lle these  if (SysTick_Config(168000000 / 1000))  {    /* Capture error */    while (1);  }Delay(1000);it does really waits for 1 second but when ı use my 1 second delay code lke these

void delay_s

 

(int delay2)

{ delay2 = delay2*168

0

0

0;

while (delay2--) { }} Delay(1000);it waits more than 10 second ı dıdn't understant why it waits more.ısn't ıt te same thing its like my main fuction has a diffrent clock speed from systemclock and it is like 8mhz or smtn like that

[Tesla DeLorean]

Posted on December 15, 2013 at 15:00

The processor doesn't natively run C code, so assuming it's going to take a single cycle per loop iteration probably isn't going to hold water. To understand where the 10 cycles go you'll want to review the code output by the compiler.

[Andrew Neil]

Posted on December 16, 2013 at 01:16

''The processor doesn't natively run C code, so assuming it's going to take a single cycle per loop iteration probably (sic?) isn't going to hold water''

I'd say it is certainly  not going to hold water!In fact, any assumption - unless fully  qualified by compiler, version, and complete settings - about a direct correspondence from 'C' source lines to machine code is unlikely to hold water.

http://www.8052.com/forum/read/162556