Enable FPU in LM4F232?

Hi,

Look in StellarisWare/boards/ek-lm4f323/sine_demo project - it is a good starting point. 

Petrei.

Hi there,

Thats what I did, in the beginning they enable the FPU with FPUEnable() and they choose for the lazystackingmethod.

So I did the same in my software but I still can't use any floats. Thats why I'm here.

Is there maybe something else that I have to set or something like that?

Ilias Mansouri said:

but nothing works...but still nothing works.

Somewhat difficult to diagnose your problems with so complete a description...  (and you took pains to repeat!)

Do you fault, get warnings etc, etc...

Usually - when floating point enters scene - stack size must substantially increase...

As Petrei guides - you must look intensely @ successful/model project - and insure that you've "modeled" that in your project.   (or - stay w/working model - and add your bits/pieces...)

Sorry for repeating myself, I'm a bit tired I gues. I didn't want to include my whole code because it is quite a piece. I'll try to filter unnecessary pieces:

int main(void)
{
tCANMsgObject sCANMessage;
unsigned char ucMsgData[6];
int temp = 0;
int i,x = 0;

unsigned int DATA_MSB, DATA_LSB;
unsigned int temp_result=0;
float temp_result_scaled=0;
float test_float = 0;

// Set the clocking
SysCtlClockSet(SYSCTL_SYSDIV_1 | SYSCTL_USE_OSC | SYSCTL_OSC_MAIN | SYSCTL_XTAL_16MHZ);

FPUEnable();
ROM_FPULazyStackingEnable();

// Inits, configures and irrelevant code

test_float = 15/2;

UARTprintf("test_float: %f \n", test_float);

The two last lines are my problem basicaly. Whenever I try to use a float, I get an ERROR on my terminal:

test_float: ERROR

So if I understand it correctly, I have to write this in my main:

ROM_FPULazyStackingEnable();
ROM_SysCtlClockSet(SYSCTL_SYSDIV_4 | SYSCTL_USE_PLL | SYSCTL_XTAL_16MHZ | SYSCTL_OSC_MAIN);

In the startup_ccs:

Enable the Systickhandler and update it in my main.c

And for the stack I'm not to sure, do I have to go to the ccs.cmd and edit the __STACK_TOP?

And there is that assembly code that I found in the datasheet:

asm(" MOVW R0, #0xE000 \n"
" MOVT R0, #0xED88 \n"
" LDR R1, [R0]\n"
" ORR R1, R1, #(0xF << 20)\n"
" STR R1, [R0]\n"
" DSB\n"
" ISB");

It seems to do the same as those FPU api's, can I use it instead?

Ilias Mansouri said:

I am trying to use floating point units

Before moving on to that, how about finishing the thread that you started here: http://e2e.ti.com/support/microcontrollers/tiva_arm/f/908/t/259190.aspx

I am terribly sorry, I've completely forgot about that thread.
Many thanks for pointing it out,  the last weeks are terribly charged here so it completely slipped out of my mind.

I deeply apology 

float test_float = 0.00

Sub 4 word post brings you "beyond close" to heralded "next level" status - your skill/efforts hereby memorialized and appreciated...

I already tried, that doesn't do the trick.
What I think is that the FPU still isn't enabled, this is how I've done it:

void SysTickHandler(void)
{
 //
 // Update our tick counter.
 //
 g_ulTickCount++;
}

int main(void)

{

 float temp_result=0;
 float temp_result_scaled=0;
 float test = 0;
 // Set the clocking
 ROM_FPULazyStackingEnable();

 ROM_SysCtlClockSet(SYSCTL_SYSDIV_4 | SYSCTL_USE_PLL | SYSCTL_XTAL_16MHZ | SYSCTL_OSC_MAIN);
 //
 // Configure SysTick to generate a periodic time tick interrupt.
 //
 ROM_SysTickPeriodSet(ROM_SysCtlClockGet() / TICKS_PER_SECOND);
 ROM_SysTickEnable();
 ROM_SysTickIntEnable();

 // Set up the serial console to use for displaying messages.
 InitConsole();

// Code

temp_result = Variable // Variable is an integer between 0 and 255

temp_result_scaled = Scale_255(temp_result); // Scale Variable to 0 - 100%

UARTprintf("Scaled Result: %f  \n", temp_result_scaled);   // Error       

sprintf(str, " %f", temp_result_scaled);            // Solution to UARTprintf that Bobby kindly suggested

UARTprintf("String scaled: %s  \n", str); // If Variable = 255 then %s = 100 else %s = 0

}

/*Scaling function */

float Scale_255 (unsigned int value)
{

 float temp = 0;
 char str[80];

unsigned long ulLastTickCount=0;

 // Wait for the next timer tick.
 
 while(ulLastTickCount == g_ulTickCount)
 {
 }
 ulLastTickCount = g_ulTickCount;
 temp = (value/255)*100;

return temp;

}

I've found the error.

It was a combination of multiple errors:

- printf doesn't print floats -> thanks  Bobby
- Stacksize wasn't properly set -> thanks  cb1_mobile
- FPU wasn't enabled -> thanks Petrei

Thanks guys, I've spent more than 9hours straight searching for those stupid errors. 
I could'nt have found them without your precious help.

Sincerely,

Ilias 

Rare to see "helped poster" respond w/such gratitude - thank you/appreciated.

You of course get "lion's share" of credit for persisting - and willingness to accept (and try) suggestions.

Our group finds that - most always - your efforts will speed, ease and enhance by first "searching for & finding" nearest applicable code-file example.  Best if you can "duplicate this file" (leave original intact/unmodified) and slowly/cautiously add "one bit at a time" (to the duplicated code-file) to extend that program to your specific requirement. 

Bon chance mon ami - merci... 

Ilias Mansouri said:

printf doesn't print floats ->

printf() does print floats.  But you were using UARTprintf - which does not!

UARTprintf  is specifically advertised & documented as a simplified, light-weight version of printf - and one of the simplifications is the omission of support for floats!

cb1_mobile,

Thank you for the accolades.  It is an honor to walk amongst you.

@greenja,

Honor all mine - great to have you back in this Stellaris (moment of silence) re-brand fold...

(past - many BCal flights: LAX - Gatwick (direct) -> Biggin Hill)  Good times...Again Congrats & Cheers...