Thread: Bit shifting with float variables

I have to shift bits in a float variable, but it is an illegal operation apperantly. So, I tried to read i nthe binary number into a long int variable, do all the shifting and it works. The problem is that when I try to print the value stored in the long variable it is treated as an int, assigning that value to the float and then printing doesn't work either. Is there another way to either read in the value from the long int into the float without variable type converstion or to bit shift in a float?

Thanks

If you really want to bit-shift a float, you could use a method called fixed-point. Say you want to hold the number 1.23 (and bit-shift it later). You would store 123 in an int, and every time you accessed the variable you would divide the value by 100:

Code:

int i = 123;
i = i << 3;
printf("%f", i/100.0);

I hope this helps.

I am trying to interpret a binary file and unless I put the bits in the rught position and read it as a float, the numbers mean nothing. I know what order they should go in but I don't know what eah number is, so dividing by anything won't work.
Her's the code I have so far, but the convertion from tempValue to Value doesn't work at all. It spits out an incorrect int, loosing all the decimal places.

Thanks for the replies so far.

Code:

# include <stdio.h>
# define BYTE21 0X0000ffff
# define BYTE43 0Xffff0000

int
main (void)
{
	FILE *inp; 
	
	
    float Value; 
    long TempValue;
    long new21, new43;

	inp = fopen("test_outp.baf", "rb");

	if(!inp)
		printf("The file could not be opened.\n");

	fseek(inp, sizeof(char)*3116,SEEK_SET); 


	fread(&TempValue,sizeof(long),1,inp);

	new21 = (TempValue & BYTE21) << 16;
	new43 = (TempValue & BYTE43) >> 16;

	TempValue = new21 | new43;

	Value = TempValue;
		
	printf("%f \n", Value);  

    return (0);

}

Originally Posted by quzah

You can't use bitwise operations on floating point numbers. Why must you use a float again? You were a bit vague.

Quzah.

I have to read in real numbers. If for say the 4 bytes I'm reading contain 5.25 only a float will read them properly, other variables, ex. int , wil read the binary 5.25 as 1034147594, becasue they don't know that the last 23 bits are the decimal part of the number. And I have to bit-shift. Fun, I know.

Say the value in the file is a3 3d 0a d7. To interpret correctly it should be read as 3d a3 d7 0a. If I scan in directly from file it means nothing. If I read it into a long int then bit-shift, put the bytes in the right order. if I fprint the in value it prints as 1034147594, if I assign the value in the int to a float variable it prints as 1034147594.0000000.
I tried scanning in a random real number into the float and hten printing it, it works, but without bit-shifting.

Originally Posted by Dave_Sinkula

After the bit-shifting of the integer, get a float value by casting the integer's address to a float* and dereferencing it.

Code:

#include <stdio.h>

int main (void)
{
   float value;
   unsigned long temp = 0xD70A3DA3; /* endianness */
   printf("temp = %08X\n", temp);
   temp = ((temp & 0xFFFF) << 16) | ((temp & 0xFFFF0000) >> 16);
   printf("temp = %08X\n", temp);
   value = *(float*)&temp;
   printf("value = %g\n", value);
   return 0;
}

/* my output
temp = D70A3DA3
temp = 3DA3D70A
value = 0.08
*/

It works like magic. I tried that before but I couldn't figure out how to cast and what to cast it to properly.
I actually wrote another function to swap the bytes in the float. A pointer to the float is passed to it and it swaps the bytes using poiter addresses and returns a revered float.

thanks a lot though,

A.

oh ok. i should have realized that.