Reinterpret bytes as float in C (IEEE 754 single-precision binary)

Question

I want to reinterpret 4 bytes as IEEE 754 single-precision binary in C.

To obtain the bytes that represent float, I used:

num = *(uint32_t*)&MyFloatNumber;
aux[0] = num  & 0xFF;
aux[1] = (num >> 8) & 0xFF;
aux[2] = (num >> 16)  & 0xFF;
aux[3] = (num >> 24)  & 0xFF;

num is a uint32_t.
aux[] is int[4].

To reinterpret the bytes as a float, I’m trying:

Buff = *(float*)&aux;

In that second case nothing apears on "Buff"

Buff is a float.

What am I doing wrong in the second case?

Hmmm `aux[]` is`int[4];` and code is using that to `convert BYTE to FLOAT`?? — chux - Reinstate Monica, Jan 02 '18 at 22:07
Is this the [XY problem?](https://meta.stackexchange.com/questions/66377/what-is-the-xy-problem)? — Weather Vane, Jan 02 '18 at 22:10
Neither of these conversions does a useful thing--I suspect you actually do have some meaningful conversion in mind, but I can't decipher it from this question. So what are you actually trying to accomplish? — Lee Daniel Crocker, Jan 02 '18 at 22:18

chux - Reinstate Monica · Answer 1 · 2018-01-02T22:32:23.577

3

2 problems:

Buff = *(float*)&aux; attempts to use the address of an array of 4 int as a pointer to a float. aux[] is perhaps 16 bytes long and a IEEE 754 single-precision binary float is expected to be 4 bytes.

Both casts: (uint32_t*) and (float*) invoke undefined behavior as well as alignment and anti-aliasing issues. Better to use a union.

int main(void) {
  union {
    float f;
    unsigned char uc[sizeof(float)];
  } x;

  // float to bytes
  x.f = 1.23f;
  printf("%x %x %x %x\n", x.uc[0], x.uc[1], x.uc[2], x.uc[3]);

  // bytes to float
  x.uc[0] = 0xA4;
  x.uc[1] = 0x70;
  x.uc[2] = 0x9D;
  x.uc[3] = 0x3F;
  printf("%.8e\n", x.f);
}

Output

a4 70 9d 3f
1.23000002e+00

edited Jan 02 '18 at 22:32

answered Jan 02 '18 at 22:12

chux - Reinstate Monica

143,097
13
135
256

I'm using a microcontroller that does not allow the use of the printf function, you would have another way to do the conversion (bytes to float)? – Guilherme Jan 09 '18 at 00:07
@Guilherme Even without `printf("%.8e\n", x.f);`, the converted `float` is in `x.f`. `printf("%.8e\n", x.f);` is only for display purposes. – chux - Reinstate Monica Jan 09 '18 at 01:56

Reinterpret bytes as float in C (IEEE 754 single-precision binary)

1 Answers1