I have a 4byte hex number, when I convert it to float it is -0.5(I checked references and it's correct), but printf gives very wrong value:
int f=0xBF000000; printf("%f",(float)f); // prints -1090519040.000000 I can't figure out what is wrong here! All calculators give the correct value as -0.5, but printf gives above value.
Your int has the value -1090519040. If you print it you get this value:
printf("%d", f); // -1090519040 If you typecast it to float, it will take this value and change the type but not the value. So it will result in -1090519040.0000.
If you actually want to interpret your int as a float, you can use a pointer:
float* p = &f; printf("%f",*p); // -0.500000 Here you have a float pointer that points to the address of your integer and if you print it it will interpret the bits as a float.
Comment from Eric Postpischil:
float *p = &f;is not a proper way to reinterpret the bytes of an object in C, as it violates the aliasing rule (C 2011 [N1570] 6.5 7, an object’s value shall be accessed only through its effective type, a character type, or certain others).
Another (better) way would be to use memcopy and copy your int into another float:
int a = 0xBF000000; float b = 0; memcpy(&b, &a, 4); printf("%f", b); // -0.500000 
3204448256 and it is implementation defined to which value this is converted if it doesn't fit into an int. C has no negative constants.
float *p = &f; is not a proper way to reinterpret the bytes of an object in C, as it violates the aliasing rule (C 2011 [N1570] 6.5 7, an object’s value shall be accessed only through its effective type, a character type, or certain others). Proper ways are to copy the bytes (as with the memcpy shown in this answer) or to reinterpret through a union (6.5.2.3 3).
If you like the access to every single Byte of your float-value you could use a union.
typedef union { struct { uint8_t Byte1; uint8_t Byte2; uint8_t Byte3; uint8_t Byte4; }; float_t FloatValue; } myFloatT; 
printf("%f",*(float*)(&f));)