CODEWITHSUNDEEP
cbinaryieee-754ieee
Casey
Casey

Reputation: 552

Stubborn bit won't be set to one.

I have this code I use for getting the mantissa, or value, of an IEEE binary number. 

iFloat_t floatGetVal (iFloat_t x) {
  iFloat_t mantissa = (BITS == 16) ? (x & 0x03FF)
                      : (x & 0x007FFFFF);
    debug("%s: getVal before implicit 1", getBinary(mantissa));
    //mantissa = (BITS == 16) ? (mantissa | 0x04)
    //                : (mantissa | 0x008);
    mantissa = x | 0000010000000000;
    debug("%s: getVal after implicit 1", getBinary(mantissa));
    mantissa = (BITS == 16) ? (mantissa & 0x07FF)
                      : (mantissa & 0x00FFFFFF);
    if(floatGetSign(x) == 1) {
        mantissa = ~mantissa + 1;
    }
    return mantissa;
}

My issue is when I try to get the value from the number 63.125, here is the corresponding output:

DEBUG iFloat.c[31] floatGetVal() 0000-0011-1110-0100: getVal before implicit 1
DEBUG iFloat.c[35] floatGetVal() 0101-0011-1110-0100: getVal after implicit 1
DEBUG iFloat.c[81] floatAdd() 0000-0011-1110-0100: bits of val_y after assignment

This is my expected output:

DEBUG iFloat.c[31] floatGetVal() 0000-0011-1110-0100: getVal before implicit 1
DEBUG iFloat.c[35] floatGetVal() 0101-0111-1110-0100: getVal after implicit 1
DEBUG iFloat.c[81] floatAdd() 0000-0111-1110-0100: bits of val_y after assignment

Here is my full code:

https://pastebin.com/w4UVzmUe

Upvotes: 0

Views: 143

Answers (2)

zch
zch

Reputation: 15278

0000010000000000 is an octal literal, not a binary one, use hexadecimal 0x0400.

Upvotes: 7

P.W
P.W

Reputation: 26800

There are no binary constants in standard C. So you cannot have something like 0b0000010000000000. But GCC has an extension to support them with 0b or 0B prefixes.

If you are not using GCC, then better go with the suggestion provided by @zch.

Upvotes: 1

Related Questions