C++ How to convert a vector of size 4 into a 32 bit float

Question

What is the fastest way to convert a vector of size 4 into a 32 bit float?

My failed attempt:

static bool vec2float32(std::vector<uint8_t> bytes, float &result)
{
    if(bytes.size() != 4) return false;
    uint8_t sign = (bytes.at(0) & 0x10000000); //will be 1 or 0
    uint8_t exponent = (bytes.at(0) & 0x01111111);
    uint16_t mantissa = (bytes.at(1) << (2*8)) + (bytes.at(2) << (1*8)) + (bytes.at(3) << (0*8));

    result = (2^(exponent - 127)) * mantissa;
    if(sign == 1) result = result * -1;
    return true;
}

memcpy, all major compilers handle it optimally. Btw, you have bugs in the 0xXXX constants. You used binary, but 0x means hexadecimal. — geza, Jun 29 '18 at 09:42

score 0 · Answer 1 · answered Jun 29 '18 at 09:33

0

From the vector you can get a pointer to the first byte. If the bytes are already in the correct order, then you can copy the bytes directly into a float variable.

answered Jun 29 '18 at 09:33

Some programmer dude

400,186
35
402
621

score 0 · Answer 2 · answered Oct 22 '18 at 18:34

I believe this should be the fastest:

return *reinterpret_cast<float*>(&bytes[0]);

I think its technically undefined behavior. But most compilers should output what you expect here. The &bytes[0] is guaranteed to work because std::vector is guaranteed to be contiguous.

C++ How to convert a vector of size 4 into a 32 bit float

2 Answers2