float16.h

Go to the documentation of this file.

#pragma once
 
namespace audio_tools {
 
class float16 {
 public:
  float16() = default;
  float16(float in) { this->value = float16::float_to_half(in); }
  float16(const float16 &value16) { this->value = value16.value; }
  inline operator float() { return half_to_float(value); }
  explicit inline operator double() {
    return (double)float16::half_to_float(value);
  }
  explicit inline operator int() { return (int)float16::half_to_float(value); }
  explicit inline operator int() const {
    return (int)float16::half_to_float(value);
  }
  inline bool operator<(float16 other) const { return float() < (float)other; }
  inline bool operator<=(float16 other) const {
    return float() <= (float)other;
  }
  inline bool operator>(float16 other) const { return float() > (float)other; }
  inline bool operator>=(float16 other) const {
    return float() >= (float)other;
  }
  inline bool operator==(float16 other) const {
    return float() == (float)other;
  }
  inline bool operator!=(float16 other) const {
    return float() != (float)other;
  }
 
 protected:
  uint16_t value = 0;
 
  static uint32_t as_uint(const float x) { return *(uint *)&x; }
  static float as_float(const uint32_t x) { return *(float *)&x; }
 
  static float half_to_float(
      const uint16_t x) {  // IEEE-754 16-bit floating-point format (without
                           // infinity): 1-5-10, exp-15, +-131008.0,
                           // +-6.1035156E-5, +-5.9604645E-8, 3.311 digits
    const uint32_t e = (x & 0x7C00) >> 10;  // exponent
    const uint32_t m = (x & 0x03FF) << 13;  // mantissa
    const uint32_t v =
        as_uint((float)m) >>
        23;  // evil log2 bit hack to count leading zeros in denormalized format
    return as_float((x & 0x8000) << 16 | (e != 0) * ((e + 112) << 23 | m) |
                    ((e == 0) & (m != 0)) *
                        ((v - 37) << 23 |
                         ((m << (150 - v)) &
                          0x007FE000)));  // sign : normalized : denormalized
  }
  static uint16_t float_to_half(
      const float x) {  // IEEE-754 16-bit floating-point format (without
                        // infinity): 1-5-10, exp-15, +-131008.0,
                        // +-6.1035156E-5, +-5.9604645E-8, 3.311 digits
    const uint32_t b =
        as_uint(x) + 0x00001000;  // round-to-nearest-even: add last bit after
                                  // truncated mantissa
    const uint32_t e = (b & 0x7F800000) >> 23;  // exponent
    const uint32_t m = b & 0x007FFFFF;  // mantissa; in line below: 0x007FF000 =
                                        // 0x00800000-0x00001000 = decimal
                                        // indicator flag - initial rounding
    return (b & 0x80000000) >> 16 |
           (e > 112) * ((((e - 112) << 10) & 0x7C00) | m >> 13) |
           ((e < 113) & (e > 101)) *
               ((((0x007FF000 + m) >> (125 - e)) + 1) >> 1) |
           (e > 143) * 0x7FFF;  // sign : normalized : denormalized : saturate
  }
};
 
inline float operator+(float16 one, float16 two) {
  return (float)one + (float)two;
}
inline float operator-(float16 one, float16 two) {
  return (float)one - (float)two;
}
inline float operator*(float16 one, float16 two) {
  return (float)one * (float)two;
}
inline float operator/(float16 one, float16 two) {
  return (float)one / (float)two;
}
inline float operator+(float16 one, float two) { return (float)one + two; }
inline float operator-(float16 one, float two) { return (float)one - two; }
inline float operator*(float16 one, float two) { return (float)one * two; }
inline float operator/(float16 one, float two) { return (float)one / two; }
inline float operator+(float one, float16 two) { return two + float(one); }
inline float operator-(float one, float16 two) { return two - float(one); }
inline float operator*(float one, float16 two) { return two * float(one); }
inline float operator/(float one, float16 two) { return two / float(one); }
 
}  // namespace audio_tools
 
namespace std {
 
inline float floor(float16 arg) { return std::floor((float)arg); }
inline float fabs(float16 arg) { return std::fabs((float)arg); }
 
}  // namespace std