ClikaRT::dtype::float8_e4m3fnuz_t

struct

Header: ClikaRT/compute/scalar_types.h

FP8 E4M3 FNUZ: 1 sign + 4 exponent + 3 mantissa, bias 8 (rebased). FNUZ = finite, NaN, unsigned-zero: one NaN at 0x80, no infinity, no -0. Range ±240, smallest normal 2^-7, eps 2^-3. 8-bit storage.

Static member functions

from_bits()

static constexpr float8_e4m3fnuz_t from_bits(std::uint8_t b) noexcept

Reinterpret a raw 8-bit encoding without converting.

Declared in ClikaRT/compute/scalar_types.h, line 194

Member functions

float8_e4m3fnuz_t()

constexpr constexpr float8_e4m3fnuz_t() noexcept=default

Declared in ClikaRT/compute/scalar_types.h, line 185

float8_e4m3fnuz_t(float)

explicit float8_e4m3fnuz_t(float f) noexcept

Convert from fp32 (round-to-nearest-even; saturates on overflow).

Declared in ClikaRT/compute/scalar_types.h, line 188

operator float()

operator float() const noexcept

Convert to fp32 (lossless).

Declared in ClikaRT/compute/scalar_types.h, line 191

is_nan()

constexpr bool is_nan() const noexcept

True at the format's single NaN (0x80).

Declared in ClikaRT/compute/scalar_types.h, line 196

Data members

bits

std::uint8_t bits = 0

Raw 8-bit encoding (the storage/ABI form).

Declared in ClikaRT/compute/scalar_types.h, line 183