Struct C32 

#[repr(transparent)]
pub struct C32(pub [f32; 2]);

Local fp8 / fp4 wrappers (#[repr(transparent)] over u8) that satisfy cudarc’s orphan-rule constraint for unsafe impl DeviceRepr. Convertible from/to the backend-agnostic atomr_accel::dtype::* equivalents. 64-bit interleaved complex ({re, im} of f32). Layout matches cufft_sys::float2 and numpy.complex64. Phase 1.5++.

#[repr(transparent)] over [f32; 2] so unsafe impl DeviceRepr is sound (cudarc’s orphan rule blocks blanket impls on tuple-struct shapes from foreign crates) and so transmutes from Vec<C32> to / from Vec<num_complex::Complex<f32>> (also #[repr(C)] with two f32 fields) are layout-safe.

Maps to cudaDataType_t::CUDA_C_32F and cuComplex in CUDA C++.

Tuple Fields

0: [f32; 2]

Implementations

impl C32

pub const fn new(re: f32, im: f32) -> Self

Construct from real / imaginary components.

pub fn re(self) -> f32

pub fn im(self) -> f32

Trait Implementations

impl AccelDtype for C32

const KIND: DType = DType::F32

Runtime discriminant.

const SIZE: usize = 8

Bytes per element including representation padding.

const NAME: &'static str = "complex64"

Human-readable name used in tracing and error messages.

type Scalar = f32

Companion scalar type for host-side parameters (alpha/beta, mean/std, scaling factors). Self for full-precision dtypes; f32 for fp8 / fp4 wrappers because the upstream APIs accept f32 scales.

fn zero() -> Self

fn one() -> Self

fn nan() -> Option<Self>

Some(NaN) for floats, None for integers.

impl Clone for C32

fn clone(&self) -> C32

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl CudaDtype for C32

fn cuda_data_type() -> cudaDataType_t

fn cublas_compute_type() -> cublasComputeType_t

fn cuda_type_name() -> &'static str

CUDA C++ type name for NVRTC source generation.

fn cudnn_data_type() -> cudnnDataType_t

fn nccl_data_type() -> ncclDataType_t

impl Debug for C32

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for C32

fn default() -> C32

Returns the “default value” for a type.

impl From<C32> for float2

Available on crate feature cufft only.

fn from(v: C32) -> Self

Converts to this type from the input type.

impl From<float2> for C32

Available on crate feature cufft only.

fn from(v: float2) -> Self

Converts to this type from the input type.

impl PartialEq for C32

fn eq(&self, other: &C32) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for C32

impl DeviceRepr for C32

impl FftSupported for C32

impl StructuralPartialEq for C32

impl ValidAsZeroBits for C32