Struct GpuArch 

pub struct GpuArch {
    pub base: usize,
    pub suffix: Option<String>,
}

GPU architecture specification.

Supports both numeric (80, 90) and string-based (90a, 100a) formats.

Fields

base: usize

Base compute capability number (e.g., 90, 100, 120).

suffix: Option<String>

Optional suffix for accelerated variants (e.g., “a” for async).

Implementations

impl GpuArch

pub fn new(base: usize) -> Self

Create a new GPU architecture from base number.

pub fn with_suffix(base: usize, suffix: &str) -> Self

Create a new GPU architecture with suffix (e.g., 90a, 100a).

pub fn parse(s: &str) -> Result<Self>

Parse from string like “90”, “90a”, “100a”, “sm_90a”.

If no suffix is provided (e.g., “90”), auto-suffix is applied for sm_90+. To explicitly disable the suffix, use the numeric API directly.

pub fn auto_suffix(base: usize) -> Self

Create GPU arch with auto-detected suffix for newer architectures.

• >= sm_120 gets “f” suffix for f8f6f4.mma instructions (CUDA 12.9+).
• >= sm_90 gets “a” suffix for async / accelerated features.

pub fn to_nvcc_arch(&self) -> String

Get the nvcc --gpu-architecture string (e.g., “sm_90a”, “sm_80”).

pub fn to_gencode_arg(&self) -> String

Get the nvcc -gencode argument (e.g., -gencode=arch=compute_90a,code=sm_90a).

pub fn base(&self) -> usize

Get the base compute capability number.

Trait Implementations

impl Clone for GpuArch

fn clone(&self) -> GpuArch

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GpuArch

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

Formats the value using the given formatter.

impl Display for GpuArch

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

Formats the value using the given formatter.

impl Eq for GpuArch

impl From<usize> for GpuArch

fn from(base: usize) -> Self

Converts to this type from the input type.

impl Hash for GpuArch

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for GpuArch

fn eq(&self, other: &GpuArch) -> 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 StructuralPartialEq for GpuArch