Struct EpistemicGpuKernelTimingTrace 

pub struct EpistemicGpuKernelTimingTrace {
    pub cuda_event_pairs: u32,
    pub timing_sync_ops: u32,
    pub kernel_elapsed_nanos: u64,
}

CUDA-event timing captured around one epistemic GPU kernel launch.

Fields

cuda_event_pairs: u32

CUDA event pairs recorded around the launch. Runtime traces require one.

timing_sync_ops: u32

CUDA event synchronizations used to make elapsed time observable on host.

kernel_elapsed_nanos: u64

Event-measured stream elapsed time, converted from milliseconds to nanoseconds.

Implementations

impl EpistemicGpuKernelTimingTrace

pub const fn unrecorded() -> Self

Empty timing marker used before a runtime launch records CUDA events.

pub fn from_cuda_elapsed_ms(elapsed_ms: f32) -> Result<Self>

Convert CUDA’s native event elapsed time in milliseconds to a trace.

pub const fn is_recorded(&self) -> bool

Whether CUDA-event timing was recorded for this trace.

pub fn saturating_add(self, other: Self) -> Self

Saturating sum used when aggregating multi-kernel or split-batch traces.

pub fn checked_add(self, other: Self) -> Result<Self>

Checked sum used by accepted certification paths.

pub fn sum(traces: impl IntoIterator<Item = Self>) -> Self

Aggregate timing traces from a single execution or split-batch result.

pub fn checked_sum(traces: impl IntoIterator<Item = Self>) -> Result<Self>

Checked aggregate timing traces for accepted certification paths.

Trait Implementations

impl Clone for EpistemicGpuKernelTimingTrace

fn clone(&self) -> EpistemicGpuKernelTimingTrace

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for EpistemicGpuKernelTimingTrace

impl Debug for EpistemicGpuKernelTimingTrace

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

Formats the value using the given formatter.

impl Default for EpistemicGpuKernelTimingTrace

fn default() -> EpistemicGpuKernelTimingTrace

Returns the “default value” for a type.

impl Eq for EpistemicGpuKernelTimingTrace

impl PartialEq for EpistemicGpuKernelTimingTrace

fn eq(&self, other: &EpistemicGpuKernelTimingTrace) -> 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 EpistemicGpuKernelTimingTrace