Struct KernelPerfCounters 

pub struct KernelPerfCounters {
    pub dispatch_count: u64,
    pub elements_processed: u64,
    pub flop_count: u64,
    pub bytes_read: u64,
    pub bytes_written: u64,
}

Lightweight performance counters attached to a single kernel invocation.

Fields

dispatch_count: u64

Number of times the kernel was dispatched.

elements_processed: u64

Total elements processed across all dispatches.

flop_count: u64

Estimated floating-point operations (MACs counted as 2 FLOPs).

bytes_read: u64

Total bytes read from global memory (mock).

bytes_written: u64

Total bytes written to global memory (mock).

Implementations

impl KernelPerfCounters

pub fn record_dispatch( &mut self, elements: u64, flops: u64, bytes_r: u64, bytes_w: u64, )

Record one dispatch that processed n elements.

pub fn arithmetic_intensity(&self) -> f64

Arithmetic intensity (FLOPs per byte).

pub fn reset(&mut self)

Reset all counters.

Trait Implementations

impl Clone for KernelPerfCounters

fn clone(&self) -> KernelPerfCounters

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for KernelPerfCounters

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

Formats the value using the given formatter.

impl Default for KernelPerfCounters

fn default() -> KernelPerfCounters

Returns the “default value” for a type.