Struct Event

pub struct Event { /* private fields */ }

Wrapper around a CUDA event.

CUDA events are synchronization markers that can be used to monitor the device’s progress, to accurately measure timing, and to synchronize CUDA streams.

The underlying CUDA events are lazily initialized when the event is first recorded or exported to another process. After creation, only streams on the same device may record the event. However, streams on any device can wait on the event.

See the docs of torch.cuda.Event for more details.

Implementations

impl Event

pub fn new() -> Self

Create a new event.

pub fn record(&mut self, stream: Option<&Stream>)

Record the event on the current stream.

Uses the current stream if no stream is provided.

pub fn wait(&mut self, stream: Option<&Stream>)

Make all future work submitted to the given stream wait for this event.

Uses the current stream if no stream is specified.

pub fn query(&self) -> bool

Check if all work currently captured by event has completed.

pub fn elapsed_time(&self, end_event: &Event) -> Duration

Return the time elapsed.

Time reported in after the event was recorded and before the end_event was recorded.

pub fn synchronize(&self)

Wait for the event to complete. Waits until the completion of all work currently captured in this event. This prevents the CPU thread from proceeding until the event completes.

Trait Implementations

impl Debug for Event

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

Formats the value using the given formatter.