Struct GpuProfiler 

pub struct GpuProfiler { /* private fields */ }

Implementations

impl GpuProfiler

pub fn new(context: Arc<GraphicsContext>, max_queries: u32) -> Self

Create a new GPU profiler.

Arguments

• context - Graphics context (must support TIMESTAMP_QUERY)
• max_queries - Maximum number of timestamp queries per frame

Panics

Panics if the device doesn’t support timestamp queries.

pub fn begin_frame(&mut self)

Begin a new frame.

Call this at the start of each frame before recording any regions.

pub fn begin_region( &mut self, encoder: &mut CommandEncoder, label: &str, ) -> Option<ProfileRegion>

Begin a profiling region.

Arguments

• encoder - Command encoder to write the timestamp
• label - Human-readable label for this region

Returns

A ProfileRegion handle that must be passed to end_region.

pub fn end_region( &mut self, encoder: &mut CommandEncoder, region: ProfileRegion, )

End a profiling region.

Arguments

• encoder - Command encoder to write the timestamp
• region - The region handle from begin_region

pub fn resolve(&self, encoder: &mut CommandEncoder)

Resolve all queries from this frame.

Call this after all regions have been recorded, before submitting commands.

pub fn read_results(&mut self) -> &[(String, f64)]

Read profiling results synchronously.

This blocks until the results are available from the GPU. For non-blocking reads, consider using double-buffering or reading results from the previous frame.

Returns

A vector of (label, duration_ms) pairs for each completed region.

pub fn try_read_results(&mut self) -> &[(String, f64)]

Try to read profiling results without blocking.

Returns None if the results are not yet available. This is useful when you want to display results from the previous frame.

Returns

Some reference to the cached results if new data was read, or the existing cached results.

pub fn queries_used(&self) -> u32

Get the number of queries used this frame.

pub fn max_queries(&self) -> u32

Get the maximum queries per frame.

pub fn timestamp_period(&self) -> f32

Get the timestamp period in nanoseconds per tick.

Trait Implementations

impl RenderCapability for GpuProfiler

High-level GPU profiler for measuring execution times.

This profiler uses timestamp queries to measure GPU execution time for different regions of your rendering code.

Requirements

• Device must support TIMESTAMP_QUERY feature
• Must call begin_frame() at the start of each frame
• Must call resolve() before submitting commands

Example

let mut profiler = GpuProfiler::new(context.clone(), 256);

// Each frame:
profiler.begin_frame();

let region = profiler.begin_region(&mut encoder, "My Pass");
// ... do rendering ...
profiler.end_region(&mut encoder, region);

profiler.resolve(&mut encoder);

// Read results (may be from previous frame)
for (label, duration_ms) in profiler.read_results() {
    println!("{}: {:.2}ms", label, duration_ms);
}

fn requirements() -> GpuRequirements

GPU requirements (features + limits) for this component.

fn name() -> &'static str

Human-readable name for diagnostics.