Struct BatchCostParams 

pub struct BatchCostParams {
    pub c_overhead_us: f64,
    pub c_per_patch_us: f64,
    pub c_latency_us: f64,
    pub total_patches: u64,
}

Parameters for the patch batching cost model.

Fields

c_overhead_us: f64

Per-batch overhead in µs (GPU command buffer setup, draw call).

c_per_patch_us: f64

Per-patch processing cost in µs (cell serialization + copy).

c_latency_us: f64

Latency cost per deferred patch in µs (visual staleness).

total_patches: u64

Total patches to process in the frame.

Implementations

impl BatchCostParams

pub fn total_cost(&self, batch_size: u64) -> f64

Total cost for a given batch size k.

J(k) = ceil(n/k) × (c_overhead + k × c_per_patch) + (k−1) × c_latency

pub fn evaluate(&self, batch_size: u64) -> BatchCostPoint

Evaluate a single point.

pub fn optimal_batch_size(&self) -> u64

Compute optimal batch size.

The continuous optimum from dJ/dk = 0 is:

k* = sqrt(n × c_overhead / c_latency)

Because ceil(n/k) creates discontinuities, the true discrete optimum is found by enumerating candidate batch sizes at all points where ceil(n/k) changes value. There are at most O(sqrt(n)) such candidates.

pub fn optimize(&self) -> BatchCostResult

Run the full optimization and produce evidence.

Trait Implementations

impl Clone for BatchCostParams

fn clone(&self) -> BatchCostParams

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BatchCostParams

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

Formats the value using the given formatter.

impl Default for BatchCostParams

fn default() -> Self

Returns the “default value” for a type.