Struct GpuCollisionPipeline 

pub struct GpuCollisionPipeline {
    pub broadphase: GpuBroadphase,
    pub narrowphase: GpuNarrowphase,
    pub contact_cache: GpuContactCache,
    pub aabb_tree: GpuAabbTree,
    pub contacts: Vec<ContactResult>,
    pub manifolds: Vec<PersistentManifoldGpu>,
    pub cumulative_stats: CollisionKernelStats,
}

Full GPU collision pipeline: broadphase (SAP) + narrowphase (GJK) + persistent contact cache with warmstarting.

Intended as the primary collision API. Configure once, call step each physics tick to get a fresh contact list.

Fields

broadphase: GpuBroadphase

SAP broadphase.

narrowphase: GpuNarrowphase

GJK narrowphase.

contact_cache: GpuContactCache

Persistent contact cache.

aabb_tree: GpuAabbTree

AABB tree (rebuilt each frame or on-demand).

contacts: Vec<ContactResult>

Contact list from the most recent step.

manifolds: Vec<PersistentManifoldGpu>

Persistent manifolds.

cumulative_stats: CollisionKernelStats

Cumulative stats since pipeline creation.

Implementations

impl GpuCollisionPipeline

pub fn new(margin: f32, max_cache_age: u32) -> Self

Create a new GpuCollisionPipeline with sensible defaults.

pub fn step(&mut self, aabbs: &[AabbGpu])

Run one full collision step using SAP broadphase + GJK narrowphase.

Steps:

1. Rebuild the AABB tree.
2. Run SAP broadphase.
3. Run GJK narrowphase on each pair.
4. Refresh contact cache; apply warmstart data.
5. Update persistent manifolds.
6. Age out stale cache entries.

pub fn reset(&mut self)

Clear all contacts, manifolds, and the contact cache.

pub fn total_contact_points(&self) -> usize

Total active contact points across all manifolds.

pub fn frame_stats(&self) -> CollisionKernelStats

Get a snapshot of per-frame broadphase + narrowphase stats.

Trait Implementations

impl Debug for GpuCollisionPipeline

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

Formats the value using the given formatter.