Enum MaterialBindGroupAllocator 

pub enum MaterialBindGroupAllocator {
    Bindless(Box<MaterialBindGroupBindlessAllocator>),
    NonBindless(Box<MaterialBindGroupNonBindlessAllocator>),
}

A resource that places materials into bind groups and tracks their resources.

Internally, Bevy has separate allocators for bindless and non-bindless materials. This resource provides a common interface to the specific allocator in use.

Variants

Bindless(Box<MaterialBindGroupBindlessAllocator>)

The allocator used when the material is bindless.

NonBindless(Box<MaterialBindGroupNonBindlessAllocator>)

The allocator used when the material is non-bindless.

Implementations

impl MaterialBindGroupAllocator

pub fn new( render_device: &RenderDevice, label: &'static str, bindless_descriptor: Option<BindlessDescriptor>, bind_group_layout: BindGroupLayoutDescriptor, slab_capacity: Option<BindlessSlabResourceLimit>, ) -> MaterialBindGroupAllocator

Creates a new MaterialBindGroupAllocator managing the data for a single material.

pub fn get(&self, group: MaterialBindGroupIndex) -> Option<MaterialSlab<'_>>

Returns the slab with the given index, if one exists.

pub fn allocate_unprepared( &mut self, unprepared_bind_group: UnpreparedBindGroup, bind_group_layout: &BindGroupLayoutDescriptor, ) -> MaterialBindingId

Allocates an UnpreparedBindGroup and returns the resulting binding ID.

This method should generally be preferred over Self::allocate_prepared, because this method supports both bindless and non-bindless bind groups. Only use Self::allocate_prepared if you need to prepare the bind group yourself.

pub fn allocate_prepared( &mut self, prepared_bind_group: PreparedBindGroup, ) -> MaterialBindingId

Places a pre-prepared bind group into a slab.

For bindless materials, the allocator internally manages the bind groups, so calling this method will panic if this is a bindless allocator. Only non-bindless allocators support this method.

It’s generally preferred to use Self::allocate_unprepared, because that method supports both bindless and non-bindless allocators. Only use this method if you need to prepare the bind group yourself.

pub fn free(&mut self, material_binding_id: MaterialBindingId)

Deallocates the material with the given binding ID.

Any resources that are no longer referenced are removed from the slab.

pub fn prepare_bind_groups( &mut self, render_device: &RenderDevice, pipeline_cache: &PipelineCache, fallback_bindless_resources: &FallbackBindlessResources, fallback_image: &FallbackImage, )

Recreates any bind groups corresponding to slabs that have been modified since last calling MaterialBindGroupAllocator::prepare_bind_groups.

pub fn write_buffers( &mut self, render_device: &RenderDevice, render_queue: &RenderQueue, )

Uploads the contents of all buffers that this MaterialBindGroupAllocator manages to the GPU.

Non-bindless allocators don’t currently manage any buffers, so this method only has an effect for bindless allocators.

pub fn slab_count(&self) -> usize

Get number of allocated slabs for bindless material, returns 0 if it is Self::NonBindless.

pub fn slabs_size(&self) -> usize

Get total size of slabs allocated for bindless material, returns 0 if it is Self::NonBindless.

pub fn allocations(&self) -> u64

Get number of bindless material allocations in slabs, returns 0 if it is Self::NonBindless.