Skip to main content

Module gpu

Module gpu 

Source
Expand description

WebGPU renderer (PortalGpu) — depth, bloom, tensor-node projection, mesh surfaces. Cross-platform WebGPU viewport for the Qualia renderer SDK.

Phenomenal viewport: projector (depth write) → ambient → optional T2 Kawase bloom.

Re-exports§

pub use holographic::HolographicNodeGpu;
pub use holographic::HolographicPrimitive;
pub use mesh_slab::MeshSlabArenaGpu;
pub use mesh_slab::ProceduralSynthesisGpu;
pub use frame_lock::*;

Modules§

emf_cue_bridge
GPU Uniform/Voxel Packer for EMF, Acoustic & Physical Field Cues (BS-40).
frame_lock
Deterministic 64-Byte UDP Multicast Frame-Lock & Conical Frustum Slicing Protocol.
holographic
Holographic 3D Engine for Semantic Graphs — Procedural Volumetric Light Field.
ibl_cubemap
Zero-Heap Procedural IBL Cubemap Generator & GGX Specular Prefilter (Pure Rust).
indirect_cull
Canonical indirect visibility, HZB and multi-representation stream delegation.
mesh_slab
Semantic procedural header adapter and universe admission for canonical slab residency.
residency_budget
Residency metadata is maintained by Etalon; Qualia owns resource policy and lifetime.

Structs§

AlphaCoverageDiagnostics
Bounded prediction for cutoff coverage after RGBA8 quantization of each generated mip.
ClusterGridParamsGpu
112-byte uniform block for CLUSTER_LIGHT_WGSL (16 x 9 x 24 = 3456 frustum clusters).
ClusterLightPipelineGpu
Active GPU compute pipeline owner for 3D Clustered Light Culling (CLUSTER_LIGHT_WGSL).
ComputeBinding
One buffer binding descriptor for a compute dispatch.
Csm4ShadowUniform
4-Cascade Cascaded Shadow Map (CSM) GPU uniform block (288 bytes) with practical split distances and smooth inter-cascade blend band parameters.
CullUniformsGpu
Uniform buffer layout for instance_cull_indirect.wgsl (288 bytes).
DeferredSpectralResolvePipelineGpu
Active GPU compute pipeline owner for Continuous EMF Spectral Deferred Resolve (DEFERRED_RESOLVE_WGSL).
DielectricVolumeIopGpu
64-byte Inherent Optical Properties (IOP) descriptor mirroring DielectricVolumeIop in HYDROLOGIC_OPTICS_WGSL.
DrawIndexedIndirectCommand
WebGPU-standard 20-byte DrawIndexedIndirect command payload.
EmfFieldCell
Per-cell field data uploaded to the GPU storage buffer.
EmfPipeline
Complete, decoupled GPU state for EMF wave slice visualization.
EmfPipelineDesc
High-level descriptor for constructing the EMF visualization pipeline.
EmfSliceUniform
Slice parameter uniform (64 bytes, std140-aligned).
FoliageLodPipelineGpu
Active GPU compute pipeline owner for Continuous Multi-Scale Foliage LOD (FOLIAGE_LOD_WGSL).
HolographicPipelineGpu
WebGPU pipeline wrapper for the Holographic 3D Engine.
HzbPyramidView
Lightweight zero-heap view over a 2-phase Hierarchical Z-Buffer (HZB) mip chain.
IblCubemapGenerator
GPU-resident cubemap IBL generator and prefiltered environment probe state.
IblCubemapParams
Uniform parameters for IblCubemapGenerator (64 bytes).
IndirectCullPipelineGpu
GPU-resident compute pipeline and indirect stream buffers for INSTANCE_CULL_INDIRECT_WGSL.
IndirectCullStreamCounts
Output stream counts produced by cull_and_select_lod_indirect_zero_heap.
InstanceCullDescriptor
GPU-aligned per-instance culling & multi-modal routing descriptor (80 bytes, 5 x vec4).
MeshSlabArenaDescriptor
Capacity configuration for MeshSlabArenaGpu.
MeshSlabVertex
64-byte interleaved vertex record inside MeshSlabArenaGpu::vertex_buffer.
PortalGpu
ResidentAssetEntry
112-byte multi-modal resident asset entry tracking 4 LOD tiers and .10d PROC/EONT/SURF/IMPO bindings.
ShadowAtlasAllocator
Zero-heap tile allocator for a square spot/point light shadow atlas (e.g. 2048x2048 split into grid_dim x grid_dim tiles).
SkinAttrVertex
24-byte skeletal skinning vertex attribute record (JOINTS_0 + WEIGHTS_0).
SlabRange
Contiguous free span [offset .. offset + length) inside a GPU buffer slab.
SlabRangeAllocator
Zero-heap first-fit range sub-allocator with immediate bidirectional coalescing.
SpotPointShadowAtlasSlot
Tile slot in a local spot/point light shadow atlas.
SubmeshSlice
TemporalProducerAvailability
Explicit availability report for the renderer/host temporal handoff.
TemporalProducerViews
Producer views supplied by a host for one temporal frame.
TemporalResolveGpu
TemporalResolveInputs
Real producer views required to admit a temporal resolve.

Enums§

AssetRepresentationKind
Typed representation archetype for a LOD tier in ResidentAssetEntry.
ComputeBufferKind
Kind of buffer to bind at a compute binding slot.
MeshInstanceUploadError
RepresentationPolicy
Multi-modal representation policy encoded in InstanceCullDescriptor::rep_policy.
TemporalSubmissionState
Command-recording lifecycle for one renderer-owned temporal submission.
TextureColorSpace
TextureMipSemantic
TextureUploadError

Constants§

CSM_4_CASCADE_COUNT
EMF_VOLUMETRIC_WGSL
Raw WGSL shader source for the EMF volumetric visualizer.
IBL_CUBEMAP_CONVOLVE_WGSL
WGSL compute shader for 6-face procedural sky cubemap generation, diffuse irradiance convolution, 5-mip GGX importance-sampled specular prefiltering, and split-sum BRDF integration.
MAX_GPU_MATERIAL_DRAWS
Maximum ordered material draws admitted for one resident mesh in the portable renderer.

Functions§

build_hzb_mip_chain_zero_heap
Returns (total_texels_written, mip_slices, mip_count).
camera_csm_4_sun_view_projections
Computes 4 texel-snapped orthographic sun view-projection matrices for 4-cascade CSM.
checked_field_bytes
Bounded byte size calculation for a given cell count.
checked_grid_cell_count
Bounded calculation of field cell memory requirements.
compute_csm_4_splits_zero_heap
Practical Split Scheme (PSSM) 4-cascade logarithmic/uniform split distances: $$z_i = \lambda , n \left(\frac{f}{n}\right)^{i/4} + (1 - \lambda)\left(n + (f - n)\frac{i}{4}\right), \quad i \in {1, 2, 3, 4}$$
csm_cascade_blend_factor
Computes the active cascade index (0..3), next cascade index (0..3), and smooth transition blend weight $\alpha \in [0, 1]$ across the cascade overlap band (blend_band_fraction, e.g. 0.10) to eliminate hard CSM split seams.
cubemap_face_uv_to_direction
Maps a cubemap face index (0..=5 for +X, -X, +Y, -Y, +Z, -Z) and normalized face coordinates uv in [-1.0, 1.0]^2 to a unit world-space direction vector.
cull_and_select_lod_indirect_zero_heap
Zero-heap CPU/GPU parity execution of 6-plane frustum culling, 2-phase HZB depth-pyramid occlusion culling, 4-tier hybrid LOD selection with hysteresis, and multi-modal stream routing.
emf_cells_from_loaded
Convert all cells of an admitted LoadedEmf into GPU EmfFieldCell storage.
emf_field_cell_from_display
Fixed grayscale display derivative; exact f64 observations stay in the retained source.
emf_field_cell_from_optical
Convert paired scalar and optical/polarimetric display cells into a GPU cell.
emf_field_cell_from_record64
Convert one Q42EMF64 scientific 64-byte cell record directly into EmfFieldCell.
eval_procedural_sky_radiance_zero_heap
Evaluates the procedural Rayleigh + Henyey-Greenstein Mie sky radiance in direction dir on the CPU with zero heap allocations (HotZeroHeap).
evaluate_display_slice_rgba8
CPU slice evaluator with parity to GPU EmfSliceUniform::projection_mode (0 = full volumetric spectral + Stokes, 1 = nearest-cell grayscale |E| derivative).
full_spectrum_sigma_to_linear_rgb
Full-spectrum 6-band linear sRGB evaluator matching WGSL sigma_to_linear_rgb (0=CIE 1931 visible, 1=RF, 2=IR, 3=UV, 4=X-Ray, 5+=Gamma).
integrate_split_sum_brdf_zero_heap
Evaluates the split-sum BRDF integration $(s, b)$ at (n_dot_v, roughness) using Hammersley GGX importance sampling with zero heap allocations (HotZeroHeap).
particle_cap_for_mode
?::particle_cap_for_mode.
select_lod_tier_with_hysteresis
Selects the 4-tier LOD index (0..=3) from an epistemic-weighted screen-space projected radius in pixels, applying a fractional hysteresis deadband around threshold transitions.