@woosh/meep-engine 3.28.0 → 3.30.0

package/src/shade/renderer/rasterize/cull/hzb/shader_hzb_meshlet_filter_1way.js

@@ -10,6 +10,7 @@ import { AABB3_STRUCT } from "../../../shader/chunk/geometry/aabb3/AABB3_STRUCT.
 import { chunk_aabb3_project } from "../../../shader/chunk/geometry/aabb3/chunk_aabb3_project.js";
 import { chunk_aabb3_project_perspective } from "../../../shader/chunk/geometry/aabb3/chunk_aabb3_project_perspective.js";
 import { chunk_array_to_aabb3 } from "../../../shader/chunk/geometry/aabb3/chunk_array_to_aabb3.js";
+import { chunk_grid3d_to_index } from "../../../shader/chunk/grid/chunk_grid3d_to_index.js";
 import { CodeChunk } from "../../../shader/compiler/CodeChunk.js";
 import { COMPUTE_SHADER_DEFAULT_GROUP_SIZE } from "../../../shader/COMPUTE_SHADER_DEFAULT_GROUP_SIZE.js";
 import { ComputeShader } from "../../../shader/ComputeShader.js";
@@ -38,10 +39,15 @@ const body = CodeChunk.from(
 // language=WGSL
     `
 @compute @workgroup_size(${COMPUTE_SHADER_DEFAULT_GROUP_SIZE})
-fn unpack(@builtin(global_invocation_id) global_id : vec3<u32>){
-    
-    let input_index = global_id.x;
-    
+fn unpack(
+    @builtin(workgroup_id) workgroup_id : vec3<u32>,
+    @builtin(num_workgroups) group_count : vec3<u32>,
+    @builtin(local_invocation_index) local_index : u32,
+){
+
+    // folded dispatch — see IndirectComputeCommandBuildPass
+    let input_index = grid3d_to_index(workgroup_id, group_count.xy) * ${COMPUTE_SHADER_DEFAULT_GROUP_SIZE}u + local_index;
+
     if(input_index >= input.count){
         return;
     }
@@ -80,6 +86,7 @@ fn unpack(@builtin(global_invocation_id) global_id : vec3<u32>){
 }`,
     [
         AABB3_STRUCT.declaration_chunk,
+        chunk_grid3d_to_index,
         chunk_array_to_aabb3,
         chunk_aabb3_project,
         chunk_aabb3_project_perspective,

package/src/shade/renderer/rasterize/cull/hzb/shader_hzb_meshlet_filter_1way_full.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"shader_hzb_meshlet_filter_1way_full.d.ts","sourceRoot":"","sources":["../../../../../../../src/shade/renderer/rasterize/cull/hzb/shader_hzb_meshlet_filter_1way_full.js"],"names":[],"mappings":"AA6GA;;;GAGG;AACH,kDAFU,aAAa,CAMpB;8BAnG2B,kCAAkC"}
+{"version":3,"file":"shader_hzb_meshlet_filter_1way_full.d.ts","sourceRoot":"","sources":["../../../../../../../src/shade/renderer/rasterize/cull/hzb/shader_hzb_meshlet_filter_1way_full.js"],"names":[],"mappings":"AA8GA;;;GAGG;AACH,kDAFU,aAAa,CAMpB;8BApG2B,kCAAkC"}

package/src/shade/renderer/rasterize/cull/hzb/shader_hzb_meshlet_filter_1way_full.js

@@ -23,11 +23,12 @@ import { MESHLET_COLLECTION_WRITE_STRUCT } from "../../MESHLET_COLLECTION_WRITE_
 
 const resources = new ShaderResourceSetDescriptor();
 
-// TODO(chrome-153): uniform, not immediate — Dawn 153 overwrites immediate data on
-//      dispatchWorkgroupsIndirect, so this reads 65535 and culls everything. Revert 22893f0a7.
+// per-dispatch, eight bytes, read-only: immediate data rather than a uniform buffer. camera stays
+// a uniform — it is a scene buffer, not a per-dispatch value.
+resources.setImmediate("resolution", vec2u);
+
 resources.createGroup()
     .addUniform("camera", CAMERA_UNIFORM_STRUCT)
-    .addUniform("resolution", vec2u)
     .addStorageBuffer("scene_database", WebGPUArray.from(u32))
     .addStorageBuffer("meshlet_metadata", WebGPUArray.from(MESHLET_METADATA_STRUCT))
     .addTexture("texture_hzb")

package/src/shade/renderer/rasterize/expand/bucket/meshlet/graph_sort_meshlets_by_bucket_id.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"graph_sort_meshlets_by_bucket_id.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/graph_sort_meshlets_by_bucket_id.js"],"names":[],"mappings":"AAeA;;;;;;;GAOG;AACH;IAL+B,KAAK;IACA,SAAS;IAClB,cAAc,EAA9B,MAAM;IACuB,MAAM;QAuG7C"}
+{"version":3,"file":"graph_sort_meshlets_by_bucket_id.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/graph_sort_meshlets_by_bucket_id.js"],"names":[],"mappings":"AAeA;;;;;;;GAOG;AACH;IAL+B,KAAK;IACA,SAAS;IAClB,cAAc,EAA9B,MAAM;IACuB,MAAM;QA0G7C"}

package/src/shade/renderer/rasterize/expand/bucket/meshlet/graph_sort_meshlets_by_bucket_id.js

@@ -46,6 +46,8 @@ export function graph_sort_meshlets_by_bucket_id({
         const input = resources.get(data.input);
         const command = resources.get(data.command);
 
+        // folded: a meshlet collection is past what one dispatch dimension reaches — see
+        // IndirectComputeCommandBuildPass
         build_indirect_compute_dispatch_command(
             COMPUTE_SHADER_DEFAULT_GROUP_SIZE,
             encoder,
@@ -53,6 +55,7 @@ export function graph_sort_meshlets_by_bucket_id({
             0,
             command,
             0,
+            true,
         );
 
         // count by bucket

package/src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_count_by_rasterization_bucket.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"shader_meshlets_count_by_rasterization_bucket.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_count_by_rasterization_bucket.js"],"names":[],"mappings":"AAsFA,0EAIG;8BAjF2B,qCAAqC"}
+{"version":3,"file":"shader_meshlets_count_by_rasterization_bucket.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_count_by_rasterization_bucket.js"],"names":[],"mappings":"AA0FA,0EAIG;8BApF2B,qCAAqC"}

package/src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_count_by_rasterization_bucket.js

@@ -1,91 +1,95 @@
-import { assert } from "../../../../../../core/assert.js";
-import { atomic_u32, u32 } from "../../../../shader/type/WEBGPU_WGSL_PRIMITIVE_TYPES.js";
-import { WebGPUArray } from "../../../../shader/type/WebGPUArray.js";
-import { WebGPUStruct } from "../../../../shader/type/WebGPUStruct.js";
-import { MESHLET_COLLECTION_STRUCT } from "../../../MESHLET_COLLECTION_STRUCT.js";
-import { MATERIAL_METADATA_STRUCT } from "../../../../material/MATERIAL_METADATA_STRUCT.js";
-import { chunk_scene_database_access } from "../../../../scene/chunk_scene_database_access.js";
-import { CodeChunk } from "../../../../shader/compiler/CodeChunk.js";
-import { COMPUTE_SHADER_DEFAULT_GROUP_SIZE } from "../../../../shader/COMPUTE_SHADER_DEFAULT_GROUP_SIZE.js";
-import { ComputeShader } from "../../../../shader/ComputeShader.js";
-import { ShaderResourceSetDescriptor } from "../../../../shader/resource/ShaderResourceSetDescriptor.js";
-import { chunk_rasterization_material_bucket } from "../chunk_rasterization_material_bucket.js";
-import { RASTERIZATION_BUCKET_LIMIT } from "../RASTERIZATION_BUCKET_LIMIT.js";
-
-
-const OUTPUT_STRUCT = WebGPUStruct.from({
-    counts: WebGPUArray.from(atomic_u32, RASTERIZATION_BUCKET_LIMIT),
-});
-
-const resources = new ShaderResourceSetDescriptor();
-resources.createGroup()
-    .addStorageBuffer("input", MESHLET_COLLECTION_STRUCT)
-    .addStorageBuffer("scene_database", WebGPUArray.from(u32))
-    .addStorageBuffer("materials", WebGPUArray.from(MATERIAL_METADATA_STRUCT))
-    .addStorageBuffer("output", OUTPUT_STRUCT, true)
-
-
-const GROUP_SIZE = COMPUTE_SHADER_DEFAULT_GROUP_SIZE;
-
-// must fit
-assert.lessThanOrEqual(RASTERIZATION_BUCKET_LIMIT, GROUP_SIZE);
-
-const body = CodeChunk.from(
-    //language=WGSL
-    `
-var<workgroup> wg_buckets: array<atomic<u32>, ${RASTERIZATION_BUCKET_LIMIT}>;
-
-@compute @workgroup_size(${GROUP_SIZE})
-fn main(
-    @builtin(global_invocation_id) global_id : vec3<u32>,
-    @builtin(local_invocation_index) local_id : u32,
-){
-    let thread_id = global_id.x;
-
-    let input_size = min(input.count, arrayLength(&input.elements));
-
-    if( thread_id < input_size ){
-        let meshlet_def = input.elements[thread_id];
-        
-        let mesh_index = meshlet_def.mesh;
-
-        let mesh = scene_read_mesh(&scene_database, mesh_index);
-
-        let material = materials[mesh.material];
-        
-        let bucket_index = rasterization_material_bucket(
-            material.transparency_mode,
-            material.draw_mode,
-            material.draw_side
-        );
-    
-        atomicAdd(&wg_buckets[bucket_index], 1u);
-    
-    }
-    
-    workgroupBarrier();
-    
-    // write out
-    if(local_id < ${RASTERIZATION_BUCKET_LIMIT}){
-        let bucket_index = local_id;
-        let count = atomicLoad(&wg_buckets[bucket_index]);
-        
-        if(count > 0u){
-            atomicAdd(&output.counts[bucket_index], count);
-        }
-        
-    }
-    
-}
-    `,
-    [
-        chunk_rasterization_material_bucket,
-        chunk_scene_database_access,
-    ]
-)
-
-export const shader_meshlets_count_by_rasterization_bucket = ComputeShader.from({
-    label: 'Count meshlets by bucket',
-    body,
-    resources,
+import { assert } from "../../../../../../core/assert.js";
+import { chunk_grid3d_to_index } from "../../../../shader/chunk/grid/chunk_grid3d_to_index.js";
+import { atomic_u32, u32 } from "../../../../shader/type/WEBGPU_WGSL_PRIMITIVE_TYPES.js";
+import { WebGPUArray } from "../../../../shader/type/WebGPUArray.js";
+import { WebGPUStruct } from "../../../../shader/type/WebGPUStruct.js";
+import { MESHLET_COLLECTION_STRUCT } from "../../../MESHLET_COLLECTION_STRUCT.js";
+import { MATERIAL_METADATA_STRUCT } from "../../../../material/MATERIAL_METADATA_STRUCT.js";
+import { chunk_scene_database_access } from "../../../../scene/chunk_scene_database_access.js";
+import { CodeChunk } from "../../../../shader/compiler/CodeChunk.js";
+import { COMPUTE_SHADER_DEFAULT_GROUP_SIZE } from "../../../../shader/COMPUTE_SHADER_DEFAULT_GROUP_SIZE.js";
+import { ComputeShader } from "../../../../shader/ComputeShader.js";
+import { ShaderResourceSetDescriptor } from "../../../../shader/resource/ShaderResourceSetDescriptor.js";
+import { chunk_rasterization_material_bucket } from "../chunk_rasterization_material_bucket.js";
+import { RASTERIZATION_BUCKET_LIMIT } from "../RASTERIZATION_BUCKET_LIMIT.js";
+
+
+const OUTPUT_STRUCT = WebGPUStruct.from({
+    counts: WebGPUArray.from(atomic_u32, RASTERIZATION_BUCKET_LIMIT),
+});
+
+const resources = new ShaderResourceSetDescriptor();
+resources.createGroup()
+    .addStorageBuffer("input", MESHLET_COLLECTION_STRUCT)
+    .addStorageBuffer("scene_database", WebGPUArray.from(u32))
+    .addStorageBuffer("materials", WebGPUArray.from(MATERIAL_METADATA_STRUCT))
+    .addStorageBuffer("output", OUTPUT_STRUCT, true)
+
+
+const GROUP_SIZE = COMPUTE_SHADER_DEFAULT_GROUP_SIZE;
+
+// must fit
+assert.lessThanOrEqual(RASTERIZATION_BUCKET_LIMIT, GROUP_SIZE);
+
+const body = CodeChunk.from(
+    //language=WGSL
+    `
+var<workgroup> wg_buckets: array<atomic<u32>, ${RASTERIZATION_BUCKET_LIMIT}>;
+
+@compute @workgroup_size(${GROUP_SIZE})
+fn main(
+    @builtin(workgroup_id) workgroup_id : vec3<u32>,
+    @builtin(num_workgroups) group_count : vec3<u32>,
+    @builtin(local_invocation_index) local_id : u32,
+){
+    // folded dispatch — see IndirectComputeCommandBuildPass
+    let thread_id = grid3d_to_index(workgroup_id, group_count.xy) * ${GROUP_SIZE}u + local_id;
+
+    let input_size = min(input.count, arrayLength(&input.elements));
+
+    if( thread_id < input_size ){
+        let meshlet_def = input.elements[thread_id];
+        
+        let mesh_index = meshlet_def.mesh;
+
+        let mesh = scene_read_mesh(&scene_database, mesh_index);
+
+        let material = materials[mesh.material];
+        
+        let bucket_index = rasterization_material_bucket(
+            material.transparency_mode,
+            material.draw_mode,
+            material.draw_side
+        );
+    
+        atomicAdd(&wg_buckets[bucket_index], 1u);
+    
+    }
+    
+    workgroupBarrier();
+    
+    // write out
+    if(local_id < ${RASTERIZATION_BUCKET_LIMIT}){
+        let bucket_index = local_id;
+        let count = atomicLoad(&wg_buckets[bucket_index]);
+        
+        if(count > 0u){
+            atomicAdd(&output.counts[bucket_index], count);
+        }
+        
+    }
+    
+}
+    `,
+    [
+        chunk_grid3d_to_index,
+        chunk_rasterization_material_bucket,
+        chunk_scene_database_access,
+    ]
+)
+
+export const shader_meshlets_count_by_rasterization_bucket = ComputeShader.from({
+    label: 'Count meshlets by bucket',
+    body,
+    resources,
 });

package/src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_extract_bucket.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"shader_meshlets_extract_bucket.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_extract_bucket.js"],"names":[],"mappings":"AAgEA,2DAIG;8BA9D2B,qCAAqC"}
+{"version":3,"file":"shader_meshlets_extract_bucket.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_extract_bucket.js"],"names":[],"mappings":"AAwEA,2DAIG;8BArE2B,qCAAqC"}

package/src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_extract_bucket.js

@@ -1,6 +1,7 @@
 import { WebGPUArray } from "../../../../shader/type/WebGPUArray.js";
 import { WebGPUStruct } from "../../../../shader/type/WebGPUStruct.js";
 import { MESHLET_COLLECTION_STRUCT } from "../../../MESHLET_COLLECTION_STRUCT.js";
+import { chunk_grid3d_to_index } from "../../../../shader/chunk/grid/chunk_grid3d_to_index.js";
 import { U32_CONTAINER_STRUCT } from "../../../../shader/type/U32_CONTAINER_STRUCT.js";
 import { CodeChunk } from "../../../../shader/compiler/CodeChunk.js";
 import { COMPUTE_SHADER_DEFAULT_GROUP_SIZE } from "../../../../shader/COMPUTE_SHADER_DEFAULT_GROUP_SIZE.js";
@@ -13,10 +14,11 @@ import {
 
 const resources = new ShaderResourceSetDescriptor();
 
-// TODO(chrome-153): uniform, not immediate — Dawn 153 overwrites immediate data on
-//      dispatchWorkgroupsIndirect, so this reads 65535 and culls everything. Revert 22893f0a7.
+// per-dispatch, four bytes, read-only: immediate data rather than a uniform buffer — the same
+// shape shader_read_out_bucket already uses
+resources.setImmediate("bucket_index", U32_CONTAINER_STRUCT);
+
 resources.createGroup()
-    .addUniform("bucket_index", U32_CONTAINER_STRUCT)
     .addStorageBuffer("bucket_data", WebGPUStruct.from({
         headers: RASTERIZATION_BUCKET_SORT_OUTPUT_HEADER_STRUCT_READ,
         elements: WebGPUArray.from(MESHLET_DEFINITION_STRUCT)
@@ -27,8 +29,13 @@ const body = CodeChunk.from(
     //language=WGSL
     `
 @compute @workgroup_size(${COMPUTE_SHADER_DEFAULT_GROUP_SIZE})
-fn main(@builtin(global_invocation_id) global_id : vec3<u32>){
-    let thread_id = global_id.x;
+fn main(
+    @builtin(workgroup_id) workgroup_id : vec3<u32>,
+    @builtin(num_workgroups) group_count : vec3<u32>,
+    @builtin(local_invocation_index) local_index : u32,
+){
+    // folded dispatch — see IndirectComputeCommandBuildPass
+    let thread_id = grid3d_to_index(workgroup_id, group_count.xy) * ${COMPUTE_SHADER_DEFAULT_GROUP_SIZE}u + local_index;
     
     let bucket_id = bucket_index.value;
     
@@ -58,6 +65,7 @@ fn main(@builtin(global_invocation_id) global_id : vec3<u32>){
     
 }
     `, [
+        chunk_grid3d_to_index,
         MESHLET_COLLECTION_STRUCT.declaration_chunk,
     ]
 )

package/src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_rasterization_bucket_sort.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"shader_meshlets_rasterization_bucket_sort.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_rasterization_bucket_sort.js"],"names":[],"mappings":"AAqFA,sEAIE;8BAjF4B,qCAAqC"}
+{"version":3,"file":"shader_meshlets_rasterization_bucket_sort.d.ts","sourceRoot":"","sources":["../../../../../../../../src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_rasterization_bucket_sort.js"],"names":[],"mappings":"AA4FA,sEAIE;8BAvF4B,qCAAqC"}

package/src/shade/renderer/rasterize/expand/bucket/meshlet/shader_meshlets_rasterization_bucket_sort.js

@@ -1,90 +1,97 @@
-import { u32 } from "../../../../shader/type/WEBGPU_WGSL_PRIMITIVE_TYPES.js";
-import { WebGPUArray } from "../../../../shader/type/WebGPUArray.js";
-import { WebGPUStruct } from "../../../../shader/type/WebGPUStruct.js";
-import { MESHLET_COLLECTION_STRUCT } from "../../../MESHLET_COLLECTION_STRUCT.js";
-import { MATERIAL_METADATA_STRUCT } from "../../../../material/MATERIAL_METADATA_STRUCT.js";
-import { chunk_scene_database_access } from "../../../../scene/chunk_scene_database_access.js";
-import { CodeChunk } from "../../../../shader/compiler/CodeChunk.js";
-import { COMPUTE_SHADER_DEFAULT_GROUP_SIZE } from "../../../../shader/COMPUTE_SHADER_DEFAULT_GROUP_SIZE.js";
-import { ComputeShader } from "../../../../shader/ComputeShader.js";
-import { ShaderResourceSetDescriptor } from "../../../../shader/resource/ShaderResourceSetDescriptor.js";
-import { MESHLET_DEFINITION_STRUCT } from "../../../../geometry/meshlet/encoding/MESHLET_DEFINITION_STRUCT.js";
-import { chunk_rasterization_material_bucket } from "../chunk_rasterization_material_bucket.js";
-import { RASTERIZATION_BUCKET_LIMIT } from "../RASTERIZATION_BUCKET_LIMIT.js";
-import {
-    RASTERIZATION_BUCKET_SORT_OUTPUT_HEADER_STRUCT_WRITE
-} from "../RASTERIZATION_BUCKET_SORT_OUTPUT_HEADER_STRUCT_WRITE.js";
-
-
-const RASTERIZATION_MESHLET_BUCKET_SORT_OUTPUT_STRUCT_WRITE = WebGPUStruct.from({
-    headers: RASTERIZATION_BUCKET_SORT_OUTPUT_HEADER_STRUCT_WRITE,
-    elements: WebGPUArray.from(MESHLET_DEFINITION_STRUCT)
-});
-
-const OFFSETS_TYPE = WebGPUArray.from(u32, RASTERIZATION_BUCKET_LIMIT);
-
-const resources = new ShaderResourceSetDescriptor();
-resources.createGroup()
-    .addStorageBuffer("scene_database", WebGPUArray.from(u32))
-    .addStorageBuffer("materials", WebGPUArray.from(MATERIAL_METADATA_STRUCT))
-    .addStorageBuffer("input", MESHLET_COLLECTION_STRUCT)
-    .addStorageBuffer("offsets",OFFSETS_TYPE)
-    .addStorageBuffer("output", RASTERIZATION_MESHLET_BUCKET_SORT_OUTPUT_STRUCT_WRITE, true)
-
-
-const body = CodeChunk.from(
-    //language=WGSL
-    `
-@compute @workgroup_size(${COMPUTE_SHADER_DEFAULT_GROUP_SIZE})
-fn main(@builtin(global_invocation_id) global_id : vec3<u32>){
-
-    let index = global_id.x;
-    
-    let input_size = min(input.count, arrayLength(&input.elements));
-    
-    if(index >= input_size){
-        return;
-    }
-    
-    let meshlet_def = input.elements[index];
-
-    let mesh_index = meshlet_def.mesh;
-
-    let mesh = scene_read_mesh(&scene_database, mesh_index);
-
-    let material = materials[mesh.material];
-    
-    let bucket_index = rasterization_material_bucket(
-        material.transparency_mode,
-        material.draw_mode,
-        material.draw_side
-    );
-    
-    let element_offset = atomicAdd(&output.headers[bucket_index].count, 1u);
-    
-    let bucket_offset = offsets[bucket_index];
-    
-    if(element_offset == 0u){
-        // first element in bucket, write offset as well
-        output.headers[bucket_index].offset = bucket_offset;
-    }
-    
-    let output_offset = bucket_offset + element_offset;
-    
-    if(output_offset < arrayLength(&output.elements)){
-        // guard against out-of-bounds access
-        output.elements[output_offset] = meshlet_def;
-    }
-    
-}
-    `, [
-        chunk_rasterization_material_bucket,
-        chunk_scene_database_access,
-    ]
-);
-
-export const shader_meshlets_rasterization_bucket_sort = ComputeShader.from({
-    label: 'Rasterization bucket sort / meshlets',
-    resources,
-    body,
+import { u32 } from "../../../../shader/type/WEBGPU_WGSL_PRIMITIVE_TYPES.js";
+import { WebGPUArray } from "../../../../shader/type/WebGPUArray.js";
+import { WebGPUStruct } from "../../../../shader/type/WebGPUStruct.js";
+import { MESHLET_COLLECTION_STRUCT } from "../../../MESHLET_COLLECTION_STRUCT.js";
+import { chunk_grid3d_to_index } from "../../../../shader/chunk/grid/chunk_grid3d_to_index.js";
+import { MATERIAL_METADATA_STRUCT } from "../../../../material/MATERIAL_METADATA_STRUCT.js";
+import { chunk_scene_database_access } from "../../../../scene/chunk_scene_database_access.js";
+import { CodeChunk } from "../../../../shader/compiler/CodeChunk.js";
+import { COMPUTE_SHADER_DEFAULT_GROUP_SIZE } from "../../../../shader/COMPUTE_SHADER_DEFAULT_GROUP_SIZE.js";
+import { ComputeShader } from "../../../../shader/ComputeShader.js";
+import { ShaderResourceSetDescriptor } from "../../../../shader/resource/ShaderResourceSetDescriptor.js";
+import { MESHLET_DEFINITION_STRUCT } from "../../../../geometry/meshlet/encoding/MESHLET_DEFINITION_STRUCT.js";
+import { chunk_rasterization_material_bucket } from "../chunk_rasterization_material_bucket.js";
+import { RASTERIZATION_BUCKET_LIMIT } from "../RASTERIZATION_BUCKET_LIMIT.js";
+import {
+    RASTERIZATION_BUCKET_SORT_OUTPUT_HEADER_STRUCT_WRITE
+} from "../RASTERIZATION_BUCKET_SORT_OUTPUT_HEADER_STRUCT_WRITE.js";
+
+
+const RASTERIZATION_MESHLET_BUCKET_SORT_OUTPUT_STRUCT_WRITE = WebGPUStruct.from({
+    headers: RASTERIZATION_BUCKET_SORT_OUTPUT_HEADER_STRUCT_WRITE,
+    elements: WebGPUArray.from(MESHLET_DEFINITION_STRUCT)
+});
+
+const OFFSETS_TYPE = WebGPUArray.from(u32, RASTERIZATION_BUCKET_LIMIT);
+
+const resources = new ShaderResourceSetDescriptor();
+resources.createGroup()
+    .addStorageBuffer("scene_database", WebGPUArray.from(u32))
+    .addStorageBuffer("materials", WebGPUArray.from(MATERIAL_METADATA_STRUCT))
+    .addStorageBuffer("input", MESHLET_COLLECTION_STRUCT)
+    .addStorageBuffer("offsets",OFFSETS_TYPE)
+    .addStorageBuffer("output", RASTERIZATION_MESHLET_BUCKET_SORT_OUTPUT_STRUCT_WRITE, true)
+
+
+const body = CodeChunk.from(
+    //language=WGSL
+    `
+@compute @workgroup_size(${COMPUTE_SHADER_DEFAULT_GROUP_SIZE})
+fn main(
+    @builtin(workgroup_id) workgroup_id : vec3<u32>,
+    @builtin(num_workgroups) group_count : vec3<u32>,
+    @builtin(local_invocation_index) local_index : u32,
+){
+
+    // folded dispatch — see IndirectComputeCommandBuildPass
+    let index = grid3d_to_index(workgroup_id, group_count.xy) * ${COMPUTE_SHADER_DEFAULT_GROUP_SIZE}u + local_index;
+    
+    let input_size = min(input.count, arrayLength(&input.elements));
+    
+    if(index >= input_size){
+        return;
+    }
+    
+    let meshlet_def = input.elements[index];
+
+    let mesh_index = meshlet_def.mesh;
+
+    let mesh = scene_read_mesh(&scene_database, mesh_index);
+
+    let material = materials[mesh.material];
+    
+    let bucket_index = rasterization_material_bucket(
+        material.transparency_mode,
+        material.draw_mode,
+        material.draw_side
+    );
+    
+    let element_offset = atomicAdd(&output.headers[bucket_index].count, 1u);
+    
+    let bucket_offset = offsets[bucket_index];
+    
+    if(element_offset == 0u){
+        // first element in bucket, write offset as well
+        output.headers[bucket_index].offset = bucket_offset;
+    }
+    
+    let output_offset = bucket_offset + element_offset;
+    
+    if(output_offset < arrayLength(&output.elements)){
+        // guard against out-of-bounds access
+        output.elements[output_offset] = meshlet_def;
+    }
+    
+}
+    `, [
+        chunk_grid3d_to_index,
+        chunk_rasterization_material_bucket,
+        chunk_scene_database_access,
+    ]
+);
+
+export const shader_meshlets_rasterization_bucket_sort = ComputeShader.from({
+    label: 'Rasterization bucket sort / meshlets',
+    resources,
+    body,
 })

package/src/shade/renderer/rasterize/expand/graph_meshlets_filter_one_way.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"graph_meshlets_filter_one_way.d.ts","sourceRoot":"","sources":["../../../../../../src/shade/renderer/rasterize/expand/graph_meshlets_filter_one_way.js"],"names":[],"mappings":"AAKA;;;;;;;;;;;GAWG;AACH;IAT+B,KAAK;IACD,IAAI;IACZ,MAAM,EAAtB,MAAM;IACU,cAAc,EAA9B,MAAM;IACU,WAAW,EAA3B,MAAM;IACU,GAAG,EAAnB,MAAM;IACW,OAAO,GAAxB,MAAM;IACL,MAAM,CA8BjB"}
+{"version":3,"file":"graph_meshlets_filter_one_way.d.ts","sourceRoot":"","sources":["../../../../../../src/shade/renderer/rasterize/expand/graph_meshlets_filter_one_way.js"],"names":[],"mappings":"AAKA;;;;;;;;;;;GAWG;AACH;IAT+B,KAAK;IACD,IAAI;IACZ,MAAM,EAAtB,MAAM;IACU,cAAc,EAA9B,MAAM;IACU,WAAW,EAA3B,MAAM;IACU,GAAG,EAAnB,MAAM;IACW,OAAO,GAAxB,MAAM;IACL,MAAM,CAgCjB"}

package/src/shade/renderer/rasterize/expand/graph_meshlets_filter_one_way.js

@@ -29,7 +29,9 @@ export function graph_meshlets_filter_one_way({
 
     data.scene = view.scene;
 
-    const command = graph_build_indirect_dispatch_command(graph, input_meshlets, COMPUTE_SHADER_DEFAULT_GROUP_SIZE);
+    // folded: a meshlet collection is past what one dispatch dimension reaches — see
+    // IndirectComputeCommandBuildPass
+    const command = graph_build_indirect_dispatch_command(graph, input_meshlets, COMPUTE_SHADER_DEFAULT_GROUP_SIZE, true);
     const pass = graph.add("hzb filter meshlets", data, hzb_filter_meshlets_pass);
 
     data.command = pass.read(command);

package/src/shade/renderer/rasterize/expand/graph_meshlets_filter_two_way.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"graph_meshlets_filter_two_way.d.ts","sourceRoot":"","sources":["../../../../../../src/shade/renderer/rasterize/expand/graph_meshlets_filter_two_way.js"],"names":[],"mappings":"AAMA;;;;;;;;;GASG;AACH;IAP+B,KAAK;IACD,IAAI;IACZ,cAAc,EAA9B,MAAM;IACU,WAAW,EAA3B,MAAM;IACU,GAAG,EAAnB,MAAM;IACJ,MAAM,EAAE,CAyCpB"}
+{"version":3,"file":"graph_meshlets_filter_two_way.d.ts","sourceRoot":"","sources":["../../../../../../src/shade/renderer/rasterize/expand/graph_meshlets_filter_two_way.js"],"names":[],"mappings":"AAMA;;;;;;;;;GASG;AACH;IAP+B,KAAK;IACD,IAAI;IACZ,cAAc,EAA9B,MAAM;IACU,WAAW,EAA3B,MAAM;IACU,GAAG,EAAnB,MAAM;IACJ,MAAM,EAAE,CA4CpB"}

package/src/shade/renderer/rasterize/expand/graph_meshlets_filter_two_way.js

@@ -31,7 +31,10 @@ export function graph_meshlets_filter_two_way({
     data.view = view;
 
     // the 2-way filter shader uses COMPUTE_SHADER_DEFAULT_GROUP_SIZE (see filter_meshlets.2-way.compute.js)
-    const command = graph_build_indirect_dispatch_command(graph, input_meshlets, COMPUTE_SHADER_DEFAULT_GROUP_SIZE);
+    // folded: a meshlet collection is past what one dispatch dimension reaches — see
+    // IndirectComputeCommandBuildPass. shader_meshlet_filter_2way indexes for it, and this is
+    // the other graph that drives it.
+    const command = graph_build_indirect_dispatch_command(graph, input_meshlets, COMPUTE_SHADER_DEFAULT_GROUP_SIZE, true);
 
     const pass = graph.add("filter meshlets 2 way", data, pass_filter_meshlets_2way);
 

package/src/shade/renderer/rasterize/expand/mesh/shader_mesh_expand_to_meshlets_prefix.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"shader_mesh_expand_to_meshlets_prefix.d.ts","sourceRoot":"","sources":["../../../../../../../src/shade/renderer/rasterize/expand/mesh/shader_mesh_expand_to_meshlets_prefix.js"],"names":[],"mappings":"AAqFA,kEAIE;8BAjF4B,kCAAkC"}
+{"version":3,"file":"shader_mesh_expand_to_meshlets_prefix.d.ts","sourceRoot":"","sources":["../../../../../../../src/shade/renderer/rasterize/expand/mesh/shader_mesh_expand_to_meshlets_prefix.js"],"names":[],"mappings":"AAgGA,kEAIE;8BA3F4B,kCAAkC"}