@aardworx/wombat.rendering 0.9.18 → 0.9.19

package/dist/runtime/derivedModes/gpuDispatcher.d.ts

@@ -0,0 +1,46 @@
+import type { CullMode } from "../pipelineCache/index.js";
+export declare class GpuDerivedModesScene {
+    readonly device: GPUDevice;
+    private readonly entries;
+    private liveCount;
+    private dirty;
+    private inputBuf;
+    private outputBuf;
+    private stagingBuf;
+    private paramsBuf;
+    private pipeline;
+    private bindGroup;
+    private layout;
+    /** Capacity in number of u32 entries (= numROs the buffers can hold). */
+    private capacity;
+    /** Last-known arena buffer, used to detect bind-group invalidation. */
+    private arenaBuf;
+    /** Last-known declared (for kernel uniform). Common case: all RO
+     *  rules share the same declared (= the SG scope's value). */
+    private declared;
+    /** Per-RO last-known output. CPU mirror, populated by finish(). */
+    readonly lastOutput: Uint32Array;
+    constructor(device: GPUDevice);
+    get registered(): number;
+    registerRo(drawId: number, modelRef: number, declared: CullMode): void;
+    deregisterRo(drawId: number): void;
+    markDirty(): void;
+    consumeDirty(): boolean;
+    private ensureResources;
+    /** Upload current per-RO input refs + params; called when dirty. */
+    private uploadInputs;
+    /**
+     * Encode the compute dispatch + the staging copy into `enc`. Caller
+     * is responsible for `device.queue.submit([enc.finish()])`. Returns
+     * the number of ROs the kernel was dispatched against.
+     */
+    dispatch(arenaBuf: GPUBuffer, numROs: number, enc: GPUCommandEncoder): number;
+    /**
+     * mapAsync the staging buffer + copy values into `lastOutput`.
+     * Resolves on the GPU finishing the copy. Caller diffs lastOutput
+     * against its own cached values to find changed ROs.
+     */
+    finish(numROs: number): Promise<void>;
+    dispose(): void;
+}
+//# sourceMappingURL=gpuDispatcher.d.ts.map

package/dist/runtime/derivedModes/gpuDispatcher.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"gpuDispatcher.d.ts","sourceRoot":"","sources":["../../../src/runtime/derivedModes/gpuDispatcher.ts"],"names":[],"mappings":"AAwBA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,2BAA2B,CAAC;AAc1D,qBAAa,oBAAoB;IAC/B,QAAQ,CAAC,MAAM,EAAE,SAAS,CAAC;IAE3B,OAAO,CAAC,QAAQ,CAAC,OAAO,CAA+B;IACvD,OAAO,CAAC,SAAS,CAAK;IACtB,OAAO,CAAC,KAAK,CAAS;IAGtB,OAAO,CAAC,QAAQ,CAA4B;IAC5C,OAAO,CAAC,SAAS,CAA2B;IAC5C,OAAO,CAAC,UAAU,CAA0B;IAC5C,OAAO,CAAC,SAAS,CAA2B;IAC5C,OAAO,CAAC,QAAQ,CAAqC;IACrD,OAAO,CAAC,SAAS,CAA8B;IAC/C,OAAO,CAAC,MAAM,CAAuC;IACrD,yEAAyE;IACzE,OAAO,CAAC,QAAQ,CAAK;IACrB,uEAAuE;IACvE,OAAO,CAAC,QAAQ,CAA0B;IAC1C;kEAC8D;IAC9D,OAAO,CAAC,QAAQ,CAA4B;IAE5C,mEAAmE;IACnE,QAAQ,CAAC,UAAU,EAAE,WAAW,CAAC;gBAErB,MAAM,EAAE,SAAS;IAK7B,IAAI,UAAU,IAAI,MAAM,CAA2B;IAEnD,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,GAAG,IAAI;IAOtE,YAAY,CAAC,MAAM,EAAE,MAAM,GAAG,IAAI;IAOlC,SAAS,IAAI,IAAI;IACjB,YAAY,IAAI,OAAO;IAEvB,OAAO,CAAC,eAAe;IAmDvB,oEAAoE;IACpE,OAAO,CAAC,YAAY;IAWpB;;;;OAIG;IACH,QAAQ,CAAC,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,EAAE,iBAAiB,GAAG,MAAM;IAa7E;;;;OAIG;IACG,MAAM,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAS3C,OAAO,IAAI,IAAI;CAMhB"}

package/dist/runtime/derivedModes/gpuDispatcher.js

@@ -0,0 +1,170 @@
+// Scene-level dispatcher for the v1 GPU-evaluated derived-mode rule
+// (flip-cull-by-determinant; see ./gpuKernel.ts).
+//
+// Lifecycle
+// =========
+//
+//   - Construct once per heap scene with the device.
+//   - registerRo(drawId, modelRef, declaredCull): register an RO that
+//     has the GPU flip-cull rule. modelRef = byte offset into the
+//     arena where the RO's ModelTrafo mat4 lives.
+//   - deregisterRo(drawId): unregister on removeDraw.
+//   - markDirty(): caller hints that an input changed; the next
+//     `dispatch()` will run the kernel.
+//   - dispatch(arenaBuf, enc): if dirty, encode the compute pass into
+//     `enc`. Always copies the output buffer into a staging buffer for
+//     CPU readback.
+//   - finish(): mapAsync the staging buffer and resolve with the
+//     per-RO output array. The scene calls this AFTER submit and
+//     before deciding rebuckets. The promise resolves on the next
+//     microtask after the GPU completes the copy.
+//
+// The buffers grow lazily (pow2) as ROs are added.
+import { GPU_FLIP_CULL_BY_DET_WGSL, CULL_TO_U32 } from "./gpuKernel.js";
+const NO_RULE = 0xFFFFFFFF;
+const WG_SIZE = 64;
+const POW2 = (n) => { let p = 1; while (p < n)
+    p <<= 1; return Math.max(64, p); };
+export class GpuDerivedModesScene {
+    device;
+    entries = [];
+    liveCount = 0;
+    dirty = false;
+    // GPU resources — lazily allocated on first registerRo.
+    inputBuf = null;
+    outputBuf = null;
+    stagingBuf = null;
+    paramsBuf = null;
+    pipeline = null;
+    bindGroup = null;
+    layout = null;
+    /** Capacity in number of u32 entries (= numROs the buffers can hold). */
+    capacity = 0;
+    /** Last-known arena buffer, used to detect bind-group invalidation. */
+    arenaBuf = null;
+    /** Last-known declared (for kernel uniform). Common case: all RO
+     *  rules share the same declared (= the SG scope's value). */
+    declared = CULL_TO_U32.back;
+    /** Per-RO last-known output. CPU mirror, populated by finish(). */
+    lastOutput;
+    constructor(device) {
+        this.device = device;
+        this.lastOutput = new Uint32Array(0);
+    }
+    get registered() { return this.liveCount; }
+    registerRo(drawId, modelRef, declared) {
+        this.entries[drawId] = { modelRef, declaredU32: CULL_TO_U32[declared] };
+        this.liveCount++;
+        this.dirty = true;
+        this.declared = CULL_TO_U32[declared];
+    }
+    deregisterRo(drawId) {
+        if (this.entries[drawId] === undefined)
+            return;
+        this.entries[drawId] = undefined;
+        this.liveCount--;
+        this.dirty = true;
+    }
+    markDirty() { this.dirty = true; }
+    consumeDirty() { const was = this.dirty; this.dirty = false; return was; }
+    ensureResources(arenaBuf, numROs) {
+        const needCap = POW2(Math.max(1, numROs));
+        if (this.outputBuf === null || needCap > this.capacity) {
+            this.inputBuf?.destroy();
+            this.outputBuf?.destroy();
+            this.stagingBuf?.destroy();
+            const bytes = needCap * 4;
+            this.inputBuf = this.device.createBuffer({ size: bytes, usage: GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_DST, label: "gpuDerivedModes/in" });
+            this.outputBuf = this.device.createBuffer({ size: bytes, usage: GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_SRC | GPUBufferUsage.COPY_DST, label: "gpuDerivedModes/out" });
+            this.stagingBuf = this.device.createBuffer({ size: bytes, usage: GPUBufferUsage.MAP_READ | GPUBufferUsage.COPY_DST, label: "gpuDerivedModes/staging" });
+            this.lastOutput = new Uint32Array(needCap);
+            this.capacity = needCap;
+            this.bindGroup = null;
+        }
+        if (this.paramsBuf === null) {
+            this.paramsBuf = this.device.createBuffer({ size: 16, usage: GPUBufferUsage.UNIFORM | GPUBufferUsage.COPY_DST, label: "gpuDerivedModes/params" });
+        }
+        if (this.layout === null) {
+            this.layout = this.device.createBindGroupLayout({
+                label: "gpuDerivedModes/bgl",
+                entries: [
+                    { binding: 0, visibility: GPUShaderStage.COMPUTE, buffer: { type: "read-only-storage" } },
+                    { binding: 1, visibility: GPUShaderStage.COMPUTE, buffer: { type: "read-only-storage" } },
+                    { binding: 2, visibility: GPUShaderStage.COMPUTE, buffer: { type: "storage" } },
+                    { binding: 3, visibility: GPUShaderStage.COMPUTE, buffer: { type: "uniform" } },
+                ],
+            });
+        }
+        if (this.pipeline === null) {
+            const module = this.device.createShaderModule({ code: GPU_FLIP_CULL_BY_DET_WGSL, label: "gpuDerivedModes/module" });
+            this.pipeline = this.device.createComputePipeline({
+                label: "gpuDerivedModes/pipeline",
+                layout: this.device.createPipelineLayout({ bindGroupLayouts: [this.layout] }),
+                compute: { module, entryPoint: "evaluate" },
+            });
+        }
+        if (this.bindGroup === null || this.arenaBuf !== arenaBuf) {
+            this.bindGroup = this.device.createBindGroup({
+                label: "gpuDerivedModes/bindGroup",
+                layout: this.layout,
+                entries: [
+                    { binding: 0, resource: { buffer: arenaBuf } },
+                    { binding: 1, resource: { buffer: this.inputBuf } },
+                    { binding: 2, resource: { buffer: this.outputBuf } },
+                    { binding: 3, resource: { buffer: this.paramsBuf } },
+                ],
+            });
+            this.arenaBuf = arenaBuf;
+        }
+    }
+    /** Upload current per-RO input refs + params; called when dirty. */
+    uploadInputs(numROs) {
+        const refs = new Uint32Array(this.capacity).fill(NO_RULE);
+        for (let i = 0; i < numROs; i++) {
+            const e = this.entries[i];
+            if (e !== undefined)
+                refs[i] = e.modelRef;
+        }
+        this.device.queue.writeBuffer(this.inputBuf, 0, refs.buffer, refs.byteOffset, refs.byteLength);
+        const params = new Uint32Array([numROs, this.declared, 0, 0]);
+        this.device.queue.writeBuffer(this.paramsBuf, 0, params.buffer, params.byteOffset, params.byteLength);
+    }
+    /**
+     * Encode the compute dispatch + the staging copy into `enc`. Caller
+     * is responsible for `device.queue.submit([enc.finish()])`. Returns
+     * the number of ROs the kernel was dispatched against.
+     */
+    dispatch(arenaBuf, numROs, enc) {
+        this.ensureResources(arenaBuf, numROs);
+        this.uploadInputs(numROs);
+        const pass = enc.beginComputePass({ label: "gpuDerivedModes/pass" });
+        pass.setPipeline(this.pipeline);
+        pass.setBindGroup(0, this.bindGroup);
+        pass.dispatchWorkgroups(Math.ceil(numROs / WG_SIZE));
+        pass.end();
+        const bytes = numROs * 4;
+        enc.copyBufferToBuffer(this.outputBuf, 0, this.stagingBuf, 0, bytes);
+        return numROs;
+    }
+    /**
+     * mapAsync the staging buffer + copy values into `lastOutput`.
+     * Resolves on the GPU finishing the copy. Caller diffs lastOutput
+     * against its own cached values to find changed ROs.
+     */
+    async finish(numROs) {
+        if (this.stagingBuf === null || numROs === 0)
+            return;
+        const bytes = numROs * 4;
+        await this.stagingBuf.mapAsync(GPUMapMode.READ, 0, bytes);
+        const view = new Uint32Array(this.stagingBuf.getMappedRange(0, bytes).slice(0));
+        this.stagingBuf.unmap();
+        this.lastOutput.set(view, 0);
+    }
+    dispose() {
+        this.inputBuf?.destroy();
+        this.outputBuf?.destroy();
+        this.stagingBuf?.destroy();
+        this.paramsBuf?.destroy();
+    }
+}
+//# sourceMappingURL=gpuDispatcher.js.map

package/dist/runtime/derivedModes/gpuDispatcher.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"gpuDispatcher.js","sourceRoot":"","sources":["../../../src/runtime/derivedModes/gpuDispatcher.ts"],"names":[],"mappings":"AAAA,oEAAoE;AACpE,kDAAkD;AAClD,EAAE;AACF,YAAY;AACZ,YAAY;AACZ,EAAE;AACF,qDAAqD;AACrD,sEAAsE;AACtE,kEAAkE;AAClE,kDAAkD;AAClD,sDAAsD;AACtD,gEAAgE;AAChE,wCAAwC;AACxC,sEAAsE;AACtE,uEAAuE;AACvE,oBAAoB;AACpB,iEAAiE;AACjE,iEAAiE;AACjE,kEAAkE;AAClE,kDAAkD;AAClD,EAAE;AACF,mDAAmD;AAEnD,OAAO,EAAE,yBAAyB,EAAE,WAAW,EAAE,MAAM,gBAAgB,CAAC;AAGxE,MAAM,OAAO,GAAG,UAAU,CAAC;AAS3B,MAAM,OAAO,GAAG,EAAE,CAAC;AACnB,MAAM,IAAI,GAAG,CAAC,CAAS,EAAU,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC;IAAE,CAAC,KAAK,CAAC,CAAC,CAAC,OAAO,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AAElG,MAAM,OAAO,oBAAoB;IACtB,MAAM,CAAY;IAEV,OAAO,GAA4B,EAAE,CAAC;IAC/C,SAAS,GAAG,CAAC,CAAC;IACd,KAAK,GAAG,KAAK,CAAC;IAEtB,wDAAwD;IAChD,QAAQ,GAAuB,IAAI,CAAC;IACpC,SAAS,GAAsB,IAAI,CAAC;IACpC,UAAU,GAAqB,IAAI,CAAC;IACpC,SAAS,GAAsB,IAAI,CAAC;IACpC,QAAQ,GAAgC,IAAI,CAAC;IAC7C,SAAS,GAAyB,IAAI,CAAC;IACvC,MAAM,GAAkC,IAAI,CAAC;IACrD,yEAAyE;IACjE,QAAQ,GAAG,CAAC,CAAC;IACrB,uEAAuE;IAC/D,QAAQ,GAAqB,IAAI,CAAC;IAC1C;kEAC8D;IACtD,QAAQ,GAAW,WAAW,CAAC,IAAI,CAAC;IAE5C,mEAAmE;IAC1D,UAAU,CAAc;IAEjC,YAAY,MAAiB;QAC3B,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,UAAU,GAAG,IAAI,WAAW,CAAC,CAAC,CAAC,CAAC;IACvC,CAAC;IAED,IAAI,UAAU,KAAa,OAAO,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC;IAEnD,UAAU,CAAC,MAAc,EAAE,QAAgB,EAAE,QAAkB;QAC7D,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,EAAE,QAAQ,EAAE,WAAW,EAAE,WAAW,CAAC,QAAQ,CAAC,EAAE,CAAC;QACxE,IAAI,CAAC,SAAS,EAAE,CAAC;QACjB,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;QAClB,IAAI,CAAC,QAAQ,GAAG,WAAW,CAAC,QAAQ,CAAC,CAAC;IACxC,CAAC;IAED,YAAY,CAAC,MAAc;QACzB,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,KAAK,SAAS;YAAE,OAAO;QAC/C,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,SAAS,CAAC;QACjC,IAAI,CAAC,SAAS,EAAE,CAAC;QACjB,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;IACpB,CAAC;IAED,SAAS,KAAW,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,CAAC,CAAC;IACxC,YAAY,KAAc,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,CAAC,OAAO,GAAG,CAAC,CAAC,CAAC;IAE3E,eAAe,CAAC,QAAmB,EAAE,MAAc;QACzD,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC;QAC1C,IAAI,IAAI,CAAC,SAAS,KAAK,IAAI,IAAI,OAAO,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC;YACvD,IAAI,CAAC,QAAU,EAAE,OAAO,EAAE,CAAC;YAC3B,IAAI,CAAC,SAAU,EAAE,OAAO,EAAE,CAAC;YAC3B,IAAI,CAAC,UAAU,EAAE,OAAO,EAAE,CAAC;YAC3B,MAAM,KAAK,GAAG,OAAO,GAAG,CAAC,CAAC;YAC1B,IAAI,CAAC,QAAQ,GAAK,IAAI,CAAC,MAAM,CAAC,YAAY,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,cAAc,CAAC,OAAO,GAAG,cAAc,CAAC,QAAQ,EAAE,KAAK,EAAE,oBAAoB,EAAE,CAAC,CAAC;YAClJ,IAAI,CAAC,SAAS,GAAI,IAAI,CAAC,MAAM,CAAC,YAAY,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,cAAc,CAAC,OAAO,GAAG,cAAc,CAAC,QAAQ,GAAG,cAAc,CAAC,QAAQ,EAAE,KAAK,EAAE,qBAAqB,EAAE,CAAC,CAAC;YAC7K,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,MAAM,CAAC,YAAY,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,cAAc,CAAC,QAAQ,GAAG,cAAc,CAAC,QAAQ,EAAE,KAAK,EAAE,yBAAyB,EAAE,CAAC,CAAC;YACvJ,IAAoC,CAAC,UAAU,GAAG,IAAI,WAAW,CAAC,OAAO,CAAC,CAAC;YAC5E,IAAI,CAAC,QAAQ,GAAG,OAAO,CAAC;YACxB,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC;QACxB,CAAC;QACD,IAAI,IAAI,CAAC,SAAS,KAAK,IAAI,EAAE,CAAC;YAC5B,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC,MAAM,CAAC,YAAY,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,KAAK,EAAE,cAAc,CAAC,OAAO,GAAG,cAAc,CAAC,QAAQ,EAAE,KAAK,EAAE,wBAAwB,EAAE,CAAC,CAAC;QACpJ,CAAC;QACD,IAAI,IAAI,CAAC,MAAM,KAAK,IAAI,EAAE,CAAC;YACzB,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,qBAAqB,CAAC;gBAC9C,KAAK,EAAE,qBAAqB;gBAC5B,OAAO,EAAE;oBACP,EAAE,OAAO,EAAE,CAAC,EAAE,UAAU,EAAE,cAAc,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,IAAI,EAAE,mBAAmB,EAAE,EAAE;oBACzF,EAAE,OAAO,EAAE,CAAC,EAAE,UAAU,EAAE,cAAc,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,IAAI,EAAE,mBAAmB,EAAE,EAAE;oBACzF,EAAE,OAAO,EAAE,CAAC,EAAE,UAAU,EAAE,cAAc,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,IAAI,EAAE,SAAS,EAAE,EAAE;oBAC/E,EAAE,OAAO,EAAE,CAAC,EAAE,UAAU,EAAE,cAAc,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,IAAI,EAAE,SAAS,EAAE,EAAE;iBAChF;aACF,CAAC,CAAC;QACL,CAAC;QACD,IAAI,IAAI,CAAC,QAAQ,KAAK,IAAI,EAAE,CAAC;YAC3B,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,kBAAkB,CAAC,EAAE,IAAI,EAAE,yBAAyB,EAAE,KAAK,EAAE,wBAAwB,EAAE,CAAC,CAAC;YACpH,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,MAAM,CAAC,qBAAqB,CAAC;gBAChD,KAAK,EAAE,0BAA0B;gBACjC,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC,oBAAoB,CAAC,EAAE,gBAAgB,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC7E,OAAO,EAAE,EAAE,MAAM,EAAE,UAAU,EAAE,UAAU,EAAE;aAC5C,CAAC,CAAC;QACL,CAAC;QACD,IAAI,IAAI,CAAC,SAAS,KAAK,IAAI,IAAI,IAAI,CAAC,QAAQ,KAAK,QAAQ,EAAE,CAAC;YAC1D,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC,MAAM,CAAC,eAAe,CAAC;gBAC3C,KAAK,EAAE,2BAA2B;gBAClC,MAAM,EAAE,IAAI,CAAC,MAAM;gBACnB,OAAO,EAAE;oBACP,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,EAAE;oBAC9C,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,QAAS,EAAG,EAAE;oBACrD,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,SAAU,EAAE,EAAE;oBACrD,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,SAAU,EAAE,EAAE;iBACtD;aACF,CAAC,CAAC;YACH,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,oEAAoE;IAC5D,YAAY,CAAC,MAAc;QACjC,MAAM,IAAI,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC1D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAChC,MAAM,CAAC,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;YAC1B,IAAI,CAAC,KAAK,SAAS;gBAAE,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC;QAC5C,CAAC;QACD,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,QAAS,EAAE,CAAC,EAAE,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,UAAU,CAAC,CAAC;QAChG,MAAM,MAAM,GAAG,IAAI,WAAW,CAAC,CAAC,MAAM,EAAE,IAAI,CAAC,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QAC9D,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,SAAU,EAAE,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,UAAU,EAAE,MAAM,CAAC,UAAU,CAAC,CAAC;IACzG,CAAC;IAED;;;;OAIG;IACH,QAAQ,CAAC,QAAmB,EAAE,MAAc,EAAE,GAAsB;QAClE,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC;QACvC,IAAI,CAAC,YAAY,CAAC,MAAM,CAAC,CAAC;QAC1B,MAAM,IAAI,GAAG,GAAG,CAAC,gBAAgB,CAAC,EAAE,KAAK,EAAE,sBAAsB,EAAE,CAAC,CAAC;QACrE,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,QAAS,CAAC,CAAC;QACjC,IAAI,CAAC,YAAY,CAAC,CAAC,EAAE,IAAI,CAAC,SAAU,CAAC,CAAC;QACtC,IAAI,CAAC,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,CAAC,CAAC;QACrD,IAAI,CAAC,GAAG,EAAE,CAAC;QACX,MAAM,KAAK,GAAG,MAAM,GAAG,CAAC,CAAC;QACzB,GAAG,CAAC,kBAAkB,CAAC,IAAI,CAAC,SAAU,EAAE,CAAC,EAAE,IAAI,CAAC,UAAW,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC;QACvE,OAAO,MAAM,CAAC;IAChB,CAAC;IAED;;;;OAIG;IACH,KAAK,CAAC,MAAM,CAAC,MAAc;QACzB,IAAI,IAAI,CAAC,UAAU,KAAK,IAAI,IAAI,MAAM,KAAK,CAAC;YAAE,OAAO;QACrD,MAAM,KAAK,GAAG,MAAM,GAAG,CAAC,CAAC;QACzB,MAAM,IAAI,CAAC,UAAU,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,cAAc,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAChF,IAAI,CAAC,UAAU,CAAC,KAAK,EAAE,CAAC;QACvB,IAAoC,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;IAChE,CAAC;IAED,OAAO;QACL,IAAI,CAAC,QAAU,EAAE,OAAO,EAAE,CAAC;QAC3B,IAAI,CAAC,SAAU,EAAE,OAAO,EAAE,CAAC;QAC3B,IAAI,CAAC,UAAU,EAAE,OAAO,EAAE,CAAC;QAC3B,IAAI,CAAC,SAAU,EAAE,OAAO,EAAE,CAAC;IAC7B,CAAC;CACF"}

package/dist/runtime/derivedModes/gpuKernel.d.ts

@@ -0,0 +1,9 @@
+export declare const GPU_FLIP_CULL_BY_DET_WGSL = "\n// Per-RO record. Index i corresponds to drawId i in the scene's\n// flat RO table.\nstruct GpuRuleParams {\n  numROs:     u32,\n  declared:   u32,   // 0 = none, 1 = front, 2 = back  (must match CULL_MODE[] order in bitfield.ts)\n  _pad0:      u32,\n  _pad1:      u32,\n};\n\n@group(0) @binding(0) var<storage, read>       arena:        array<u32>;\n// One per-RO entry: byte offset within the arena of the ROs ModelTrafo\n// allocation header. Data is at offset + 16 (ALLOC_HEADER_PAD_TO).\n// 0xFFFFFFFFu means \"this RO does NOT have a flip-cull-by-det rule\".\n@group(0) @binding(1) var<storage, read>       roInputRefs:  array<u32>;\n@group(0) @binding(2) var<storage, read_write> roOutputs:    array<u32>;\n@group(0) @binding(3) var<uniform>             uIn:          GpuRuleParams;\n\n// Read a row-major mat4 from the arena at byte offset `refBytes + 16`\n// (header is 16 bytes wide due to ALLOC_HEADER_PAD_TO). Returns the\n// 3\u00D73 upper-left in row-major order: [m00,m01,m02, m10,m11,m12, m20,m21,m22].\nfn loadUpper3x3(refBytes: u32) -> array<f32, 9> {\n  let baseU32 = (refBytes + 16u) >> 2u;\n  // The arena is a Float32 storage of mat4 in row-major, 16 floats.\n  // Read 9 of them.\n  return array<f32, 9>(\n    bitcast<f32>(arena[baseU32 + 0u]),  bitcast<f32>(arena[baseU32 + 1u]),  bitcast<f32>(arena[baseU32 + 2u]),\n    bitcast<f32>(arena[baseU32 + 4u]),  bitcast<f32>(arena[baseU32 + 5u]),  bitcast<f32>(arena[baseU32 + 6u]),\n    bitcast<f32>(arena[baseU32 + 8u]),  bitcast<f32>(arena[baseU32 + 9u]),  bitcast<f32>(arena[baseU32 + 10u]),\n  );\n}\n\nfn det3x3(m: array<f32, 9>) -> f32 {\n  return m[0] * (m[4] * m[8] - m[5] * m[7])\n       - m[1] * (m[3] * m[8] - m[5] * m[6])\n       + m[2] * (m[3] * m[7] - m[4] * m[6]);\n}\n\n// Mirror of bitfield.ts CULL_MODE[] order: 0=none, 1=front, 2=back.\nfn flipCull(c: u32) -> u32 {\n  if (c == 1u) { return 2u; } // front -> back\n  if (c == 2u) { return 1u; } // back  -> front\n  return c;                   // none\n}\n\n@compute @workgroup_size(64)\nfn evaluate(@builtin(global_invocation_id) gid: vec3<u32>) {\n  let i = gid.x;\n  if (i >= uIn.numROs) { return; }\n  let ref = roInputRefs[i];\n  if (ref == 0xFFFFFFFFu) {\n    // No rule on this RO; leave existing output untouched. Allows a\n    // single kernel dispatch to skip non-participating ROs.\n    return;\n  }\n  let m = loadUpper3x3(ref);\n  let d = det3x3(m);\n  let declared = uIn.declared;\n  let outVal = select(declared, flipCull(declared), d < 0.0);\n  roOutputs[i] = outVal;\n}\n";
+/** Map between bitfield.ts's CULL_MODE order and the kernel's u32 enum. */
+export declare const CULL_TO_U32: {
+    readonly none: 0;
+    readonly front: 1;
+    readonly back: 2;
+};
+export declare const U32_TO_CULL: readonly ["none", "front", "back"];
+//# sourceMappingURL=gpuKernel.d.ts.map

package/dist/runtime/derivedModes/gpuKernel.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"gpuKernel.d.ts","sourceRoot":"","sources":["../../../src/runtime/derivedModes/gpuKernel.ts"],"names":[],"mappings":"AAgCA,eAAO,MAAM,yBAAyB,s+EA6DrC,CAAC;AAEF,2EAA2E;AAC3E,eAAO,MAAM,WAAW;;;;CAAgD,CAAC;AACzE,eAAO,MAAM,WAAW,oCAAqC,CAAC"}

package/dist/runtime/derivedModes/gpuKernel.js

@@ -0,0 +1,97 @@
+// GPU-evaluated derived-mode rules — minimum viable kernel.
+//
+// CPU eval (Task 2 Phase 1-3, 0.9.17) runs rule closures in JS. For
+// rules whose inputs live GPU-side (per-RO trafos, etc.) JS evaluation
+// fans out CPU work that's wasted: the trafo bytes are right there in
+// the arena. This module runs the rule on the GPU instead, in one
+// dispatch over all dirty ROs per frame.
+//
+// v1 scope is deliberately narrow: ONE hard-coded WGSL kernel for
+// "flip cullMode by sign of det(upper3x3 of an arena mat4)". Picked
+// because it's the motivating case from docs/derived-modes.md and
+// proves the GPU-eval path end-to-end. Generalizes via a traced IR
+// builder in v2 (see derived-mode-rules-plan.md Phase 5).
+//
+// Architecture
+// ============
+//
+//   - One **module-wide** `GpuDerivedModesScene` per heap scene. Holds:
+//       * `roOutputBuf`  — `array<u32>` sized to scene.numROs.
+//                          Each entry = the rule's modeKey contribution.
+//       * `roInputBuf`   — `array<u32>` (one ref-into-arena per RO)
+//                          telling the kernel where to read the input.
+//       * `stagingBuf`   — MAP_READ + COPY_DST mirror of roOutputBuf,
+//                          for the per-frame readback.
+//   - Dispatch each frame in `update()`, ONLY when something dirty.
+//   - `mapAsync(staging)` resolves on a microtask; the CPU diffs the
+//     buffer against `roLastKey[]`. Any change → mark drawId dirty;
+//     the existing rebucket flow does the rest.
+//
+// The dirty gate keeps steady-state cost ≈ 0 — when no input aval
+// marks AND no RO add/remove, we skip the dispatch + readback.
+export const GPU_FLIP_CULL_BY_DET_WGSL = /* wgsl */ `
+// Per-RO record. Index i corresponds to drawId i in the scene's
+// flat RO table.
+struct GpuRuleParams {
+  numROs:     u32,
+  declared:   u32,   // 0 = none, 1 = front, 2 = back  (must match CULL_MODE[] order in bitfield.ts)
+  _pad0:      u32,
+  _pad1:      u32,
+};
+
+@group(0) @binding(0) var<storage, read>       arena:        array<u32>;
+// One per-RO entry: byte offset within the arena of the ROs ModelTrafo
+// allocation header. Data is at offset + 16 (ALLOC_HEADER_PAD_TO).
+// 0xFFFFFFFFu means "this RO does NOT have a flip-cull-by-det rule".
+@group(0) @binding(1) var<storage, read>       roInputRefs:  array<u32>;
+@group(0) @binding(2) var<storage, read_write> roOutputs:    array<u32>;
+@group(0) @binding(3) var<uniform>             uIn:          GpuRuleParams;
+
+// Read a row-major mat4 from the arena at byte offset \`refBytes + 16\`
+// (header is 16 bytes wide due to ALLOC_HEADER_PAD_TO). Returns the
+// 3×3 upper-left in row-major order: [m00,m01,m02, m10,m11,m12, m20,m21,m22].
+fn loadUpper3x3(refBytes: u32) -> array<f32, 9> {
+  let baseU32 = (refBytes + 16u) >> 2u;
+  // The arena is a Float32 storage of mat4 in row-major, 16 floats.
+  // Read 9 of them.
+  return array<f32, 9>(
+    bitcast<f32>(arena[baseU32 + 0u]),  bitcast<f32>(arena[baseU32 + 1u]),  bitcast<f32>(arena[baseU32 + 2u]),
+    bitcast<f32>(arena[baseU32 + 4u]),  bitcast<f32>(arena[baseU32 + 5u]),  bitcast<f32>(arena[baseU32 + 6u]),
+    bitcast<f32>(arena[baseU32 + 8u]),  bitcast<f32>(arena[baseU32 + 9u]),  bitcast<f32>(arena[baseU32 + 10u]),
+  );
+}
+
+fn det3x3(m: array<f32, 9>) -> f32 {
+  return m[0] * (m[4] * m[8] - m[5] * m[7])
+       - m[1] * (m[3] * m[8] - m[5] * m[6])
+       + m[2] * (m[3] * m[7] - m[4] * m[6]);
+}
+
+// Mirror of bitfield.ts CULL_MODE[] order: 0=none, 1=front, 2=back.
+fn flipCull(c: u32) -> u32 {
+  if (c == 1u) { return 2u; } // front -> back
+  if (c == 2u) { return 1u; } // back  -> front
+  return c;                   // none
+}
+
+@compute @workgroup_size(64)
+fn evaluate(@builtin(global_invocation_id) gid: vec3<u32>) {
+  let i = gid.x;
+  if (i >= uIn.numROs) { return; }
+  let ref = roInputRefs[i];
+  if (ref == 0xFFFFFFFFu) {
+    // No rule on this RO; leave existing output untouched. Allows a
+    // single kernel dispatch to skip non-participating ROs.
+    return;
+  }
+  let m = loadUpper3x3(ref);
+  let d = det3x3(m);
+  let declared = uIn.declared;
+  let outVal = select(declared, flipCull(declared), d < 0.0);
+  roOutputs[i] = outVal;
+}
+`;
+/** Map between bitfield.ts's CULL_MODE order and the kernel's u32 enum. */
+export const CULL_TO_U32 = { "none": 0, "front": 1, "back": 2 };
+export const U32_TO_CULL = ["none", "front", "back"];
+//# sourceMappingURL=gpuKernel.js.map

package/dist/runtime/derivedModes/gpuKernel.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"gpuKernel.js","sourceRoot":"","sources":["../../../src/runtime/derivedModes/gpuKernel.ts"],"names":[],"mappings":"AAAA,4DAA4D;AAC5D,EAAE;AACF,oEAAoE;AACpE,uEAAuE;AACvE,sEAAsE;AACtE,kEAAkE;AAClE,yCAAyC;AACzC,EAAE;AACF,kEAAkE;AAClE,oEAAoE;AACpE,kEAAkE;AAClE,mEAAmE;AACnE,0DAA0D;AAC1D,EAAE;AACF,eAAe;AACf,eAAe;AACf,EAAE;AACF,wEAAwE;AACxE,+DAA+D;AAC/D,yEAAyE;AACzE,oEAAoE;AACpE,uEAAuE;AACvE,sEAAsE;AACtE,uDAAuD;AACvD,oEAAoE;AACpE,qEAAqE;AACrE,oEAAoE;AACpE,gDAAgD;AAChD,EAAE;AACF,kEAAkE;AAClE,+DAA+D;AAE/D,MAAM,CAAC,MAAM,yBAAyB,GAAG,UAAU,CAAC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CA6DnD,CAAC;AAEF,2EAA2E;AAC3E,MAAM,CAAC,MAAM,WAAW,GAAG,EAAE,MAAM,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAW,CAAC;AACzE,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,CAAU,CAAC"}

package/dist/runtime/derivedModes/rule.d.ts

@@ -17,10 +17,28 @@ export type ModeValue<A extends ModeAxis> = A extends "cull" ? CullMode : A exte
  *     want to pass declared through unchanged.
  */
 export type DerivedModeBuilder<A extends ModeAxis> = (u: Record<string, unknown>, declared: ModeValue<A>) => ModeValue<A>;
+/**
+ * Opt-in GPU evaluation marker. When present on a rule, the heap
+ * scene runs a hard-coded WGSL kernel (see `./gpuKernel.ts`) instead
+ * of the CPU closure. The closure is retained for fallback (e.g.
+ * MockGPU tests, devices without compute, or before frame 0).
+ *
+ * v1 supports one kernel: `"flipCullByDeterminant"`. v2 generalizes
+ * via a traced IR builder.
+ */
+export interface GpuRuleSpec<A extends ModeAxis = ModeAxis> {
+    readonly kernel: "flipCullByDeterminant";
+    /** Name of the input uniform the kernel reads (e.g. "ModelTrafo"). */
+    readonly inputUniform: string;
+    /** Declared (SG-scope) value the kernel mutates by. */
+    readonly declared: ModeValue<A>;
+}
 export interface DerivedModeRule<A extends ModeAxis = ModeAxis> {
     readonly __derivedModeRule: true;
     readonly axis: A;
     readonly evaluate: DerivedModeBuilder<A>;
+    /** Opt-in GPU evaluation. Absent → CPU closure path. */
+    readonly gpu?: GpuRuleSpec<A>;
     /**
      * Conservative over-approximation of the rule's output set. Used by
      * the build-time pre-warm in v2 to enumerate the pipelines a scene
@@ -56,4 +74,19 @@ export declare function isDerivedModeRule(x: unknown): x is DerivedModeRule;
 /** Convenience: flip 'back' ↔ 'front' (and pass through 'none').
  *  Useful in `(u, declared) => mirrored ? flipCull(declared) : declared`. */
 export declare function flipCull(c: CullMode): CullMode;
+/**
+ * Construct a derived-mode rule that flips cullMode based on the sign
+ * of `det(upperLeft3x3(u.<inputUniform>))`. Runs on the GPU when
+ * heapScene's compute path is available; the same logic is mirrored
+ * in the CPU closure for tests and fallback.
+ *
+ *   const rule = gpuFlipCullByDeterminant("ModelTrafo", "back");
+ *   spec.modeRules = { cull: rule };
+ *
+ * Useful for mirrored geometry: scenes that animate a per-RO trafo
+ * with potentially-negative scale don't pay CPU per-RO determinant
+ * cost — the kernel walks the arena once per frame for ALL ROs that
+ * carry this rule.
+ */
+export declare function gpuFlipCullByDeterminant(inputUniform: string, declared?: CullMode): DerivedModeRule<"cull">;
 //# sourceMappingURL=rule.d.ts.map

package/dist/runtime/derivedModes/rule.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"rule.d.ts","sourceRoot":"","sources":["../../../src/runtime/derivedModes/rule.ts"],"names":[],"mappings":"AAoBA,OAAO,KAAK,EAAE,QAAQ,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,2BAA2B,CAAC;AAE/E;;;wDAGwD;AACxD,MAAM,MAAM,QAAQ,GAChB,MAAM,GAAG,WAAW,GAAG,UAAU,GACjC,cAAc,GAAG,YAAY,GAC7B,iBAAiB,CAAC;AAEtB,kCAAkC;AAClC,MAAM,MAAM,SAAS,CAAC,CAAC,SAAS,QAAQ,IACtC,CAAC,SAAS,MAAM,GAAa,QAAQ,GACrC,CAAC,SAAS,WAAW,GAAQ,SAAS,GACtC,CAAC,SAAS,UAAU,GAAS,QAAQ,GACrC,CAAC,SAAS,cAAc,GAAK,kBAAkB,GAC/C,CAAC,SAAS,YAAY,GAAO,OAAO,GACpC,CAAC,SAAS,iBAAiB,GAAG,OAAO,GACrC,KAAK,CAAC;AAER;;;;;;;;;GASG;AACH,MAAM,MAAM,kBAAkB,CAAC,CAAC,SAAS,QAAQ,IAAI,CACnD,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EAC1B,QAAQ,EAAE,SAAS,CAAC,CAAC,CAAC,KACnB,SAAS,CAAC,CAAC,CAAC,CAAC;AAElB,MAAM,WAAW,eAAe,CAAC,CAAC,SAAS,QAAQ,GAAG,QAAQ;IAC5D,QAAQ,CAAC,iBAAiB,EAAE,IAAI,CAAC;IACjC,QAAQ,CAAC,IAAI,EAAK,CAAC,CAAC;IACpB,QAAQ,CAAC,QAAQ,EAAE,kBAAkB,CAAC,CAAC,CAAC,CAAC;IACzC;;;;;OAKG;IACH,QAAQ,CAAC,MAAM,CAAC,EAAE,aAAa,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;CAC/C;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,WAAW,CAAC,CAAC,SAAS,QAAQ,EAC5C,IAAI,EAAE,CAAC,EACP,QAAQ,EAAE,kBAAkB,CAAC,CAAC,CAAC,EAC/B,OAAO,GAAE;IAAE,QAAQ,CAAC,MAAM,CAAC,EAAE,aAAa,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAA;CAAO,GAC9D,eAAe,CAAC,CAAC,CAAC,CAKpB;AAED;mEACmE;AACnE,wBAAgB,iBAAiB,CAAC,CAAC,EAAE,OAAO,GAAG,CAAC,IAAI,eAAe,CAGlE;AAED;6EAC6E;AAC7E,wBAAgB,QAAQ,CAAC,CAAC,EAAE,QAAQ,GAAG,QAAQ,CAE9C"}
+{"version":3,"file":"rule.d.ts","sourceRoot":"","sources":["../../../src/runtime/derivedModes/rule.ts"],"names":[],"mappings":"AAoBA,OAAO,KAAK,EAAE,QAAQ,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,2BAA2B,CAAC;AAE/E;;;wDAGwD;AACxD,MAAM,MAAM,QAAQ,GAChB,MAAM,GAAG,WAAW,GAAG,UAAU,GACjC,cAAc,GAAG,YAAY,GAC7B,iBAAiB,CAAC;AAEtB,kCAAkC;AAClC,MAAM,MAAM,SAAS,CAAC,CAAC,SAAS,QAAQ,IACtC,CAAC,SAAS,MAAM,GAAa,QAAQ,GACrC,CAAC,SAAS,WAAW,GAAQ,SAAS,GACtC,CAAC,SAAS,UAAU,GAAS,QAAQ,GACrC,CAAC,SAAS,cAAc,GAAK,kBAAkB,GAC/C,CAAC,SAAS,YAAY,GAAO,OAAO,GACpC,CAAC,SAAS,iBAAiB,GAAG,OAAO,GACrC,KAAK,CAAC;AAER;;;;;;;;;GASG;AACH,MAAM,MAAM,kBAAkB,CAAC,CAAC,SAAS,QAAQ,IAAI,CACnD,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EAC1B,QAAQ,EAAE,SAAS,CAAC,CAAC,CAAC,KACnB,SAAS,CAAC,CAAC,CAAC,CAAC;AAElB;;;;;;;;GAQG;AACH,MAAM,WAAW,WAAW,CAAC,CAAC,SAAS,QAAQ,GAAG,QAAQ;IACxD,QAAQ,CAAC,MAAM,EAAE,uBAAuB,CAAC;IACzC,sEAAsE;IACtE,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAC;IAC9B,uDAAuD;IACvD,QAAQ,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC;CACjC;AAED,MAAM,WAAW,eAAe,CAAC,CAAC,SAAS,QAAQ,GAAG,QAAQ;IAC5D,QAAQ,CAAC,iBAAiB,EAAE,IAAI,CAAC;IACjC,QAAQ,CAAC,IAAI,EAAK,CAAC,CAAC;IACpB,QAAQ,CAAC,QAAQ,EAAE,kBAAkB,CAAC,CAAC,CAAC,CAAC;IACzC,wDAAwD;IACxD,QAAQ,CAAC,GAAG,CAAC,EAAK,WAAW,CAAC,CAAC,CAAC,CAAC;IACjC;;;;;OAKG;IACH,QAAQ,CAAC,MAAM,CAAC,EAAE,aAAa,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;CAC/C;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,WAAW,CAAC,CAAC,SAAS,QAAQ,EAC5C,IAAI,EAAE,CAAC,EACP,QAAQ,EAAE,kBAAkB,CAAC,CAAC,CAAC,EAC/B,OAAO,GAAE;IAAE,QAAQ,CAAC,MAAM,CAAC,EAAE,aAAa,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAA;CAAO,GAC9D,eAAe,CAAC,CAAC,CAAC,CAKpB;AAED;mEACmE;AACnE,wBAAgB,iBAAiB,CAAC,CAAC,EAAE,OAAO,GAAG,CAAC,IAAI,eAAe,CAGlE;AAED;6EAC6E;AAC7E,wBAAgB,QAAQ,CAAC,CAAC,EAAE,QAAQ,GAAG,QAAQ,CAE9C;AAED;;;;;;;;;;;;;GAaG;AACH,wBAAgB,wBAAwB,CACtC,YAAY,EAAE,MAAM,EACpB,QAAQ,GAAE,QAAiB,GAC1B,eAAe,CAAC,MAAM,CAAC,CAwBzB"}

package/dist/runtime/derivedModes/rule.js

@@ -52,4 +52,43 @@ export function isDerivedModeRule(x) {
 export function flipCull(c) {
     return c === "back" ? "front" : c === "front" ? "back" : "none";
 }
+/**
+ * Construct a derived-mode rule that flips cullMode based on the sign
+ * of `det(upperLeft3x3(u.<inputUniform>))`. Runs on the GPU when
+ * heapScene's compute path is available; the same logic is mirrored
+ * in the CPU closure for tests and fallback.
+ *
+ *   const rule = gpuFlipCullByDeterminant("ModelTrafo", "back");
+ *   spec.modeRules = { cull: rule };
+ *
+ * Useful for mirrored geometry: scenes that animate a per-RO trafo
+ * with potentially-negative scale don't pay CPU per-RO determinant
+ * cost — the kernel walks the arena once per frame for ALL ROs that
+ * carry this rule.
+ */
+export function gpuFlipCullByDeterminant(inputUniform, declared = "back") {
+    return {
+        __derivedModeRule: true,
+        axis: "cull",
+        domain: ["back", "front", "none"],
+        gpu: { kernel: "flipCullByDeterminant", inputUniform, declared },
+        evaluate: (u, declaredFromCtx) => {
+            const m = u[inputUniform];
+            // Accept Trafo3d ({forward: M44d}), M44 ({_data}), or row-major
+            // Float32Array. Reads upper-left 3x3.
+            const data = m.forward?._data ??
+                m._data ??
+                m;
+            if (data === undefined)
+                return declaredFromCtx;
+            const m00 = data[0], m01 = data[1], m02 = data[2];
+            const m10 = data[4], m11 = data[5], m12 = data[6];
+            const m20 = data[8], m21 = data[9], m22 = data[10];
+            const det = m00 * (m11 * m22 - m12 * m21)
+                - m01 * (m10 * m22 - m12 * m20)
+                + m02 * (m10 * m21 - m11 * m20);
+            return det < 0 ? flipCull(declaredFromCtx) : declaredFromCtx;
+        },
+    };
+}
 //# sourceMappingURL=rule.js.map

package/dist/runtime/derivedModes/rule.js.map

@@ -1 +1 @@
-{"version":3,"file":"rule.js","sourceRoot":"","sources":["../../../src/runtime/derivedModes/rule.ts"],"names":[],"mappings":"AAAA,kEAAkE;AAClE,oEAAoE;AACpE,iBAAiB;AACjB,EAAE;AACF,mEAAmE;AACnE,kEAAkE;AAClE,oEAAoE;AACpE,oEAAoE;AACpE,iEAAiE;AACjE,iEAAiE;AACjE,sBAAsB;AACtB,EAAE;AACF,iEAAiE;AACjE,oEAAoE;AACpE,4DAA4D;AAC5D,EAAE;AACF,uEAAuE;AACvE,oEAAoE;AACpE,oDAAoD;AAmDpD;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,WAAW,CACzB,IAAO,EACP,QAA+B,EAC/B,UAA6D,EAAE;IAE/D,MAAM,GAAG,GAAuB,OAAO,CAAC,MAAM,KAAK,SAAS;QAC1D,CAAC,CAAC,EAAE,iBAAiB,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE;QACrE,CAAC,CAAC,EAAE,iBAAiB,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC;IAChD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;mEACmE;AACnE,MAAM,UAAU,iBAAiB,CAAC,CAAU;IAC1C,OAAO,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,IAAI;WAClC,CAAqC,CAAC,iBAAiB,KAAK,IAAI,CAAC;AAC3E,CAAC;AAED;6EAC6E;AAC7E,MAAM,UAAU,QAAQ,CAAC,CAAW;IAClC,OAAO,CAAC,KAAK,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,KAAK,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC;AAClE,CAAC"}
+{"version":3,"file":"rule.js","sourceRoot":"","sources":["../../../src/runtime/derivedModes/rule.ts"],"names":[],"mappings":"AAAA,kEAAkE;AAClE,oEAAoE;AACpE,iBAAiB;AACjB,EAAE;AACF,mEAAmE;AACnE,kEAAkE;AAClE,oEAAoE;AACpE,oEAAoE;AACpE,iEAAiE;AACjE,iEAAiE;AACjE,sBAAsB;AACtB,EAAE;AACF,iEAAiE;AACjE,oEAAoE;AACpE,4DAA4D;AAC5D,EAAE;AACF,uEAAuE;AACvE,oEAAoE;AACpE,oDAAoD;AAsEpD;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,WAAW,CACzB,IAAO,EACP,QAA+B,EAC/B,UAA6D,EAAE;IAE/D,MAAM,GAAG,GAAuB,OAAO,CAAC,MAAM,KAAK,SAAS;QAC1D,CAAC,CAAC,EAAE,iBAAiB,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE;QACrE,CAAC,CAAC,EAAE,iBAAiB,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC;IAChD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;mEACmE;AACnE,MAAM,UAAU,iBAAiB,CAAC,CAAU;IAC1C,OAAO,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,IAAI;WAClC,CAAqC,CAAC,iBAAiB,KAAK,IAAI,CAAC;AAC3E,CAAC;AAED;6EAC6E;AAC7E,MAAM,UAAU,QAAQ,CAAC,CAAW;IAClC,OAAO,CAAC,KAAK,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,KAAK,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC;AAClE,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,MAAM,UAAU,wBAAwB,CACtC,YAAoB,EACpB,WAAqB,MAAM;IAE3B,OAAO;QACL,iBAAiB,EAAE,IAAI;QACvB,IAAI,EAAE,MAAM;QACZ,MAAM,EAAE,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,CAAC;QACjC,GAAG,EAAE,EAAE,MAAM,EAAE,uBAAuB,EAAE,YAAY,EAAE,QAAQ,EAAE;QAChE,QAAQ,EAAE,CAAC,CAAC,EAAE,eAAe,EAAE,EAAE;YAC/B,MAAM,CAAC,GAAG,CAAC,CAAC,YAAY,CAAoG,CAAC;YAC7H,gEAAgE;YAChE,sCAAsC;YACtC,MAAM,IAAI,GACP,CAAiD,CAAC,OAAO,EAAE,KAAK;gBAChE,CAAmC,CAAC,KAAK;gBACzC,CAAkC,CAAC;YACtC,IAAI,IAAI,KAAK,SAAS;gBAAE,OAAO,eAAe,CAAC;YAC/C,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,EAAE,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,EAAE,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,CAAC;YACrD,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,EAAE,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,EAAE,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,CAAC;YACrD,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,EAAE,GAAG,GAAG,IAAI,CAAC,CAAC,CAAE,EAAE,GAAG,GAAG,IAAI,CAAC,EAAE,CAAE,CAAC;YACtD,MAAM,GAAG,GAAG,GAAG,GAAG,CAAC,GAAG,GAAG,GAAG,GAAG,GAAG,GAAG,GAAG,CAAC;kBAC7B,GAAG,GAAG,CAAC,GAAG,GAAG,GAAG,GAAG,GAAG,GAAG,GAAG,CAAC;kBAC7B,GAAG,GAAG,CAAC,GAAG,GAAG,GAAG,GAAG,GAAG,GAAG,GAAG,CAAC,CAAC;YAC1C,OAAO,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC,eAAe,CAAC;QAC/D,CAAC;KACF,CAAC;AACJ,CAAC"}

package/dist/runtime/heapScene.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"heapScene.d.ts","sourceRoot":"","sources":["../../src/runtime/heapScene.ts"],"names":[],"mappings":"AA4CA,OAAO,EAA2B,KAAK,GAAG,EAAE,MAAM,uBAAuB,CAAC;AAC1E,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAwB,aAAa,EAAa,MAAM,2BAA2B,CAAC;AAC3F,OAAO,KAAK,EAAE,IAAI,EAAE,IAAI,EAAgD,MAAM,2BAA2B,CAAC;AAC1G,OAAO,KAAK,EAAE,MAAM,EAAkB,MAAM,yBAAyB,CAAC;AACtE,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;AAG9D,OAAO,EAEc,KAAK,oBAAoB,EAE7C,MAAM,iBAAiB,CAAC;AAMzB,OAAO,EAEL,KAAK,SAAS,EAAwB,KAAK,SAAS,EACrD,MAAM,6BAA6B,CAAC;AA6DrC;;;GAGG;AACH,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,SAAS,EAAE,YAAY,CAAC;IACjC,QAAQ,CAAC,OAAO,EAAI,YAAY,CAAC;IACjC,QAAQ,CAAC,OAAO,EAAI,WAAW,CAAC;CACjC;AAgJD,MAAM,WAAW,YAAY;IAC3B;;;OAGG;IACH,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB;;;;;;;;;;;OAWG;IACH,QAAQ,CAAC,MAAM,EAAE;QAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,OAAO,CAAA;KAAE,CAAC;IACtE;;;;;;;;;;OAUG;IACH,QAAQ,CAAC,kBAAkB,CAAC,EAAE;QAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,OAAO,CAAA;KAAE,CAAC;IACnF;;;;OAIG;IACH,QAAQ,CAAC,aAAa,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG,MAAM,CAAC;IAC/C;;;OAGG;IACH,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC,WAAW,CAAC,GAAG,WAAW,CAAC;IAClD;;;;;;OAMG;IACH,QAAQ,CAAC,QAAQ,CAAC,EAAE,cAAc,CAAC;IACnC;;;;;;;;;;;;OAYG;IACH,QAAQ,CAAC,aAAa,CAAC,EAAE,aAAa,CAAC;IACvC;;;;;;;;;;;;;;;;;;;;;;;OAuBG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE;QACnB,QAAQ,CAAC,IAAI,CAAC,EAAY,OAAO,wBAAwB,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC;QACnF,QAAQ,CAAC,SAAS,CAAC,EAAO,OAAO,wBAAwB,EAAE,eAAe,CAAC,WAAW,CAAC,CAAC;QACxF,QAAQ,CAAC,QAAQ,CAAC,EAAQ,OAAO,wBAAwB,EAAE,eAAe,CAAC,UAAU,CAAC,CAAC;QACvF,QAAQ,CAAC,YAAY,CAAC,EAAI,OAAO,wBAAwB,EAAE,eAAe,CAAC,cAAc,CAAC,CAAC;QAC3F,QAAQ,CAAC,UAAU,CAAC,EAAM,OAAO,wBAAwB,EAAE,eAAe,CAAC,YAAY,CAAC,CAAC;QACzF,QAAQ,CAAC,eAAe,CAAC,EAAE,OAAO,wBAAwB,EAAE,eAAe,CAAC,iBAAiB,CAAC,CAAC;KAChG,CAAC;CACH;AAED,MAAM,WAAW,cAAc;IAC7B,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,MAAM,CAAC;IAClB,aAAa,EAAE,MAAM,CAAC;IACtB,4DAA4D;IAC5D,aAAa,EAAI,MAAM,CAAC;IACxB,eAAe,EAAE,MAAM,CAAC;IACxB,eAAe,EAAE,MAAM,CAAC;IACxB,cAAc,EAAG,MAAM,CAAC;CACzB;AAED,MAAM,WAAW,SAAS;IACxB;;;;;;OAMG;IACH,KAAK,CAAC,WAAW,EAAE,OAAO,kBAAkB,EAAE,YAAY,EAAE,KAAK,EAAE,aAAa,GAAG,IAAI,CAAC;IACxF;;;;OAIG;IACH,MAAM,CAAC,KAAK,EAAE,aAAa,GAAG,IAAI,CAAC;IACnC;;;;;;;;OAQG;IACH,cAAc,CAAC,OAAO,EAAE,oBAAoB,GAAG,IAAI,CAAC;IACpD;;;;;;;OAOG;IACH,iBAAiB,CAAC,GAAG,EAAE,iBAAiB,EAAE,KAAK,EAAE,aAAa,GAAG,IAAI,CAAC;IACtE;;;OAGG;IACH,OAAO,CAAC,IAAI,EAAE,YAAY,GAAG,MAAM,CAAC;IACpC,mFAAmF;IACnF,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;IAC/B,QAAQ,CAAC,KAAK,EAAE,cAAc,CAAC;IAC/B,OAAO,IAAI,IAAI,CAAC;CACjB;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,MAAM,cAAc,GACtB;IACE,QAAQ,CAAC,IAAI,EAAE,YAAY,CAAC;IAC5B,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC;IAC3B,QAAQ,CAAC,WAAW,CAAC,EAAE,cAAc,CAAC;IACtC,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC;CAC5B,GACD;IACE,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC;IACvB,QAAQ,CAAC,MAAM,EAAE,gBAAgB,CAAC;IAClC,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB,wDAAwD;IACxD,QAAQ,CAAC,MAAM,EAAE,GAAG,CAAC;IACrB,oDAAoD;IACpD,QAAQ,CAAC,IAAI,EAAE,GAAG,CAAC;IACnB,mEAAmE;IACnE,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC;IAC3B;;;;;OAKG;IACH,QAAQ,CAAC,IAAI,EAAE,SAAS,CAAC;IACzB;;;OAGG;IACH,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB;;;;;OAKG;IACH,QAAQ,CAAC,OAAO,EAAE,MAAM,IAAI,CAAC;IAC7B;;;;;;;;;;OAUG;IACH,QAAQ,CAAC,UAAU,CAAC,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IACrC;;;;;OAKG;IACH,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,QAAQ,KAAK,OAAO,6BAA6B,EAAE,gBAAgB,CAAC;CAChG,CAAC;AAEN;;;;;;;;GAQG;AACH,MAAM,WAAW,qBAAqB;IACpC;;;;;;;;;;OAUG;IACH,QAAQ,CAAC,oBAAoB,CAAC,EAAE,oBAAoB,CAAC;IACrD;;;;;;;OAOG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,SAAS,CAAC;IAC/B;;;;;;;;OAQG;IACH,QAAQ,CAAC,iBAAiB,CAAC,EAAE,OAAO,CAAC;IACrC;;;;;;;;;;;OAWG;IACH,QAAQ,CAAC,qBAAqB,CAAC,EAAE,OAAO,CAAC;CAC1C;AAED,wBAAgB,cAAc,CAC5B,MAAM,EAAE,SAAS,EACjB,GAAG,EAAE,OAAO,iCAAiC,EAAE,oBAAoB,EACnE,YAAY,EAAE,SAAS,YAAY,EAAE,GAAG,IAAI,CAAC,YAAY,CAAC,EAC1D,IAAI,GAAE,qBAA0B,GAC/B,SAAS,CA+tGX"}
+{"version":3,"file":"heapScene.d.ts","sourceRoot":"","sources":["../../src/runtime/heapScene.ts"],"names":[],"mappings":"AA4CA,OAAO,EAA2B,KAAK,GAAG,EAAE,MAAM,uBAAuB,CAAC;AAC1E,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAwB,aAAa,EAAa,MAAM,2BAA2B,CAAC;AAC3F,OAAO,KAAK,EAAE,IAAI,EAAE,IAAI,EAAgD,MAAM,2BAA2B,CAAC;AAC1G,OAAO,KAAK,EAAE,MAAM,EAAkB,MAAM,yBAAyB,CAAC;AACtE,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;AAG9D,OAAO,EAEc,KAAK,oBAAoB,EAE7C,MAAM,iBAAiB,CAAC;AAMzB,OAAO,EAEL,KAAK,SAAS,EAAwB,KAAK,SAAS,EACrD,MAAM,6BAA6B,CAAC;AA+DrC;;;GAGG;AACH,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,SAAS,EAAE,YAAY,CAAC;IACjC,QAAQ,CAAC,OAAO,EAAI,YAAY,CAAC;IACjC,QAAQ,CAAC,OAAO,EAAI,WAAW,CAAC;CACjC;AAgJD,MAAM,WAAW,YAAY;IAC3B;;;OAGG;IACH,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB;;;;;;;;;;;OAWG;IACH,QAAQ,CAAC,MAAM,EAAE;QAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,OAAO,CAAA;KAAE,CAAC;IACtE;;;;;;;;;;OAUG;IACH,QAAQ,CAAC,kBAAkB,CAAC,EAAE;QAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,OAAO,CAAA;KAAE,CAAC;IACnF;;;;OAIG;IACH,QAAQ,CAAC,aAAa,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG,MAAM,CAAC;IAC/C;;;OAGG;IACH,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC,WAAW,CAAC,GAAG,WAAW,CAAC;IAClD;;;;;;OAMG;IACH,QAAQ,CAAC,QAAQ,CAAC,EAAE,cAAc,CAAC;IACnC;;;;;;;;;;;;OAYG;IACH,QAAQ,CAAC,aAAa,CAAC,EAAE,aAAa,CAAC;IACvC;;;;;;;;;;;;;;;;;;;;;;;OAuBG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE;QACnB,QAAQ,CAAC,IAAI,CAAC,EAAY,OAAO,wBAAwB,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC;QACnF,QAAQ,CAAC,SAAS,CAAC,EAAO,OAAO,wBAAwB,EAAE,eAAe,CAAC,WAAW,CAAC,CAAC;QACxF,QAAQ,CAAC,QAAQ,CAAC,EAAQ,OAAO,wBAAwB,EAAE,eAAe,CAAC,UAAU,CAAC,CAAC;QACvF,QAAQ,CAAC,YAAY,CAAC,EAAI,OAAO,wBAAwB,EAAE,eAAe,CAAC,cAAc,CAAC,CAAC;QAC3F,QAAQ,CAAC,UAAU,CAAC,EAAM,OAAO,wBAAwB,EAAE,eAAe,CAAC,YAAY,CAAC,CAAC;QACzF,QAAQ,CAAC,eAAe,CAAC,EAAE,OAAO,wBAAwB,EAAE,eAAe,CAAC,iBAAiB,CAAC,CAAC;KAChG,CAAC;CACH;AAED,MAAM,WAAW,cAAc;IAC7B,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,MAAM,CAAC;IAClB,aAAa,EAAE,MAAM,CAAC;IACtB,4DAA4D;IAC5D,aAAa,EAAI,MAAM,CAAC;IACxB,eAAe,EAAE,MAAM,CAAC;IACxB,eAAe,EAAE,MAAM,CAAC;IACxB,cAAc,EAAG,MAAM,CAAC;CACzB;AAED,MAAM,WAAW,SAAS;IACxB;;;;;;OAMG;IACH,KAAK,CAAC,WAAW,EAAE,OAAO,kBAAkB,EAAE,YAAY,EAAE,KAAK,EAAE,aAAa,GAAG,IAAI,CAAC;IACxF;;;;OAIG;IACH,MAAM,CAAC,KAAK,EAAE,aAAa,GAAG,IAAI,CAAC;IACnC;;;;;;;;OAQG;IACH,cAAc,CAAC,OAAO,EAAE,oBAAoB,GAAG,IAAI,CAAC;IACpD;;;;;;;OAOG;IACH,iBAAiB,CAAC,GAAG,EAAE,iBAAiB,EAAE,KAAK,EAAE,aAAa,GAAG,IAAI,CAAC;IACtE;;;OAGG;IACH,OAAO,CAAC,IAAI,EAAE,YAAY,GAAG,MAAM,CAAC;IACpC,mFAAmF;IACnF,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;IAC/B,QAAQ,CAAC,KAAK,EAAE,cAAc,CAAC;IAC/B,OAAO,IAAI,IAAI,CAAC;CACjB;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,MAAM,cAAc,GACtB;IACE,QAAQ,CAAC,IAAI,EAAE,YAAY,CAAC;IAC5B,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC;IAC3B,QAAQ,CAAC,WAAW,CAAC,EAAE,cAAc,CAAC;IACtC,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC;CAC5B,GACD;IACE,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC;IACvB,QAAQ,CAAC,MAAM,EAAE,gBAAgB,CAAC;IAClC,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB,wDAAwD;IACxD,QAAQ,CAAC,MAAM,EAAE,GAAG,CAAC;IACrB,oDAAoD;IACpD,QAAQ,CAAC,IAAI,EAAE,GAAG,CAAC;IACnB,mEAAmE;IACnE,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC;IAC3B;;;;;OAKG;IACH,QAAQ,CAAC,IAAI,EAAE,SAAS,CAAC;IACzB;;;OAGG;IACH,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB;;;;;OAKG;IACH,QAAQ,CAAC,OAAO,EAAE,MAAM,IAAI,CAAC;IAC7B;;;;;;;;;;OAUG;IACH,QAAQ,CAAC,UAAU,CAAC,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IACrC;;;;;OAKG;IACH,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,QAAQ,KAAK,OAAO,6BAA6B,EAAE,gBAAgB,CAAC;CAChG,CAAC;AAEN;;;;;;;;GAQG;AACH,MAAM,WAAW,qBAAqB;IACpC;;;;;;;;;;OAUG;IACH,QAAQ,CAAC,oBAAoB,CAAC,EAAE,oBAAoB,CAAC;IACrD;;;;;;;OAOG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,SAAS,CAAC;IAC/B;;;;;;;;OAQG;IACH,QAAQ,CAAC,iBAAiB,CAAC,EAAE,OAAO,CAAC;IACrC;;;;;;;;;;;OAWG;IACH,QAAQ,CAAC,qBAAqB,CAAC,EAAE,OAAO,CAAC;CAC1C;AAED,wBAAgB,cAAc,CAC5B,MAAM,EAAE,SAAS,EACjB,GAAG,EAAE,OAAO,iCAAiC,EAAE,oBAAoB,EACnE,YAAY,EAAE,SAAS,YAAY,EAAE,GAAG,IAAI,CAAC,YAAY,CAAC,EAC1D,IAAI,GAAE,qBAA0B,GAC/B,SAAS,CAyyGX"}

package/dist/runtime/heapScene.js

@@ -55,6 +55,8 @@ import { GrowBuffer, MIN_BUFFER_BYTES, POW2, ALIGN16, } from "./heapScene/growBu
 import { UniformPool, IndexPool, DrawHeap, AttributeArena, IndexAllocator, insertSortedFreeBlock, buildArenaState, arenaBytes, writeAttribute, asAval, isBufferView, asFloat32, ALLOC_HEADER_BYTES, ALLOC_HEADER_PAD_TO, ENC_V3F_TIGHT, SEM_POSITIONS, SEM_NORMALS, } from "./heapScene/pools.js";
 import { encodeModeKey } from "./pipelineCache/index.js";
 import { snapshotDescriptor, ModeKeyTracker } from "./derivedModes/modeKeyCpu.js";
+import { GpuDerivedModesScene } from "./derivedModes/gpuDispatcher.js";
+import { U32_TO_CULL } from "./derivedModes/gpuKernel.js";
 import { addMarkingCallback } from "@aardworx/wombat.adaptive";
 export function buildHeapScene(device, sig, initialDraws, opts = {}) {
     const atlasPool = opts.atlasPool;
@@ -335,6 +337,13 @@ export function buildHeapScene(device, sig, initialDraws, opts = {}) {
     /** Per-RO §7 registration handles, keyed by global drawId.
      *  Drained on removeDraw to release slots + records. */
     const derivedByDrawId = new Map();
+    // ─── GPU-eval mode rules (Task 2 Phase 5, narrow) ─────────────────
+    // Lazily allocated on first registration; one dispatcher per scene.
+    let gpuModesScene;
+    /** Last per-RO GPU output we routed to a bucket. Diff against
+     *  newly-readback values to decide rebuckets without re-bucketing
+     *  un-changed ROs. */
+    const gpuModesLastKey = new Map();
     /** The derived-uniform rule for a uniform binding on this RO, if any: an explicit
      *  `derivedUniform(...)` value in `spec.inputs`, or a standard trafo recipe by name
      *  (recipes apply only when their base-trafo leaves are actually bound on this RO —
@@ -1410,6 +1419,21 @@ export function buildHeapScene(device, sig, initialDraws, opts = {}) {
         });
         drawIdToModeTracker[drawId] = tracker;
         tracker.forEachLeaf((av) => subscribeModeLeaf(av, drawId));
+        // ─── GPU-eval mode rule registration (Task 2 Phase 5) ────────────
+        // If spec.modeRules.cull is GPU-flavoured, register the RO with
+        // the scene-wide dispatcher. Provide the arena byte offset of the
+        // RO's input uniform so the kernel reads from the right slot.
+        {
+            const gpuRule = spec.modeRules?.cull?.gpu;
+            if (gpuRule !== undefined && gpuRule.kernel === "flipCullByDeterminant") {
+                const inputRef = perDrawRefs.get(gpuRule.inputUniform);
+                if (inputRef !== undefined) {
+                    if (gpuModesScene === undefined)
+                        gpuModesScene = new GpuDerivedModesScene(device);
+                    gpuModesScene.registerRo(drawId, inputRef, gpuRule.declared);
+                }
+            }
+        }
         // ─── §7 derived-uniforms registration ────────────────────────────
         // A uniform binding on this RO is either a value (aval/constant) or a rule —
         // collect the rules (explicit `derivedUniform(...)` values + standard trafo
@@ -1578,6 +1602,10 @@ export function buildHeapScene(device, sig, initialDraws, opts = {}) {
         }
         drawIdToModeTracker[drawId] = undefined;
         dirtyModeKeyDrawIds.delete(drawId);
+        if (gpuModesScene !== undefined) {
+            gpuModesScene.deregisterRo(drawId);
+            gpuModesLastKey.delete(drawId);
+        }
         stats.totalDraws--;
         stats.geometryBytes = arenaBytes(arena);
         // GC the bucket if it just emptied (common after reactive
@@ -1912,6 +1940,55 @@ export function buildHeapScene(device, sig, initialDraws, opts = {}) {
      * we just swap the bind group + dispatch shape per bucket.
      */
     function encodeComputePrep(enc, token) {
+        // GPU-eval derived-mode rules (Task 2 Phase 5): dispatch the
+        // hardcoded determinant-flip-cull kernel + start readback.
+        // Resolved on the next frame's update() — one-frame lag for the
+        // first appearance of a state transition. CPU fallback runs at
+        // addDraw time so the initial bucket assignment is still correct.
+        if (gpuModesScene !== undefined && gpuModesScene.registered > 0) {
+            const numROs = nextDrawId; // upper bound; sparse entries are skipped via 0xFFFFFFFF sentinel
+            gpuModesScene.dispatch(arena.attrs.buffer, numROs, enc);
+            // mapAsync the staging buffer in the background. When it
+            // resolves, diff against gpuModesLastKey to schedule rebuckets.
+            gpuModesScene.finish(numROs).then(() => {
+                const out = gpuModesScene.lastOutput;
+                for (let i = 0; i < numROs; i++) {
+                    const cur = out[i] ?? 0xFFFFFFFF;
+                    const prev = gpuModesLastKey.get(i);
+                    if (cur === 0xFFFFFFFF)
+                        continue; // no rule on this RO
+                    if (prev === cur)
+                        continue;
+                    gpuModesLastKey.set(i, cur);
+                    // Project the kernel's u32 enum back to a CullMode value,
+                    // store on the tracker's modeRules so the next snapshot
+                    // picks it up, then schedule a rebucket.
+                    const tracker = drawIdToModeTracker[i];
+                    const spec = drawIdToSpec[i];
+                    if (tracker === undefined || spec === undefined)
+                        continue;
+                    const cull = U32_TO_CULL[cur];
+                    if (cull === undefined)
+                        continue;
+                    // Inject the GPU result by patching the spec's CPU rule to
+                    // return a constant. Slightly hacky but contained: the next
+                    // recompute() will read this and route to the right bucket.
+                    const patched = {
+                        ...spec.modeRules,
+                        cull: {
+                            __derivedModeRule: true, axis: "cull",
+                            evaluate: () => cull,
+                        },
+                    };
+                    spec.modeRules = patched;
+                    dirtyModeKeyDrawIds.add(i);
+                }
+            }).catch((e) => {
+                // mapAsync can reject if the device was lost or the scene was
+                // disposed mid-readback. Log but don't crash the frame.
+                console.warn("gpuModesScene readback failed:", e);
+            });
+        }
         // §7: derived-uniforms dispatch first — writes into per-bucket
         // drawHeap regions before the scan reads anything. One dispatcher
         // per bucket; constituents are shared so dirty state propagates
@@ -2043,6 +2120,7 @@ export function buildHeapScene(device, sig, initialDraws, opts = {}) {
         arena.indices.destroy();
         for (const b of buckets)
             destroyBucketResources(b);
+        gpuModesScene?.dispose();
     }
     // Test-only escape hatch for inspecting megacall bucket state. Not
     // part of the public API surface — keep cast at use-site.