feat: LOD 网格分级——meshopt 解析期简化 + 按距离/级联选级

vendored meshoptimizer 1.2（submodule 环境故障改直接 vendor）。Mesh 多 LOD：各级索引连续拼进同一 index buffer（共享顶点，LodRange 区间选择），解析期 meshopt_simplify 生成链（LOD0=原始，每级目标减半，相对误差 1e-2，小网格复用上级）；都在 worker 线程，加载零额外卡顿。 SceneRenderer 按 (mesh, LOD) 聚合 batch（lookup key = 指针<<8|lod）：主视图按屏占比近似 radius/dist 选级（阈值 10/20 倍半径）；阴影级联按级联序降级（越远越简）。Draw 按 batch.lod 选索引区间。 A/B 对比（100 背包 10×10，RTX 5060 Laptop，vsync off）： - 纯背包：LOD 开 93-95 fps / 关 90-107 fps —— 无显著差异 - 加 SunTemple：LOD 开 52-56 / 关 51-55 —— 无显著差异结论：当前场景瓶颈不在顶点数（5060 顶点吞吐富余），LOD 基础设施保留（几何复杂资产/低端 GPU/更大场景生效），帧率问题转向 Tracy GPU 时间线定位真实瓶颈。坑：LodRange 记录"LOD 不足复用最后一级"时 first 不得继续累加（buffer 无重复数据，累加必越界——VUID-08798 抓的）。
2026-09-24 20:53:19 +08:00
parent 34b593f734
commit 5bdfb53566
125 changed files with 67608 additions and 29 deletions
@@ -0,0 +1,116 @@
+#ifndef __wasi__
+#error This file contains libc stubs for WASI SDK and should only be used in non-Emscripten WebAssembly builds
+#endif
+
+#include <assert.h>
+#include <stddef.h>
+#include <stdint.h>
+
+extern unsigned char __heap_base;
+static intptr_t sbrkp = intptr_t(&__heap_base);
+
+static const int WASM_PAGE_SIZE = 64 * 1024;
+
+extern "C" void* sbrk(intptr_t increment)
+{
+	intptr_t sbrko = sbrkp;
+
+	increment = (increment + 3) & ~3;
+	sbrkp += increment;
+
+	size_t heap_size = __builtin_wasm_memory_size(0) * WASM_PAGE_SIZE;
+
+	if (size_t(sbrkp) > heap_size)
+	{
+		size_t diff = (sbrkp - heap_size + WASM_PAGE_SIZE - 1) / WASM_PAGE_SIZE;
+
+		if (__builtin_wasm_memory_grow(0, diff) == size_t(-1))
+			return (void*)-1;
+	}
+
+	return (void*)sbrko;
+}
+
+extern "C" void* memcpy(void* destination, const void* source, size_t num)
+{
+	char* d = (char*)destination;
+	const char* s = (const char*)source;
+
+	if (((uintptr_t(d) | uintptr_t(s)) & 3) == 0)
+	{
+		while (num > 15)
+		{
+			((uint32_t*)d)[0] = ((uint32_t*)s)[0];
+			((uint32_t*)d)[1] = ((uint32_t*)s)[1];
+			((uint32_t*)d)[2] = ((uint32_t*)s)[2];
+			((uint32_t*)d)[3] = ((uint32_t*)s)[3];
+			d += 16;
+			s += 16;
+			num -= 16;
+		}
+
+		while (num > 3)
+		{
+			((uint32_t*)d)[0] = ((uint32_t*)s)[0];
+			d += 4;
+			s += 4;
+			num -= 4;
+		}
+	}
+
+	while (num > 0)
+	{
+		*d++ = *s++;
+		num--;
+	}
+
+	return destination;
+}
+
+extern "C" void* memset(void* ptr, int value, size_t num)
+{
+	uint32_t v32 = ~0u / 255 * uint8_t(value);
+
+	char* d = (char*)ptr;
+
+	if ((uintptr_t(d) & 3) == 0)
+	{
+		while (num > 15)
+		{
+			((uint32_t*)d)[0] = v32;
+			((uint32_t*)d)[1] = v32;
+			((uint32_t*)d)[2] = v32;
+			((uint32_t*)d)[3] = v32;
+			d += 16;
+			num -= 16;
+		}
+
+		while (num > 3)
+		{
+			((uint32_t*)d)[0] = v32;
+			d += 4;
+			num -= 4;
+		}
+	}
+
+	while (num > 0)
+	{
+		*d++ = char(value);
+		num--;
+	}
+
+	return ptr;
+}
+
+void* operator new(size_t size)
+{
+	return sbrk((size + 7) & ~7);
+}
+
+void operator delete(void* ptr) throw()
+{
+	void* brk = reinterpret_cast<void*>(sbrkp);
+	assert(ptr <= brk);
+
+	sbrk((char*)ptr - (char*)brk);
+}