网站开发中的开版什么意思,工程建设监理网站,国内做网站的大公司有哪些,如何免费建网站博文介绍了最基本的实现原理#xff0c;有些老板懒得折腾#xff0c;所以特意熬了几个秃头的夜把RVO、BRG、GPU动画、海量物体目标搜索等高度封装成了开箱即用的插件。 划重点#xff01;#xff01;此方案是绕开Entities(ECS)#xff0c;不用写一行ECS代码#xff0c;现… 博文介绍了最基本的实现原理有些老板懒得折腾所以特意熬了几个秃头的夜把RVO、BRG、GPU动画、海量物体目标搜索等高度封装成了开箱即用的插件。 划重点此方案是绕开Entities(ECS)不用写一行ECS代码现有MonoBehavior开发工作流享受Entities渲染的性能。已有项目也能使用此方案开挂无需代码重构。Wechat: SunSparkStudio
性能压力测试包自取 PC 10万动画人同屏对抗https://pan.baidu.com/s/1CgpTV0TuFagobtAf7k38OA?pwdxwsf 安卓 1万动画人同屏避障 https://pan.baidu.com/s/1RkXyVXt_he5uCgizTSA42A?pwdk0ji
插件使用视频教程 Unity低成本性能开挂 gpu动画dots graphics万人同屏渲染 jobs rvo2 弹幕游戏海量单位索敌 Unity万人同屏海量物体合批渲染 Unity万人同屏海量物体目标搜索 海量单位高性能索敌攻击方案压测 移动端测试AOT和HybridCLR热更性能对比 博文开发测试环境
UnityUnity 2022.3.10f1URP 14.0.8Burst 1.8.8Jobs 0.70.0-preview.7热更HybridCLR 4.0.6PCWin11CPU i7-13700KFGPU 3070 8GRAM 32G;移动端Android骁龙8 gen2RAM 12G;
上篇博文通过最基本的自定义BRG(Batch Renderer Group) RVO避障实现了10万人同屏动态避障【Unity】十万人同屏寻路? 基于Dots技术的多线程RVO2避障_TopGames的博客-CSDN博客 上篇博文的BRG功能并不完善不支持多种Mesh和Material渲染没有拆分渲染批次(Batch)没有裁剪(Culling)处理(即不会剔除相机视口外的物体)。并且没有根据渲染数量拆分多个Draw Command。
BRG代码写起来并不容易此博文基于Unity中国DOTS技术主管开源的一个BRG示例工程修改这里仅描述主要的实现原理
参考文章Unity Open Day 北京站-技术专场深入理解 Entities Gr - 技术专栏 - Unity官方开发者社区
BRG示例工程https://github.com/vinsli/batch-rendererhttps://github.com/vinsli/batch-renderer 另外也可以看看Unity官方BRG测试案例官方案例是把当前已经创建的MeshRenderer禁用然后用BRG接管渲染
https://github.com/Unity-Technologies/Graphics/tree/master/Tests/SRPTests/Packages/com.unity.testing.brg 实现目标
为了方便使用BRG功能需要封装一个BatchRenderComponent脚本。不使用ECS仅使用传统创建GameObject的方式与BRG无缝衔接也就是对于海量物体不使用Unity Renderer组件用封装后的BRG无感知接管物体的渲染最大程度上不改变传统工作流的同时大幅提升性能
其中GameObject是只有Transform组件的空物体渲染由BatchRenderComponent接管进行合批渲染。由于Transform只能在主线程使用当数量级庞大时每帧修改Transform位置/旋转会导致掉帧所以仅当这个GameObject挂载子节点(如,子节点包含特效需要跟随人物移动)时才需要开启每帧同步Transform位置这样可以大幅节省开销。
通过此方法可以完美绕开Entities(ECS)同时也绕开了Entities(ECS)对开发成本和效率影响以及ECS当前不支持从文件名加载资源、不支持热更的痛点。
最终效果
PC端5W人, AOT模式(不使用HybridCLR)开启阴影: Android端5K人AOT模式(不使用HybridCLR)开启阴影: Android端5K人, HybridCLR热更模式开启阴影 测试中一帧创建1000个物体时会有明显卡顿这是Unity实体化GameObject本身的性能问题实际项目中海量物体的创建并不要求实时性可以通过队列分散到多帧创建以解决卡顿。 一支持多Mesh/多Material 使用BRG必须开启SRP Batcher, SRP支持相同Material合批。因此支持多Material就需要根据不同Material拆分Batch针对不同Material使用多个Batch渲染。
每个物体需要向GPU上传以下数据
两个3x4矩阵决定物体渲染的位置/旋转/缩放_BaseColor物体混合颜色_ClipId, GPU动画id, 用于切换动画;
int objectToWorldID Shader.PropertyToID(unity_ObjectToWorld);
int worldToObjectID Shader.PropertyToID(unity_WorldToObject);
int colorID Shader.PropertyToID(_BaseColor);
int gpuAnimClipId Shader.PropertyToID(_ClipId);
如果Shader还需要动态修改其它参数需要自行扩展根据参数所占内存还需要重新组织内存分配
注意必须在Shader Graph中把参数类型设置为Hybrid Per Installed否则无法正常将数据传递给shader 将每个物体依赖的数据组织成一个struct便于管理RendererNode由于要在Jobs中使用所以必须为struct类型
using Unity.Mathematics;
using static Unity.Mathematics.math;
using Unity.Burst;[BurstCompile]
public struct RendererNode
{public RendererNodeId Id { get; private set; }public bool Enable{get{return active visible;}}/// summary/// 是否启用/// /summarypublic bool active;/// summary/// 是否在视口内/// /summarypublic bool visible;/// summary/// 位置/// /summarypublic float3 position;/// summary/// 旋转/// /summarypublic quaternion rotation;/// summary/// 缩放/// /summarypublic float3 localScale;/// summary/// 顶点颜色/// /summarypublic float4 color;/// summary/// 动画id/// /summarypublic float4 animClipId;/// summary/// Mesh的原始AABB(无缩放)/// /summarypublic AABB unscaleAABB;/// summary/// 受缩放影响的AABB/// /summarypublic AABB aabb{get{//var result unscaleAABB;//result.Extents * localScale;return unscaleAABB;}}public bool IsEmpty{get{return unscaleAABB.Size.Equals(Unity.Mathematics.float3.zero);}}public static readonly RendererNode Empty new RendererNode();public RendererNode(RendererNodeId id, float3 position, quaternion rotation, float3 localScale, AABB meshAABB){this.Id id;this.position position;this.rotation rotation;this.localScale localScale;this.unscaleAABB meshAABB;this.color float4(1);this.active false;this.visible true;this.animClipId 0;}/// summary/// 构建矩阵/// /summary/// returns/returns[BurstCompile]public float4x4 BuildMatrix(){return Unity.Mathematics.float4x4.TRS(position, rotation, localScale);}
}初始化渲染数据Buffer列表
为了维护数据简单并避免物体数量变化后频繁重新创建列表所以根据RendererResource的Capacity大小维护一个固定长度的列表。并且根据不同的RendererResource拆分多个渲染批次
private void CreateRendererDataCaches(){m_BatchesVisibleCount.Clear();int index 0;foreach (var rendererRes in m_RendererResources){var drawKey rendererRes.Key;m_BatchesVisibleCount.Add(drawKey, 0);NativeListint perBatchNodes;if (!m_DrawBatchesNodeIndexes.ContainsKey(drawKey)){perBatchNodes new NativeListint(2048, Allocator.Persistent);m_DrawBatchesNodeIndexes.Add(drawKey, perBatchNodes);NativeQueueBatchDrawCommand batchDrawCommands new NativeQueueBatchDrawCommand(Allocator.Persistent);m_BatchDrawCommandsPerDrawKey.Add(drawKey, batchDrawCommands);}else{perBatchNodes m_DrawBatchesNodeIndexes[drawKey];}for (int i 0; i rendererRes.capacity; i){var color SpawnUtilities.ComputeColor(i, rendererRes.capacity);var aabb rendererRes.mesh.bounds.ToAABB();var node new RendererNode(new RendererNodeId(drawKey, index), Unity.Mathematics.float3.zero, Unity.Mathematics.quaternion.identity, float3(1), aabb);node.color color;perBatchNodes.Add(index);m_AllRendererNodes[index] node;}}}
组织拆分后每个Batch的数据
由于不同硬件性能不同单个Draw Command数量是有上限的所以还需要根据渲染数量拆分至多个BatchDrawCommand。这里主要是对内存的直接操作需要正确计算内存偏移否则会导致程序崩溃。
private void GenerateBatches(){
#if UNITY_ANDROID || UNITY_IOSint kBRGBufferMaxWindowSize 16 * 256 * 256;
#elseint kBRGBufferMaxWindowSize 16 * 1024 * 1024;
#endifconst int kItemSize (2 * 3 2); //每个物体2个3*4矩阵,1个颜色值,1个动画id,内存大小共8个float4m_MaxItemPerBatch ((kBRGBufferMaxWindowSize / kSizeOfFloat4) - 4) / kItemSize; // -4 float4 for 64 first 0 bytes ( BRG contrainst )// if (_maxItemPerBatch instanceCount)// _maxItemPerBatch instanceCount;foreach (var drawKey in m_DrawBatchesNodeIndexes.GetKeyArray(Allocator.Temp)){if (!m_BatchesPerDrawKey.ContainsKey(drawKey)){m_BatchesPerDrawKey.Add(drawKey, new NativeListint(128, Allocator.Persistent));}var instanceCountPerDrawKey m_DrawBatchesNodeIndexes[drawKey].Length;m_WorldToObjectPerDrawKey.Add(drawKey, new NativeArrayfloat4(instanceCountPerDrawKey * 3, Allocator.Persistent));m_ObjectToWorldPerDrawKey.Add(drawKey, new NativeArrayfloat4(instanceCountPerDrawKey * 3, Allocator.Persistent));var maxItemPerDrawKeyBatch m_MaxItemPerBatch instanceCountPerDrawKey ? instanceCountPerDrawKey : m_MaxItemPerBatch;//gather batch count per drawkeyint batchAlignedSizeInFloat4 BufferSizeForInstances(kBytesPerInstance, maxItemPerDrawKeyBatch, kSizeOfFloat4, 4 * kSizeOfFloat4) / kSizeOfFloat4;var batchCountPerDrawKey (instanceCountPerDrawKey maxItemPerDrawKeyBatch - 1) / maxItemPerDrawKeyBatch;//create instance data buffervar instanceDataCountInFloat4 batchCountPerDrawKey * batchAlignedSizeInFloat4;var instanceData new GraphicsBuffer(GraphicsBuffer.Target.Raw, GraphicsBuffer.UsageFlags.LockBufferForWrite, instanceDataCountInFloat4, kSizeOfFloat4);m_InstanceDataPerDrawKey.Add(drawKey, instanceData);//generate srp batchesint left instanceCountPerDrawKey;for (int i 0; i batchCountPerDrawKey; i){int instanceOffset i * maxItemPerDrawKeyBatch;int gpuOffsetInFloat4 i * batchAlignedSizeInFloat4;var batchInstanceCount left maxItemPerDrawKeyBatch ? maxItemPerDrawKeyBatch : left;var drawBatch new SrpBatch{DrawKey drawKey,GraphicsBufferOffsetInFloat4 gpuOffsetInFloat4,InstanceOffset instanceOffset,InstanceCount batchInstanceCount};m_BatchesPerDrawKey[drawKey].Add(m_DrawBatches.Length);m_DrawBatches.Add(drawBatch);left - batchInstanceCount;}}int objectToWorldID Shader.PropertyToID(unity_ObjectToWorld);int worldToObjectID Shader.PropertyToID(unity_WorldToObject);int colorID Shader.PropertyToID(_BaseColor);var batchMetadata new NativeArrayMetadataValue(4, Allocator.Temp, NativeArrayOptions.UninitializedMemory);for (int i 0; i m_DrawBatches.Length; i){var drawBatch m_DrawBatches[i];var instanceData m_InstanceDataPerDrawKey[drawBatch.DrawKey];var baseOffset drawBatch.GraphicsBufferOffsetInFloat4 * kSizeOfFloat4;int gpuAddressOffset baseOffset 64;batchMetadata[0] CreateMetadataValue(objectToWorldID, gpuAddressOffset, true); // matricesgpuAddressOffset kSizeOfPackedMatrix * drawBatch.InstanceCount;batchMetadata[1] CreateMetadataValue(worldToObjectID, gpuAddressOffset, true); // inverse matricesgpuAddressOffset kSizeOfPackedMatrix * drawBatch.InstanceCount;batchMetadata[2] CreateMetadataValue(colorID, gpuAddressOffset, true); // colorsif (BatchRendererGroup.BufferTarget BatchBufferTarget.ConstantBuffer){drawBatch.BatchID m_BRG.AddBatch(batchMetadata, instanceData.bufferHandle, (uint)BatchRendererGroup.GetConstantBufferOffsetAlignment(), (uint)BatchRendererGroup.GetConstantBufferMaxWindowSize());}else{drawBatch.BatchID m_BRG.AddBatch(batchMetadata, instanceData.bufferHandle);}m_DrawBatches[i] drawBatch;}}
二Culling剔除视口外物体渲染
使用Jobs判定物体AABB包围盒是否在相机视口内把视口外RendererNode的visible标记为false
[BurstCompile]private unsafe struct CullingJob : IJobParallelFor{[ReadOnly]public NativeArrayPlane CullingPlanes;[DeallocateOnJobCompletion][ReadOnly]public NativeArraySrpBatch Batches;[ReadOnly]public NativeArrayRendererNode Nodes;[ReadOnly]public NativeListint NodesIndexes;[ReadOnly][NativeDisableContainerSafetyRestriction]public NativeArrayfloat4 ObjectToWorldMatrices;[ReadOnly]public int DrawKeyOffset;[WriteOnly][NativeDisableUnsafePtrRestriction]public int* VisibleInstances;[WriteOnly]public NativeQueueBatchDrawCommand.ParallelWriter DrawCommands;public void Execute(int index){var batchesPtr (SrpBatch*)Batches.GetUnsafeReadOnlyPtr();var objectToWorldMatricesPtr (float4*)ObjectToWorldMatrices.GetUnsafeReadOnlyPtr();ref var srpBatch ref batchesPtr[index];int visibleCount 0;int batchOffset DrawKeyOffset srpBatch.InstanceOffset;int idx 0;for (int instanceIdx 0; instanceIdx srpBatch.InstanceCount; instanceIdx){idx srpBatch.InstanceOffset instanceIdx;int nodeIndex NodesIndexes[idx];var node Nodes[nodeIndex];if (!node.active) continue;//Assume only have 1 culling split and have 6 culling planesvar matrixIdx idx * 3;var worldAABB Transform(ref objectToWorldMatricesPtr[matrixIdx], ref objectToWorldMatricesPtr[matrixIdx 1], ref objectToWorldMatricesPtr[matrixIdx 2], node.aabb);if (!(node.visible (Intersect(CullingPlanes, ref worldAABB) ! FrustumPlanes.IntersectResult.Out)))continue;VisibleInstances[batchOffset visibleCount] instanceIdx;visibleCount;}if (visibleCount 0){var drawKey srpBatch.DrawKey;DrawCommands.Enqueue(new BatchDrawCommand{visibleOffset (uint)batchOffset,visibleCount (uint)visibleCount,batchID srpBatch.BatchID,materialID drawKey.MaterialID,meshID drawKey.MeshID,submeshIndex (ushort)drawKey.SubmeshIndex,splitVisibilityMask 0xff,flags BatchDrawCommandFlags.None,sortingPosition 0});}}} 三添加/移除渲染物体
添加Renderer实际上就是从RendererNode列表中找出空闲的RendererNode用来存放数据。从上万长度的数组中找出空闲索引也是比较消耗性能的所以这里使用Jobs查找每次查找出N个空闲索引存入队列待使用直到用完后再进行下次查找。
移除Renderer就是把RendererNode的active设置为false置为空闲状态
public RendererNodeId AddRenderer(int resourceIdx, float3 pos, quaternion rot, float3 scale){if (resourceIdx 0 || resourceIdx m_RendererResources.Count){return RendererNodeId.Null;}var rendererRes m_RendererResources[resourceIdx];var inactiveList m_InactiveRendererNodes[rendererRes.Key];if (inactiveList.Count 1){var nodesIndexes m_DrawBatchesNodeIndexes[rendererRes.Key];var jobs new GetInactiveRendererNodeJob{Nodes m_AllRendererNodes.AsReadOnly(),NodesIndexes nodesIndexes,RequireCount 2048,Outputs inactiveList};jobs.Schedule().Complete();}if (!inactiveList.TryDequeue(out int inactiveNodeIndex)){Log.Warning(添加Renderer失败, Inactive renderer node is null);return RendererNodeId.Null;}var renderer m_AllRendererNodes[inactiveNodeIndex];renderer.position pos;renderer.rotation rot;renderer.localScale scale;renderer.active true;//renderer.color float4(1);renderer.visible true;renderer.animClipId 0;m_AllRendererNodes[inactiveNodeIndex] renderer;m_BatchesVisibleCount[rendererRes.Key];m_TotalVisibleCount;return renderer.Id;}
四同步RVO位置数据到RendererNode: 由于已经把所有RendererNode组织到了一个NativeArray里所以可以非常容易得使用Jobs批量同步渲染位置、旋转等信息 /// summary/// 通过JobSystem更新渲染数据/// /summary/// param nameagents/paraminternal void SetRendererData(NativeArrayAgentData agentsData){var job new SyncRendererNodeTransformJob{AgentDataArr agentsData,Nodes m_AllRendererNodes};job.Schedule(agentsData.Length, 64).Complete();}[BurstCompile]private struct SyncRendererNodeTransformJob : IJobParallelFor{[ReadOnly] public NativeArrayAgentData AgentDataArr;[NativeDisableParallelForRestriction]public NativeArrayRendererNode Nodes;public void Execute(int index){var agentDt AgentDataArr[index];var node Nodes[agentDt.rendererIndex];node.position agentDt.worldPosition;node.rotation agentDt.worldQuaternion;node.animClipId agentDt.animationIndex;Nodes[agentDt.rendererIndex] node;}}
五创建空物体并绑定RVO Agent和RendererNode: 上述BRG封装完成后就可以非常简单得创建空物体在空物体脚本中绑定RVO Agent和RendererNode其中RVO Agent用于自动寻路RendererNode相当于是空物体的渲染组件
空物体挂载脚本如下
using UnityEngine;
using UnityGameFramework.Runtime;public class RvoBRGEntity : EntityBase
{float m_MoveSpeed 5f;Transform m_FollowTarget;RVOAgent mAgent;RendererNodeId m_RenderId;protected override void OnShow(object userData){base.OnShow(userData);m_FollowTarget Params.GetVarTransform(FollowTarget);mAgent GF.RVO.AddAgent(this);//为空GameObject添加 RVO AgentmAgent.maxSpeed m_MoveSpeed;m_RenderId GF.BatchRender.AddRenderer(0, mAgent.pos, mAgent.rotation, CachedTransform.localScale); //为空GameObject添加一个BRG渲染节点mAgent.rendererIndex m_RenderId.Index;mAgent.SyncTransform false; //默认不同步当前Transform}protected override void OnUpdate(float elapseSeconds, float realElapseSeconds){base.OnUpdate(elapseSeconds, realElapseSeconds);mAgent.targetPosition m_FollowTarget.position; //更新RVO Agent目标点mAgent.animationIndex mAgent.IsMoving ? 1 : 0;}protected override void OnHide(bool isShutdown, object userData){if (!isShutdown){GF.RVO.RemoveAgent(mAgent);GF.BatchRender.RemoveRenderer(m_RenderId);}base.OnHide(isShutdown, userData);}
}六海量物体目标搜索:
海量物体就会面临如何获取范围内的敌人或获取最近的攻击目标在数万数量级面前即使是遍历所有目标进行距离判定还是非常耗时的操作大部分情况下需要每帧进行目标搜索。
推荐几个高性能jobs开源库
1. Neighbor Search Grid, 可用于高性能最近目标搜索、范围内目标搜索GitHub - nezix/NeighborSearchGridBurst: Neighbor search using a grid based approach C# job system Burst
2. KDTree Jobs版可用于高性能最近目标搜索、范围内目标搜索GitHub - ArthurBrussee/KNN: Fast K-Nearest Neighbour Library for Unity DOTS
3. QuadTree可用于做高性能碰撞检测GitHub - marijnz/NativeQuadtree: A Quadtree Native Collection for Unity DOTS 强烈推荐Neighbor Search Grid其原理是将海量单位划分到格子区域从查询点最近的格子进行目标搜索同样支持单次查询多个点。
使用方法
1. 查询最近目标
以查询最近的RVO Agent为例
/// summary/// 给定多个point,一次查询各个点最近的Agent/// /summary/// param namepoints/param/// param namenearestAgents/param/// returns/returnspublic int GetNearestAgents(Vector3[] points, RVOAgent[] nearestAgents){if (points.Length ! nearestAgents.Length || !m_orca.TryGetFirst(-1, out IAgentProvider agentProvider, true)){return 0;}m_orca.currentHandle.Complete();//确保RVO Jobs已经完成int arrayLength agentProvider.outputAgents.Length;if (arrayLength 1 || arrayLength ! m_agents.Count) return 0;NativeArrayAgentData tempAgentDatas new NativeArrayAgentData(agentProvider.outputAgents, Allocator.TempJob);var tempPositions new NativeArrayfloat3(arrayLength, Allocator.TempJob, NativeArrayOptions.UninitializedMemory);var job new GetAgentsPositionArrayJob{Input tempAgentDatas,Output tempPositions};job.Schedule(arrayLength, 64).Complete();if (m_GirdInitialized){if (m_GridSearch.PositionsLength ! job.Output.Length){m_GridSearch.clean();m_GridSearch.initGrid(job.Output);}else{m_GridSearch.updatePositions(job.Output);}}else{m_GridSearch.initGrid(job.Output);m_GirdInitialized true;}var indexes m_GridSearch.searchClosestPoint(points);int resultCount 0;for (int i 0; i indexes.Length; i){var index indexes[i];if (index -1) continue;var searchAgent m_agents[tempAgentDatas[index].index];if (searchAgent null) continue;nearestAgents[resultCount] searchAgent;}tempPositions.Dispose();tempAgentDatas.Dispose();return resultCount;}
2. 查询范围内多个目标
/// summary/// 获取给定点半径内的Agents/// /summary/// param namepoint/param/// param nameradius/param/// param nameresults/param/// returns/returnspublic int OverlapSphere(Vector3 point, float radius, RVOAgent[] results){if (!m_orca.TryGetFirst(-1, out IAgentProvider agentProvider, true)){return 0;}m_orca.currentHandle.Complete();//确保RVO Jobs已经完成int arrayLength agentProvider.outputAgents.Length;if (arrayLength 1 || arrayLength ! m_agents.Count) return 0;NativeArrayAgentData tempAgentDatas new NativeArrayAgentData(agentProvider.outputAgents, Allocator.TempJob);var tempPositions new NativeArrayfloat3(arrayLength, Allocator.TempJob, NativeArrayOptions.UninitializedMemory);var job new GetAgentsPositionArrayJob{Input tempAgentDatas,Output tempPositions};job.Schedule(arrayLength, 64).Complete();if (m_GirdInitialized){if (m_GridSearch.PositionsLength ! job.Output.Length){m_GridSearch.clean();m_GridSearch.initGrid(job.Output);}else{m_GridSearch.updatePositions(job.Output);}}else{m_GridSearch.initGrid(job.Output);m_GirdInitialized true;}m_QueryPoints[0] point;var indexes m_GridSearch.searchWithin(m_QueryPoints, radius, results.Length);int resultCount 0;for (int i 0; i indexes.Length; i){if (indexes[i] -1) continue;int index indexes[i];int agentIndex tempAgentDatas[index].index;if (agentIndex 0 agentIndex arrayLength){results[resultCount] m_agents[agentIndex];}}tempPositions.Dispose();tempAgentDatas.Dispose();return resultCount;}
总结
海量物体同屏大多数时间并非所有物体都在视口内所以BRG增加剔除功能后性能会得到大幅提升。
通过对RVO和BRG的封装可以非常简单得与传统开发方式无缝结合比使用Entities更加简单并且同样能享受到dots技术优势。
不足的是还没有为BRG实现LOD, 海量物体渲染CPU和GPU是木桶效应共同决定了帧数。JobsBurst解决了CPU瓶颈但GPU瓶颈还需要从多个方向去解决其中使用LOD对远处物体使用低顶点Mesh已降低GPU压力是效果显著的一种方式。
关于移动平台事实上BRG对移动平台无明显增益Jobs才是性能提升的关键使用Jobs计算数据和上传给GPU大大降低了GPU等待CPU的时间可惜的是移动平台下Jobs工作线程谜之少带来的增益非常有限。 由于HybridCLR不支持Burst加速HybridCLR下Jobs代码以解释方式执行所以相比AOT性能大打折扣。对于热更项目要想发挥Burst加成就需要把Jobs代码放到AOT执行了。以demo为例dots代码放到AOT后与纯AOT性能相差无几5K人全部在视口内也能流畅运行相对于移动平台在实际项目中同时出现在视口内的物体基本不会超过3K有了Culling功能后视口内外共1W物体也能流程运行。
