#pragma kernel MainUpdateIndexBuffer16 #pragma kernel MainUpdateIndexBuffer32 #pragma kernel MainUpdateVertexBuffer #include "GeometryPoolDefs.cs.hlsl" #include "GeometryPool.hlsl" #pragma multi_compile UV0_DIM2 UV0_DIM3 UV0_DIM4 #pragma multi_compile UV1_DIM2 UV1_DIM3 UV1_DIM4 //#pragma enable_d3d11_debug_symbols #define GROUP_SIZE_X 256 #define GROUP_SIZE_X_HALF (GROUP_SIZE_X >> 1) int _DispatchIndexOffset; int _InputIBBaseOffset; int _InputIBCount; int _InputFirstVertex; int _OutputIBOffset; ByteAddressBuffer _InputIndexBuffer; RWStructuredBuffer _OutputIndexBuffer; [numthreads(GROUP_SIZE_X,1,1)] void MainUpdateIndexBuffer32(uint3 dispatchThreadID : SV_DispatchThreadID, int3 groupID : SV_GroupID) { uint bufferOffset = _DispatchIndexOffset + dispatchThreadID.x; if (bufferOffset >= (uint)_InputIBCount) return; uint indexVal = _InputIndexBuffer.Load((_InputIBBaseOffset + bufferOffset) << 2u); _OutputIndexBuffer[(uint)_OutputIBOffset + bufferOffset] = indexVal - _InputFirstVertex; } groupshared uint _ldsIndexCache[GROUP_SIZE_X_HALF + 1]; [numthreads(GROUP_SIZE_X,1,1)] void MainUpdateIndexBuffer16(uint3 dispatchThreadID : SV_DispatchThreadID, uint groupIndex : SV_GroupIndex, uint3 groupID : SV_GroupID) { if (groupIndex < (GROUP_SIZE_X_HALF + 1)) _ldsIndexCache[groupIndex] = _InputIndexBuffer.Load((GROUP_SIZE_X_HALF * (groupID.x) + groupIndex + (uint)(_InputIBBaseOffset + _DispatchIndexOffset) / 2) << 2u); GroupMemoryBarrierWithGroupSync(); uint bufferOffset = _DispatchIndexOffset + dispatchThreadID.x; if (bufferOffset >= (uint)_InputIBCount) return; uint localOffset = groupIndex + ((_InputIBBaseOffset + _DispatchIndexOffset) & 0x1); uint pair = _ldsIndexCache[localOffset >> 1u]; uint value = (localOffset & 0x1) ? (pair >> 16) : (pair & 0xffff); _OutputIndexBuffer[(uint)_OutputIBOffset + bufferOffset] = value - _InputFirstVertex; } int _InputVBCount; int _InputBaseVertexOffset; int _DispatchVertexOffset; int _OutputVBSize; int _OutputVBOffset; int _InputPosBufferStride; int _InputPosBufferOffset; int _InputUv0BufferStride; int _InputUv0BufferOffset; int _InputUv1BufferStride; int _InputUv1BufferOffset; int _InputNormalBufferStride; int _InputNormalBufferOffset; ByteAddressBuffer _PosBuffer; ByteAddressBuffer _Uv0Buffer; ByteAddressBuffer _Uv1Buffer; ByteAddressBuffer _NormalBuffer; RWStructuredBuffer _OutputVB; int _AttributesMask; [numthreads(GROUP_SIZE_X, 1, 1)] void MainUpdateVertexBuffer(uint3 dispatchThreadID : SV_DispatchThreadID) { int inputVertexOffset = _DispatchVertexOffset + (int)dispatchThreadID.x; if ((int)inputVertexOffset >= _InputVBCount) return; inputVertexOffset += _InputBaseVertexOffset; GeoPoolVertex vtx = (GeoPoolVertex)0; if (_AttributesMask & GEOPOOLVERTEXATTRIBS_POSITION) vtx.pos = asfloat(_PosBuffer.Load3(_InputPosBufferOffset + (inputVertexOffset * _InputPosBufferStride))); if (_AttributesMask & GEOPOOLVERTEXATTRIBS_UV0) { #if UV0_DIM2 vtx.uv0.xy = asfloat(_Uv0Buffer.Load2(_InputUv0BufferOffset + (inputVertexOffset * _InputUv0BufferStride))); #elif UV0_DIM3 vtx.uv0.xyz = asfloat(_Uv0Buffer.Load3(_InputUv0BufferOffset + (inputVertexOffset * _InputUv0BufferStride))); #elif UV0_DIM4 vtx.uv0 = asfloat(_Uv0Buffer.Load4(_InputUv0BufferOffset + (inputVertexOffset * _InputUv0BufferStride))); #endif } if (_AttributesMask & GEOPOOLVERTEXATTRIBS_UV1) { #if UV1_DIM2 vtx.uv1.xy = asfloat(_Uv1Buffer.Load2(_InputUv1BufferOffset + (inputVertexOffset * _InputUv1BufferStride))); #elif UV1_DIM3 vtx.uv1.xyz = asfloat(_Uv1Buffer.Load3(_InputUv1BufferOffset + (inputVertexOffset * _InputUv1BufferStride))); #elif UV1_DIM4 vtx.uv1 = asfloat(_Uv1Buffer.Load4(_InputUv1BufferOffset + (inputVertexOffset * _InputUv1BufferStride))); #endif } if (_AttributesMask & GEOPOOLVERTEXATTRIBS_NORMAL) vtx.N = asfloat(_NormalBuffer.Load3(_InputNormalBufferOffset + (inputVertexOffset * _InputNormalBufferStride))); uint outputVertexOffset = _DispatchVertexOffset + (int)dispatchThreadID.x; GeometryPool::StoreVertex(_OutputVBOffset + outputVertexOffset, vtx, _OutputVBSize, _OutputVB); }