Dynamitos/Seele
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Polymorphic Map and threadpool

Browse Source
This commit is contained in:
Dynamitos
2021-11-01 20:25:16 +01:00
parent 9c48c48f8c
commit 1b7ab9b1f1
20 changed files with 1442 additions and 1280 deletions
Download Patch File Download Diff File Expand all files Collapse all files
res/shaders/LightCulling.slang
+72 -72
View File
@@ -3,18 +3,18 @@ import Common;
struct ComputeShaderInput
{
uint3 groupID : SV_GroupID;
uint3 groupThreadID : SV_GroupThreadID;
uint3 dispatchThreadID : SV_DispatchThreadID;
uint groupIndex : SV_GroupIndex;
uint3 groupID : SV_GroupID;
uint3 groupThreadID : SV_GroupThreadID;
uint3 dispatchThreadID : SV_DispatchThreadID;
uint groupIndex : SV_GroupIndex;
};
layout(set = INDEX_VIEW_PARAMS, binding = 1)
cbuffer DispatchParams
{
uint3 numThreadGroups;
uint pad0;
uint3 numThreads;
uint pad1;
uint3 numThreadGroups;
uint pad0;
uint3 numThreads;
uint pad1;
}
layout(set = INDEX_VIEW_PARAMS, binding = 2)
@@ -52,90 +52,90 @@ groupshared uint tLightList[1024];
void oAppendLight(uint lightIndex)
{
uint index;
InterlockedAdd(oLightCount, 1, index);
if(index < 1024)
{
oLightList[index] = lightIndex;
}
uint index;
InterlockedAdd(oLightCount, 1, index);
if(index < 1024)
{
oLightList[index] = lightIndex;
}
}
void tAppendLight(uint lightIndex)
{
uint index;
InterlockedAdd(tLightCount, 1, index);
if(index < 1024)
{
tLightList[index] = lightIndex;
}
uint index;
InterlockedAdd(tLightCount, 1, index);
if(index < 1024)
{
tLightList[index] = lightIndex;
}
}
[numthreads(BLOCK_SIZE, BLOCK_SIZE, 1)]
[shader("compute")]
void cullLights(ComputeShaderInput in)
{
int3 texCoord = int3(in.dispatchThreadID.xy, 0);
float fDepth = depthTextureVS.Load(texCoord).r;
int3 texCoord = int3(in.dispatchThreadID.xy, 0);
float fDepth = depthTextureVS.Load(texCoord).r;
uint uDepth = asuint(fDepth);
if(in.groupIndex == 0)
{
uMinDepth = 0xffffffff;
uMaxDepth = 0x0;
oLightCount = 0;
tLightCount = 0;
uint uDepth = asuint(fDepth);
if(in.groupIndex == 0)
{
uMinDepth = 0xffffffff;
uMaxDepth = 0x0;
oLightCount = 0;
tLightCount = 0;
groupFrustum = frustums[in.groupID.x + (in.groupID.y * numThreadGroups.x)];
}
}
GroupMemoryBarrierWithGroupSync();
InterlockedMin(uMinDepth, uDepth);
InterlockedMax(uMaxDepth, uDepth);
GroupMemoryBarrierWithGroupSync();
InterlockedMin(uMinDepth, uDepth);
InterlockedMax(uMaxDepth, uDepth);
GroupMemoryBarrierWithGroupSync();
GroupMemoryBarrierWithGroupSync();
float fMinDepth = asfloat(uMinDepth);
float fMaxDepth = asfloat(uMaxDepth);
float fMinDepth = asfloat(uMinDepth);
float fMaxDepth = asfloat(uMaxDepth);
float minDepthVS = clipToView(float4(0, 0, fMinDepth, 1)).z;
float maxDepthVS = clipToView(float4(0, 0, fMaxDepth, 1)).z;
float nearClipVS = clipToView(float4(0, 0, 0, 1.0f)).z;
float minDepthVS = clipToView(float4(0, 0, fMinDepth, 1)).z;
float maxDepthVS = clipToView(float4(0, 0, fMaxDepth, 1)).z;
float nearClipVS = clipToView(float4(0, 0, 0, 1.0f)).z;
Plane minPlane = {float3(0, 0, -1), -minDepthVS};
Plane minPlane = {float3(0, 0, -1), -minDepthVS};
for ( uint i = in.groupIndex; i < numPointLights; i += BLOCK_SIZE * BLOCK_SIZE )
{
{
PointLight light = pointLights[i];
//TODO: why doesn't this check go through?
//if(light.insideFrustum(groupFrustum, nearClipVS, maxDepthVS))
{
tAppendLight(i);
//TODO: why doesn't this check go through?
//if(light.insideFrustum(groupFrustum, nearClipVS, maxDepthVS))
{
tAppendLight(i);
if(!light.insidePlane(minPlane))
{
oAppendLight(i);
}
}
{
oAppendLight(i);
}
}
}
GroupMemoryBarrierWithGroupSync();
if(in.groupIndex == 0)
{
InterlockedAdd(oLightIndexCounter[0], (uint)oLightCount, oLightIndexStartOffset);
oLightGrid[in.groupID.xy] = uint2(oLightIndexStartOffset, oLightCount);
InterlockedAdd(tLightIndexCounter[0], (uint)tLightCount, tLightIndexStartOffset);
tLightGrid[in.groupID.xy] = uint2(tLightIndexStartOffset, tLightCount);
}
GroupMemoryBarrierWithGroupSync();
for (uint j = in.groupIndex; j < (uint)oLightCount; j += BLOCK_SIZE * BLOCK_SIZE)
{
oLightIndexList[oLightIndexStartOffset + j] = oLightList[j];
}
// For transparent geometry.
for ( uint k = in.groupIndex; k < (uint)tLightCount; k += BLOCK_SIZE * BLOCK_SIZE )
{
tLightIndexList[tLightIndexStartOffset + k] = tLightList[k];
}
}
GroupMemoryBarrierWithGroupSync();
if(in.groupIndex == 0)
{
InterlockedAdd(oLightIndexCounter[0], (uint)oLightCount, oLightIndexStartOffset);
oLightGrid[in.groupID.xy] = uint2(oLightIndexStartOffset, oLightCount);
InterlockedAdd(tLightIndexCounter[0], (uint)tLightCount, tLightIndexStartOffset);
tLightGrid[in.groupID.xy] = uint2(tLightIndexStartOffset, tLightCount);
}
GroupMemoryBarrierWithGroupSync();
for (uint j = in.groupIndex; j < (uint)oLightCount; j += BLOCK_SIZE * BLOCK_SIZE)
{
oLightIndexList[oLightIndexStartOffset + j] = oLightList[j];
}
// For transparent geometry.
for ( uint k = in.groupIndex; k < (uint)tLightCount; k += BLOCK_SIZE * BLOCK_SIZE )
{
tLightIndexList[tLightIndexStartOffset + k] = tLightList[k];
}
}