Use a single wavefront group for the volumetric lighting shader

ScriptableRenderPipeline/HDRenderPipeline/HDRP/Lighting/Volumetrics/Resources/VolumetricLighting.compute

 
 #include "../../../ShaderConfig.cs.hlsl"
 #if (SHADEROPTIONS_VOLUMETRIC_LIGHTING_PRESET == 1)
-    // E.g. for 1080p: (1920/8)x(1080/8)x(128) =  4,147,200 voxels
+    // E.g. for 1080p: (1920/8)x(1080/8)x(64)  =  2,073,600 voxels
-    // E.g. for 1080p: (1920/4)x(1080/4)x(256) = 33,177,600 voxels
+    // E.g. for 1080p: (1920/4)x(1080/4)x(128) = 16,588,800 voxels
     #define VBUFFER_TILE_SIZE   4
     #define VBUFFER_SLICE_COUNT 128
 #endif
 
-#define GROUP_SIZE_1D 16
-#define GROUP_SIZE_2D (GROUP_SIZE_1D * GROUP_SIZE_1D)
+#define GROUP_SIZE_1D 8
 
 #if (SHADEROPTIONS_VOLUMETRIC_LIGHTING_PRESET != 0) // Switch between the full and the empty shader
 
     }
 }
 
-[numthreads(GROUP_SIZE_2D, 1, 1)]
+[numthreads(GROUP_SIZE_1D, GROUP_SIZE_1D, 1)]
-                        uint  groupThreadId : SV_GroupThreadID)
+                        uint2 groupThreadId : SV_GroupThreadID)
 {
     // Perform compile-time checks.
     if (!IsPower2(VBUFFER_TILE_SIZE) || !IsPower2(TILE_SIZE_CLUSTERED)) return;
     uint laneIndex = groupThreadId % 64;
     uint quadIndex = laneIndex / 4;
 
-    // Arrange threads in the Morton order to optimally match the memory layout of GCN tiles.
-    uint2 groupCoord  = DecodeMorton2D(groupThreadId);
+    uint2 groupCoord  = groupThreadId;
     uint2 groupOffset = groupId * GROUP_SIZE_1D;
     uint2 voxelCoord  = groupOffset + groupCoord;
     uint2 tileCoord   = voxelCoord * VBUFFER_TILE_SIZE / TILE_SIZE_CLUSTERED;
     if (voxelsPerClusterTile >= 64)
     {
         // TODO: this is a compile-time test, make sure the compiler actually scalarizes.
-        tileCoord = WaveReadFirstLane(tileCoord);
+        tileCoord = groupOffset * VBUFFER_TILE_SIZE / TILE_SIZE_CLUSTERED;
-    UNITY_BRANCH if (voxelCoord.x >= (uint)_VBufferResolution.x ||
-                 voxelCoord.y >= (uint)_VBufferResolution.y)
+    UNITY_BRANCH
+    if (voxelCoord.x >= (uint)_VBufferResolution.x ||
+        voxelCoord.y >= (uint)_VBufferResolution.y)
     {
         return;
     }

ScriptableRenderPipeline/HDRenderPipeline/HDRP/Lighting/Volumetrics/VolumetricLighting.cs

             cmd.SetComputeTextureParam(m_VolumetricLightingCS, kernel, HDShaderIDs._VBufferLightingHistory,  vBuffer.GetLightingHistoryBuffer());  // Read
             }
 
-            // The shader defines GROUP_SIZE_1D = 16.
-            cmd.DispatchCompute(m_VolumetricLightingCS, kernel, (w + 15) / 16, (h + 15) / 16, 1);
+            // The shader defines GROUP_SIZE_1D = 8.
+            cmd.DispatchCompute(m_VolumetricLightingCS, kernel, (w + 7) / 8, (h + 7) / 8, 1);
         }
     }
 } // class VolumetricLightingModule