Add WaveReadFirstLane()

ScriptableRenderPipeline/Core/ShaderLibrary/API/PSSL.hlsl

 // This file assume SHADER_API_D3D11 is defined
 
 #define INTRINSIC_BITFIELD_EXTRACT
+#define INTRINSIC_WAVEREADFIRSTLANE
+#define WaveReadFirstLane ReadFirstLane
 #define Mad24 mad24
 #define INTRINSIC_MED3
 #define INTRINSIC_MINMAX3

ScriptableRenderPipeline/Core/ShaderLibrary/Common.hlsl

     return BitFieldExtract(data, 1u, bitPos) != 0;
 }
 
+#ifndef INTRINSIC_WAVEREADFIRSTLANE
+// Warning: for correctness, the value you pass to the function must be constant across the wave!
+uint WaveReadFirstLane(uint scalarValue)
+{
+    return scalarValue;
+}
+#endif
+
 #ifndef INTRINSIC_CLAMP
 // TODO: should we force all clamp to be intrinsic by default ?
 // Some platform have one instruction clamp

ScriptableRenderPipeline/HDRenderPipeline/Material/Lit/Resources/SubsurfaceScattering.compute

                           uint  groupThreadId : SV_GroupThreadID)
 {
     // Note: any factor of 64 is a suitable wave size for our algorithm.
-    uint waveIndex = groupThreadId / 64;
+    uint waveIndex = WaveReadFirstLane(groupThreadId / 64);
     uint laneIndex = groupThreadId % 64;
     uint quadIndex = laneIndex / 4;