Dynamic Tessellation - Practical Rendering and Computation with Direct3D 11 - page 439

Figure 9.21. Computer shader phase five.
is a good metric of how coplanar the 256 individual height samples are—lower values imply
a flatter surface, and higher values imply a noisier and more varying patch. This single value
and the plane's normal vector are written out as a float4 in the output texture, so 256 height
map samples have been reduced to four numbers. Listing 9.16 shows the last phase of the
computer shader.
if(GI == 0)
{ // Let the first thread compute the standard deviation for
// this patch. The 'average' is really just going to be 0.0
// as we want the deviation from the plane and any point on the
// plane now has a 'height' of zero.
float stddev = 0.0f;
for(int i = 0; i < 16*16; ++i)
stddev += pow(groupResults[i] ,2) ;
stddev /= ((16.0f * 16.0f) - 1.0f);
stddev = sqrt(stddev);
// Then write the normal vector and standard deviation
// to the output buffer for use by the Domain and Hull Shaders
bufferResults[uint2(Gid.x, Gid.y)] = float4(plane.xyz, stddev);
}
Listing 9.16. Phase five of the compute shader.
Integrating the Compute Shader
The previous section details the actual compute shader that implements our algorithm, but
the application must still coordinate this work.