Hack around a convergence problem in the SOR code, by stopping the over-relaxation...

[nageru] / sor.frag

diff --git a/sor.frag b/sor.frag
index dc4a5b8af4fdb19910819ffb39a96ee06246163f..6c5333fa5ead7c960a96361ae16fd501638d7a6a 100644 (file)
--- a/sor.frag
+++ b/sor.frag
@@ -6,13 +6,30 @@ out vec2 diff_flow;
 uniform sampler2D diff_flow_tex, smoothness_x_tex, smoothness_y_tex;
 uniform usampler2D equation_tex;
 
+// See pack_floats_shared() in equations.frag.
+vec2 unpack_floats_shared(uint c)
+{
+       // Recover the exponent, and multiply it in. Add one because
+       // we have denormalized mantissas, then another one because we
+       // already reduced the exponent by one. Then subtract 20, because
+       // we are going to shift up the number by 20 below to recover the sign bits.
+       float normalizer = uintBitsToFloat(((c >> 1) & 0x7f800000u) - (18 << 23));
+       normalizer *= (1.0 / 2047.0);
+
+       // Shift the values up so that we recover the sign bit, then normalize.
+       float a = int(uint(c & 0x000fffu) << 20) * normalizer;
+       float b = int(uint(c & 0xfff000u) << 8) * normalizer;
+
+       return vec2(a, b);
+}
+
 void main()
 {
        uvec4 equation = texture(equation_tex, tc);
        float inv_A11 = uintBitsToFloat(equation.x);
        float A12 = uintBitsToFloat(equation.y);
        float inv_A22 = uintBitsToFloat(equation.z);
-       vec2 b = unpackHalf2x16(equation.w);
+       vec2 b = unpack_floats_shared(equation.w);
 
        // Subtract the missing terms from the right-hand side
        // (it couldn't be done earlier, because we didn't know
@@ -28,7 +45,10 @@ void main()
        b += smooth_d * textureOffset(diff_flow_tex, tc, ivec2( 0, -1)).xy;
        b += smooth_u * textureOffset(diff_flow_tex, tc, ivec2( 0,  1)).xy;
 
-       const float omega = 1.6;
+       // FIXME: omega=1.6 seems to make our entire system diverge.
+       // Is this because we do Gauss-Seidel instead of Jacobi?
+       // Figure out what's going on.
+       const float omega = 1.0;
        diff_flow = texture(diff_flow_tex, tc).xy;
 
        // From https://en.wikipedia.org/wiki/Successive_over-relaxation.