evas - gl - optimize dither shader - use integer vecs and go to 2x2

a 2x2 matrix visually here is about as good as 4x4. at least in what i
see, but on low end gpu's it can halve the cost. in fact i was watching the
gpu on my old i5-4200u drop down to 340-410mhz (no dithering is 320-360mhz).
i got to 630-660mhz with the original 4x4 code.

the 4x4 is still there ifdefed out. perhaps i can bring it back with a
high-quality dither option, but 2x3 i think is good enough.

2 files changed, 41 insertions, 22 deletions

diff --git a/src/modules/evas/engines/gl_common/shader/evas_gl_shaders.x b/src/modules/evas/engines/gl_common/shader/evas_gl_shaders.x
index b534f66c41..411a680e78 100644
--- a/src/modules/evas/engines/gl_common/shader/evas_gl_shaders.x
+++ b/src/modules/evas/engines/gl_common/shader/evas_gl_shaders.x
@@ -99,26 +99,35 @@ static const char fragment_glsl[] =
    "#endif\n"
    "// ----------------------------------------------------------------------------\n"
    "#ifdef SHD_DITHER\n"
-   "const mat4 dm = mat4(vec4( 0,  8,  2, 10),\n"
-   "                     vec4(12,  4, 14,  6),\n"
-   "                     vec4( 3, 11,  1,  9),\n"
-   "                     vec4(15,  7, 13,  5));\n"
-   "float dither_closest(vec2 pos, float val)\n"
+   "#ifdef DM_HIQ\n"
+   "#define DMMOD 4.0\n"
+   "#define DMDIV (16.0*256.0)\n"
+   "const mat4 dm = mat4(vec4( 0.0/DMDIV,  8.0/DMDIV,  2.0/DMDIV, 10.0/DMDIV),\n"
+   "                     vec4(12.0/DMDIV,  4.0/DMDIV, 14.0/DMDIV,  6.0/DMDIV),\n"
+   "                     vec4( 3.0/DMDIV, 11.0/DMDIV,  1.0/DMDIV,  9.0/DMDIV),\n"
+   "                     vec4(15.0/DMDIV,  7.0/DMDIV, 13.0/DMDIV,  5.0/DMDIV));\n"
+   "#else\n"
+   "#define DMMOD 2.0\n"
+   "#define DMDIV (4.0*256.0)\n"
+   "const mat2 dm = mat2(vec2( 0.0/DMDIV,  2.0/DMDIV),\n"
+   "                     vec2( 3.0/DMDIV,  1.0/DMDIV));\n"
+   "#endif\n"
+   "float dither_closest(ivec2 pos, float val)\n"
    "{\n"
-   "   float limit = dm[int(pos.x)][int(pos.y)] / 16.0;\n"
+   "   float limit = dm[pos.x][pos.y];\n"
    "   if (val <= limit) return 0.0;\n"
-   "   return 1.0;\n"
+   "   return (1.0 / 256.0);\n"
    "}\n"
-   "float dither_8bit(vec2 modpos, float val)\n"
+   "float dither_8bit(ivec2 modpos, float val)\n"
    "{\n"
    "   float val_quant = float(floor(val * 255.0)) / 255.0;\n"
-   "   float val_delta = (val - val_quant) / (1.0 / 256.0);\n"
+   "   float val_delta = val - val_quant;\n"
    "   float val_roundup = dither_closest(modpos, val_delta);\n"
-   "   return val_quant + (val_roundup * (1.0 / 256.0));\n"
+   "   return val_quant + val_roundup;\n"
    "}\n"
    "vec4 dither(vec4 col, vec2 pos)\n"
    "{\n"
-   "   vec2 modpos = vec2(mod(float(pos.x), 4.0), mod(float(pos.y), 4.0));\n"
+   "   ivec2 modpos = ivec2(int(mod(pos.x, DMMOD)), int(mod(pos.y, DMMOD)));\n"
    "   return vec4(dither_8bit(modpos, col.r),\n"
    "               dither_8bit(modpos, col.g),\n"
    "               dither_8bit(modpos, col.b),\n"
diff --git a/src/modules/evas/engines/gl_common/shader/fragment.glsl b/src/modules/evas/engines/gl_common/shader/fragment.glsl
index b715c8de99..5b1f8c24bc 100644
--- a/src/modules/evas/engines/gl_common/shader/fragment.glsl
+++ b/src/modules/evas/engines/gl_common/shader/fragment.glsl
@@ -98,29 +98,39 @@ uniform float blur_div;
 // ----------------------------------------------------------------------------
 
 #ifdef SHD_DITHER
-const mat4 dm = mat4(vec4( 0,  8,  2, 10),
-                     vec4(12,  4, 14,  6),
-                     vec4( 3, 11,  1,  9),
-                     vec4(15,  7, 13,  5));
 
-float dither_closest(vec2 pos, float val)
+#ifdef DM_HIQ
+#define DMMOD 4.0
+#define DMDIV (16.0*256.0)
+const mat4 dm = mat4(vec4( 0.0/DMDIV,  8.0/DMDIV,  2.0/DMDIV, 10.0/DMDIV),
+                     vec4(12.0/DMDIV,  4.0/DMDIV, 14.0/DMDIV,  6.0/DMDIV),
+                     vec4( 3.0/DMDIV, 11.0/DMDIV,  1.0/DMDIV,  9.0/DMDIV),
+                     vec4(15.0/DMDIV,  7.0/DMDIV, 13.0/DMDIV,  5.0/DMDIV));
+#else
+#define DMMOD 2.0
+#define DMDIV (4.0*256.0)
+const mat2 dm = mat2(vec2( 0.0/DMDIV,  2.0/DMDIV),
+                     vec2( 3.0/DMDIV,  1.0/DMDIV));
+#endif
+
+float dither_closest(ivec2 pos, float val)
 {
-   float limit = dm[int(pos.x)][int(pos.y)] / 16.0;
+   float limit = dm[pos.x][pos.y];
    if (val <= limit) return 0.0;
-   return 1.0;
+   return (1.0 / 256.0);
 }
 
-float dither_8bit(vec2 modpos, float val)
+float dither_8bit(ivec2 modpos, float val)
 {
    float val_quant = float(floor(val * 255.0)) / 255.0;
-   float val_delta = (val - val_quant) / (1.0 / 256.0);
+   float val_delta = val - val_quant;
    float val_roundup = dither_closest(modpos, val_delta);
-   return val_quant + (val_roundup * (1.0 / 256.0));
+   return val_quant + val_roundup;
 }
 
 vec4 dither(vec4 col, vec2 pos)
 {
-   vec2 modpos = vec2(mod(float(pos.x), 4.0), mod(float(pos.y), 4.0));
+   ivec2 modpos = ivec2(int(mod(pos.x, DMMOD)), int(mod(pos.y, DMMOD)));
    return vec4(dither_8bit(modpos, col.r),
                dither_8bit(modpos, col.g),
                dither_8bit(modpos, col.b),