shark: Shader cleanup

diff --git a/title/shark-shaders/shaders/display_transform.comp.glsl b/title/shark-shaders/shaders/display_transform.comp.glsl
index be48b9171404d4ee15d62a3dc050344470ed2b22..113d314b4726e7dbdbdc57400b5a99c436e365c9 100644 (file)
--- a/title/shark-shaders/shaders/display_transform.comp.glsl
+++ b/title/shark-shaders/shaders/display_transform.comp.glsl
@@ -27,18 +27,17 @@ vec3 tony_mc_mapface(vec3 stimulus) {
 layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
 
 void main() {
-    const vec3 stimulus = imageLoad(color_layer, ivec2(gl_GlobalInvocationID.xy)).rgb;
-    const vec3 transformed = tony_mc_mapface(stimulus);
-    vec3 composited = srgb_oetf(transformed);
-
     const uvec2 tile_coord = gl_WorkGroupID.xy / 4;
     const uint tile_index = tile_coord.y * uniforms.tile_stride + tile_coord.x;
     const uint tile_base = tile_index * TILE_STRIDE;
-
     TilesRead tiles_read = TilesRead(uniforms.tiles);
-
     const uint lo = tiles_read.values[tile_base + TILE_BITMAP_RANGE_LO_OFFSET];
     const uint hi = tiles_read.values[tile_base + TILE_BITMAP_RANGE_HI_OFFSET];
+
+    const vec3 stimulus = imageLoad(color_layer, ivec2(gl_GlobalInvocationID.xy)).rgb;
+    const vec3 transformed = tony_mc_mapface(stimulus);
+    vec3 composited = srgb_oetf(transformed);
+
     if (lo <= hi) {
         const vec4 ui = imageLoad(ui_layer_read, ivec2(gl_GlobalInvocationID.xy)).rgba;
         composited = ui.rgb + (composited * (1.0 - ui.a));

diff --git a/title/shark-shaders/shaders/primitive_2d_bin.comp.glsl b/title/shark-shaders/shaders/primitive_2d_bin.comp.glsl
index 0010d718504801f9ddb97e9533d67eb861e43791..a826daef1601488393f9172a5c12a4aa94078f0f 100644 (file)
--- a/title/shark-shaders/shaders/primitive_2d_bin.comp.glsl
+++ b/title/shark-shaders/shaders/primitive_2d_bin.comp.glsl
@@ -45,17 +45,17 @@ void main() {
             continue;
         }
 
-        ivec2 bin_start = ivec2(floor(max(min(primitives_min, uniforms.screen_resolution), 0.0) / TILE_SIZE));
-        ivec2 bin_end = ivec2(floor((max(min(primitives_max, uniforms.screen_resolution), 0.0) + (TILE_SIZE - 1)) / TILE_SIZE));
+        const ivec2 tile_start = ivec2(floor(max(min(primitives_min, uniforms.screen_resolution), 0.0) / TILE_SIZE));
+        const ivec2 tile_end = ivec2(floor((max(min(primitives_max, uniforms.screen_resolution), 0.0) + (TILE_SIZE - 1)) / TILE_SIZE));
 
-        for (int y = bin_start.y; y < bin_end.y; y++) {
-            for (int x = bin_start.x; x < bin_end.x; x++) {
-                const uvec2 bin_coord = uvec2(x, y);
-                const uint bin_index = bin_coord.y * uniforms.tile_stride + bin_coord.x;
-                const vec2 bin_min = bin_coord * TILE_SIZE;
-                const vec2 bin_max = min(bin_min + TILE_SIZE, uniforms.screen_resolution);
+        for (int y = tile_start.y; y < tile_end.y; y++) {
+            for (int x = tile_start.x; x < tile_end.x; x++) {
+                const uvec2 tile_coord = uvec2(x, y);
+                const uint tile_index = tile_coord.y * uniforms.tile_stride + tile_coord.x;
+                const vec2 tile_min = tile_coord * TILE_SIZE;
+                const vec2 tile_max = min(tile_min + TILE_SIZE, uniforms.screen_resolution);
 
-                const bool intersects = !(any(lessThan(bin_max, primitive_min)) || any(greaterThan(bin_min, primitive_max)));
+                const bool intersects = !(any(lessThan(tile_max, primitive_min)) || any(greaterThan(tile_min, primitive_max)));
                 const uvec4 ballot = subgroupBallot(intersects);
 
                 if (ballot.x == 0 && ballot.y == 0) {
@@ -63,20 +63,20 @@ void main() {
                 }
 
                 if (ballot.x != 0) {
-                    uniforms.tiles.values[bin_index * TILE_STRIDE + TILE_BITMAP_L0_OFFSET + gl_WorkGroupID.x * 32 + word_index + 0] = ballot.x;
+                    uniforms.tiles.values[tile_index * TILE_STRIDE + TILE_BITMAP_L0_OFFSET + gl_WorkGroupID.x * 32 + word_index + 0] = ballot.x;
                 }
 
                 if (ballot.y != 0) {
-                    uniforms.tiles.values[bin_index * TILE_STRIDE + TILE_BITMAP_L0_OFFSET + gl_WorkGroupID.x * 32 + word_index + 1] = ballot.y;
+                    uniforms.tiles.values[tile_index * TILE_STRIDE + TILE_BITMAP_L0_OFFSET + gl_WorkGroupID.x * 32 + word_index + 1] = ballot.y;
                 }
 
                 if (subgroupElect()) {
-                    uniforms.tiles.values[bin_index * TILE_STRIDE + TILE_BITMAP_L1_OFFSET + gl_WorkGroupID.x] |=
+                    uniforms.tiles.values[tile_index * TILE_STRIDE + TILE_BITMAP_L1_OFFSET + gl_WorkGroupID.x] |=
                         (uint(ballot.x != 0) << (word_index + 0)) |
                         (uint(ballot.y != 0) << (word_index + 1));
 
-                    atomicMin(uniforms.tiles.values[bin_index * TILE_STRIDE + TILE_BITMAP_RANGE_LO_OFFSET], gl_WorkGroupID.x);
-                    atomicMax(uniforms.tiles.values[bin_index * TILE_STRIDE + TILE_BITMAP_RANGE_HI_OFFSET], gl_WorkGroupID.x);
+                    atomicMin(uniforms.tiles.values[tile_index * TILE_STRIDE + TILE_BITMAP_RANGE_LO_OFFSET], gl_WorkGroupID.x);
+                    atomicMax(uniforms.tiles.values[tile_index * TILE_STRIDE + TILE_BITMAP_RANGE_HI_OFFSET], gl_WorkGroupID.x);
                 }
             }
         }