Add compute shader support from TSnake41/raylib

examples/resources/gol.glsl

@@ -0,0 +1,57 @@
+#version 430
+
+#define GOL_WIDTH 1024
+
+layout (local_size_x = 16, local_size_y = 16, local_size_x = 1) in;
+
+layout(std430, binding = 1) readonly restrict buffer golLayout {
+  int golBuffer[]; // golBuffer[x, y] = golBuffer[x + gl_NumWorkGroups.x * y]
+};
+
+layout(std430, set = 1, binding = 2) writeonly restrict buffer golLayout2 {
+  int golBufferDest[]; // golBufferDest[x, y] = golBufferDest[x + gl_NumWorkGroups.x * y]
+};
+
+#define fetchGol(x, y) ((((x) < 0) || ((y) < 0) || ((x) > GOL_WIDTH) || ((y) > GOL_WIDTH)) \
+  ? (0) \
+  : golBuffer[(x) + GOL_WIDTH * (y)])
+
+#define setGol(x, y, value) golBufferDest[(x) + GOL_WIDTH * (y)] = value
+
+void main()
+{
+  uint neighbour_count = 0;
+  uint x = gl_GlobalInvocationID.x;
+  uint y = gl_GlobalInvocationID.y;
+
+  // Top left
+  neighbour_count += fetchGol(x - 1, y - 1);
+  
+  // Top middle
+  neighbour_count += fetchGol(x, y - 1);
+  
+  // Top right
+  neighbour_count += fetchGol(x + 1, y - 1);
+  
+  // Left
+  neighbour_count += fetchGol(x - 1, y);
+
+  // Right
+  neighbour_count += fetchGol(x + 1, y);
+  
+  // Bottom left
+  neighbour_count += fetchGol(x - 1, y + 1);
+  
+  // Bottom middle
+  neighbour_count += fetchGol(x, y + 1);
+  
+  // Bottom right
+  neighbour_count += fetchGol(x + 1, y + 1);
+
+  if (neighbour_count == 3)
+    setGol(x, y, 1);
+  else if (neighbour_count == 2)
+    setGol(x, y, fetchGol(x, y));
+  else
+    setGol(x, y, 0);
+}

examples/shaders_compute_shader.lua

@@ -0,0 +1,105 @@
+local ffi = require "ffi"
+local width, height = 1024, 1024
+rl.InitWindow(width, height, "raylib-lua [core] example - compute shader")
+local computeShaderCode
+do
+  local f = io.open("resources/gol.glsl", "rb")
+  assert(f, "Can't read resources/gol.glsl file")
+  computeShaderCode = f:read "*a"
+  f:close()
+end
+
+local computeShader = rl.rlCompileShader(computeShaderCode, rl.RL_COMPUTE_SHADER)
+local csProgram = rl.rlLoadComputeShaderProgram(computeShader)
+print(csProgram)
+
+local ssbo_size = ffi.sizeof("int32_t[?]", width * height)
+local ssbo_baseBuffer = ffi.new("int32_t[?]", width * height)
+for i=0,width * height - 1 do
+  ssbo_baseBuffer[i] = 0
+end
+
+local shader_framebuffer = rl.new("Color[?]", width * height)
+local ssbo_image = rl.new "Image" ---@type Image
+
+ssbo_image.data = shader_framebuffer
+ssbo_image.width = width
+ssbo_image.height = height
+ssbo_image.format = rl.PIXELFORMAT_UNCOMPRESSED_R8G8B8A8
+ssbo_image.mipmaps = 1
+
+local ssboA = rl.rlLoadShaderBuffer(ssbo_size, ssbo_baseBuffer, rl.RL_STREAM_COPY)
+local ssboB = rl.rlLoadShaderBuffer(ssbo_size, ssbo_baseBuffer, rl.RL_STREAM_COPY)
+
+-- Create a texture to apply shader
+local renderTexture = rl.LoadRenderTexture(width, height)
+
+local renderShader = rl.LoadShaderFromMemory(nil, [[
+#version 430
+out vec4 finalColor;
+in vec2 fragTexCoord;
+
+layout(std430, binding = 1) readonly buffer golLayout {
+  int golBuffer[];
+};
+
+uniform vec2 res;
+
+void main()
+{
+  ivec2 coords = ivec2(fragTexCoord * res);
+
+  if (golBuffer[coords.x + coords.y * uint(res.x)] == 1)
+    finalColor = vec4(1.0);
+  else
+    finalColor = vec4(0.0, 0.0, 0.0, 1.0);
+}
+]])
+
+local resolution = ffi.new("float[2]", width, height)
+local res_uniform = rl.GetShaderLocation(renderShader, "res")
+
+while not rl.WindowShouldClose() do
+  rl.BeginDrawing()
+
+  if rl.IsMouseButtonDown(rl.MOUSE_BUTTON_LEFT) then
+    rl.rlReadShaderBufferElements(ssboB, ssbo_baseBuffer, ssbo_size, 0)
+
+    -- Correct colors
+    for i=0,(width * height)-1 do
+      local toggled = ssbo_baseBuffer[i]
+
+      shader_framebuffer[i].r = toggled * 255
+      shader_framebuffer[i].g = toggled * 255
+      shader_framebuffer[i].b = toggled * 255
+      shader_framebuffer[i].a = toggled * 255
+    end
+
+    rl.ImageDrawRectangleV(ssbo_image, rl.GetMousePosition(), rl.new("Vector2", 25, 25), rl.WHITE)
+
+    for x=0,ssbo_image.width-1 do
+      for y=0,ssbo_image.height-1 do
+        ssbo_baseBuffer[x + y * width] = (shader_framebuffer[x + y * width].r > 0)
+      end
+    end
+    rl.rlUpdateShaderBufferElements(ssboB, ssbo_baseBuffer, ssbo_size, 0)
+  else
+    rl.rlEnableShader(csProgram)
+    rl.rlBindShaderBuffer(ssboA, 1)
+    rl.rlBindShaderBuffer(ssboB, 2)
+    rl.rlComputeShaderDispatch(width / 16, height / 16, 1)
+    rl.rlDisableShader()
+  end
+
+  rl.ClearBackground(rl.BLANK)
+  rl.SetShaderValue(renderShader, res_uniform, resolution, rl.SHADER_UNIFORM_VEC2)
+  rl.BeginShaderMode(renderShader)
+  rl.DrawTexture(renderTexture.texture, 0, 0, rl.WHITE)
+  rl.EndShaderMode()
+  rl.DrawFPS(0, 0)
+  rl.EndDrawing()
+
+  ssboA, ssboB = ssboB, ssboA
+end
+
+rl.CloseWindow()

makefile

@@ -6,7 +6,7 @@ LUA ?= luajit/src/luajit
 
 WINDRES ?= windres
 
-CFLAGS += -Iluajit/src -Iraylib/src -Iraygui/src
+CFLAGS += -Iluajit/src -Iraylib/src -Iraygui/src -DSUPPORT_COMPUTE_SHADERS
 LDFLAGS += luajit/src/libluajit.a raylib/src/libraylib.a
 
 MODULES := raymath rlgl easings gestures physac raygui

src/raylib.lua

@@ -809,6 +809,47 @@ ffi.cdef [[
   } rlRenderBatch;
 ]]
 
+-- rlgl defines
+rl.RL_TEXTURE_WRAP_S = 0x2802
+rl.RL_TEXTURE_WRAP_T = 0x2803
+rl.RL_TEXTURE_MAG_FILTER = 0x2800
+rl.RL_TEXTURE_MIN_FILTER = 0x2801
+rl.RL_TEXTURE_FILTER_NEAREST = 0x2600
+rl.RL_TEXTURE_FILTER_LINEAR = 0x2601
+rl.RL_TEXTURE_FILTER_MIP_NEAREST = 0x2700
+rl.RL_TEXTURE_FILTER_NEAREST_MIP_LINEAR = 0x2702
+rl.RL_TEXTURE_FILTER_LINEAR_MIP_NEAREST = 0x2701
+rl.RL_TEXTURE_FILTER_MIP_LINEAR = 0x2703
+rl.RL_TEXTURE_FILTER_ANISOTROPIC = 0x3000
+rl.RL_TEXTURE_WRAP_REPEAT = 0x2901
+rl.RL_TEXTURE_WRAP_CLAMP = 0x812F
+rl.RL_TEXTURE_WRAP_MIRROR_REPEAT = 0x8370
+rl.RL_TEXTURE_WRAP_MIRROR_CLAMP = 0x8742
+
+rl.RL_MODELVIEW = 0x1700
+rl.RL_PROJECTION = 0x1701
+rl.RL_TEXTURE = 0x1702
+rl.RL_LINES = 0x0001
+rl.RL_TRIANGLES = 0x0004
+rl.RL_QUADS = 0x0007
+
+rl.RL_UNSIGNED_BYTE = 0x1401
+rl.RL_FLOAT = 0x1406
+
+rl.RL_STREAM_DRAW = 0x88E0
+rl.RL_STREAM_READ = 0x88E1
+rl.RL_STREAM_COPY = 0x88E2
+rl.RL_STATIC_DRAW = 0x88E4
+rl.RL_STATIC_READ = 0x88E5
+rl.RL_STATIC_COPY = 0x88E6
+rl.RL_DYNAMIC_DRAW = 0x88E8
+rl.RL_DYNAMIC_READ = 0x88E9
+rl.RL_DYNAMIC_COPY = 0x88EA
+
+rl.RL_FRAGMENT_SHADER = 0x8B30
+rl.RL_VERTEX_SHADER = 0x8B31
+rl.RL_COMPUTE_SHADER = 0x91B9
+
 -- Physac cdef
 ffi.cdef [[
   typedef enum PhysicsShapeType { PHYSICS_CIRCLE = 0, PHYSICS_POLYGON } PhysicsShapeType;

tools/rlgl.h

@@ -116,6 +116,16 @@ void rlSetUniform(int locIndex, const void *value, int uniformType, int count)
 void rlSetUniformMatrix(int locIndex, Matrix mat)
 void rlSetUniformSampler(int locIndex, unsigned int textureId)
 void rlSetShader(unsigned int id, int *locs)
+unsigned int rlLoadComputeShaderProgram(int shaderId)
+void rlComputeShaderDispatch(unsigned int groupX, unsigned int groupY, unsigned int groupZ)
+unsigned int rlLoadShaderBuffer(unsigned long long size, const void *data, int usageHint)
+void rlUnloadShaderBuffer(unsigned int ssboId)
+void rlUpdateShaderBufferElements(unsigned int id, const void *data, unsigned long long dataSize, unsigned long long offset)
+unsigned long long rlGetShaderBufferSize(unsigned int id)
+void rlReadShaderBufferElements(unsigned int id, void *dest, unsigned long long count, unsigned long long offset)
+void rlBindShaderBuffer(unsigned int id, unsigned int index)
+void rlCopyBuffersElements(unsigned int destId, unsigned int srcId, unsigned long long destOffset, unsigned long long srcOffset, unsigned long long count)
+void rlBindImageTexture(unsigned int id, unsigned int index, unsigned int format, int readonly)
 Matrix rlGetMatrixModelview(void)
 Matrix rlGetMatrixProjection(void)
 Matrix rlGetMatrixTransform(void)