nothing works yet

src/CMakeLists.txt

@@ -66,6 +66,7 @@ if (ENABLE_OGLRENDERER)
         GPU_OpenGL.cpp
         GPU_OpenGL_shaders.h
         GPU3D_OpenGL.cpp
+        GPU3D_Compute.cpp
         GPU3D_OpenGL_shaders.h
         OpenGLSupport.cpp)
 

src/GPU3D.h

@@ -169,6 +169,7 @@ extern std::unique_ptr<Renderer3D> CurrentRenderer;
 
 #ifdef OGLRENDERER_ENABLED
 #include "GPU3D_OpenGL.h"
+#include "GPU3D_Compute.h"
 #endif
 
 #endif

src/GPU3D_Compute.h

@@ -0,0 +1,263 @@
+/*
+    Copyright 2016-2022 melonDS team
+
+    This file is part of melonDS.
+
+    melonDS is free software: you can redistribute it and/or modify it under
+    the terms of the GNU General Public License as published by the Free
+    Software Foundation, either version 3 of the License, or (at your option)
+    any later version.
+
+    melonDS is distributed in the hope that it will be useful, but WITHOUT ANY
+    WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+    FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License along
+    with melonDS. If not, see http://www.gnu.org/licenses/.
+*/
+
+#ifndef GPU3D_COMPUTE
+#define GPU3D_COMPUTE
+
+#include "GPU3D.h"
+
+#include "OpenGLSupport.h"
+
+#include "NonStupidBitfield.h"
+
+#include <unordered_map>
+
+namespace GPU3D
+{
+
+class ComputeRenderer : public Renderer3D
+{
+public:
+    ComputeRenderer();
+    ~ComputeRenderer() override;
+
+    bool Init() override;
+    void DeInit() override;
+    void Reset() override;
+
+    void SetRenderSettings(GPU::RenderSettings& settings) override;
+
+    void VCount144() override;
+
+    void RenderFrame() override;
+    void RestartFrame() override;
+    u32* GetLine(int line) override;
+
+    //dk::Fence FrameReady = {};
+    //dk::Fence FrameReserveFence = {};
+private:
+    GLuint ShaderInterpXSpans[2];
+    GLuint ShaderBinCombined;
+    GLuint ShaderDepthBlend[2];
+    GLuint ShaderRasteriseNoTexture[2];
+    GLuint ShaderRasteriseNoTextureToon[2];
+    GLuint ShaderRasteriseNoTextureHighlight[2];
+    GLuint ShaderRasteriseUseTextureDecal[2];
+    GLuint ShaderRasteriseUseTextureModulate[2];
+    GLuint ShaderRasteriseUseTextureToon[2];
+    GLuint ShaderRasteriseUseTextureHighlight[2];
+    GLuint ShaderRasteriseShadowMask[2];
+    GLuint ShaderClearCoarseBinMask;
+    GLuint ShaderClearIndirectWorkCount;
+    GLuint ShaderCalculateWorkListOffset;
+    GLuint ShaderSortWork;
+    GLuint ShaderFinalPass[8];
+
+    GLuint YSpanIndicesTextureMemory;
+    GLuint YSpanIndicesTexture;
+    GLuint YSpanSetupMemory;
+    GLuint XSpanSetupMemory;
+    GLuint BinResultMemory;
+    GLuint RenderPolygonMemory;
+    GLuint TileMemory;
+    GLuint FinalTileMemory;
+
+    u32 DummyLine[256] = {};
+
+    struct SpanSetupY
+    {
+        // Attributes
+        s32 Z0, Z1, W0, W1;
+        s32 ColorR0, ColorG0, ColorB0;
+        s32 ColorR1, ColorG1, ColorB1;
+        s32 TexcoordU0, TexcoordV0;
+        s32 TexcoordU1, TexcoordV1;
+
+        // Interpolator
+        s32 I0, I1;
+        s32 Linear;
+        s32 IRecip;
+        s32 W0n, W0d, W1d;
+
+        // Slope
+        s32 Increment;
+
+        s32 X0, X1, Y0, Y1;
+        s32 XMin, XMax;
+        s32 DxInitial;
+
+        s32 XCovIncr;
+        u32 IsDummy, __pad1;
+    };
+    struct SpanSetupX
+    {
+        s32 X0, X1;
+
+        s32 EdgeLenL, EdgeLenR, EdgeCovL, EdgeCovR;
+
+        s32 XRecip;
+
+        u32 Flags;
+
+        s32 Z0, Z1, W0, W1;
+        s32 ColorR0, ColorG0, ColorB0;
+        s32 ColorR1, ColorG1, ColorB1;
+        s32 TexcoordU0, TexcoordV0;
+        s32 TexcoordU1, TexcoordV1;
+
+        s32 CovLInitial, CovRInitial;
+    };
+    struct SetupIndices
+    {
+        u16 PolyIdx, SpanIdxL, SpanIdxR, Y;
+    };
+    struct RenderPolygon
+    {
+        u32 FirstXSpan;
+        s32 YTop, YBot;
+
+        s32 XMin, XMax;
+        s32 XMinY, XMaxY;
+
+        u32 Variant;
+        u32 Attr;
+
+        float TextureLayer;
+        u32 __pad0, __pad1;
+    };
+
+    static const int TileSize = 8;
+    static const int CoarseTileCountX = 8;
+    static const int CoarseTileCountY = 4;
+    static const int CoarseTileW = CoarseTileCountX * TileSize;
+    static const int CoarseTileH = CoarseTileCountY * TileSize;
+
+    static const int TilesPerLine = 256/TileSize;
+    static const int TileLines = 192/TileSize;
+
+    static const int BinStride = 2048/32;
+    static const int CoarseBinStride = BinStride/32;
+
+    static const int MaxWorkTiles = TilesPerLine*TileLines*48;
+    static const int MaxVariants = 256;
+
+    static const int UniformIdxCurVariant = 0;
+    static const int UniformIdxTextureSize = 1;
+
+    struct BinResult
+    {
+        u32 VariantWorkCount[MaxVariants*4];
+        u32 SortedWorkOffset[MaxVariants];
+
+        u32 SortWorkWorkCount[4];
+        u32 UnsortedWorkDescs[MaxWorkTiles*2];
+        u32 SortedWork[MaxWorkTiles*2];
+
+        u32 BinnedMaskCoarse[TilesPerLine*TileLines*CoarseBinStride];
+        u32 BinnedMask[TilesPerLine*TileLines*BinStride];
+        u32 WorkOffsets[TilesPerLine*TileLines*BinStride];
+    };
+
+    struct Tiles
+    {
+        u32 ColorTiles[MaxWorkTiles*TileSize*TileSize];
+        u32 DepthTiles[MaxWorkTiles*TileSize*TileSize];
+        u32 AttrStencilTiles[MaxWorkTiles*TileSize*TileSize];
+    };
+
+    struct FinalTiles
+    {
+        u32 ColorResult[256*192*2];
+        u32 DepthResult[256*192*2];
+        u32 AttrResult[256*192*2];
+    };
+
+    // eh those are pretty bad guesses
+    // though real hw shouldn't be eable to render all 2048 polygons on every line either
+    static const int MaxYSpanIndices = 64*2048;
+    static const int MaxYSpanSetups = 6144*2;
+    SetupIndices YSpanIndices[MaxYSpanIndices];
+    SpanSetupY YSpanSetups[MaxYSpanSetups];
+    RenderPolygon RenderPolygons[2048];
+
+    struct TexArrayEntry
+    {
+        u16 TexArrayIdx;
+        u16 LayerIdx;
+    };
+    struct TexArray
+    {
+        GLuint Image;
+        u32 ImageDescriptor;
+    };
+
+    struct TexCacheEntry
+    {
+        u32 DescriptorIdx;
+        u32 LastVariant; // very cheap way to make variant lookup faster
+
+        u32 TextureRAMStart[2], TextureRAMSize[2];
+        u32 TexPalStart, TexPalSize;
+        u8 WidthLog2, HeightLog2;
+        TexArrayEntry Texture;
+
+        u64 TextureHash[2];
+        u64 TexPalHash;
+    };
+    std::unordered_map<u64, TexCacheEntry> TexCache;
+
+    struct MetaUniform
+    {
+        u32 NumPolygons;
+        u32 NumVariants;
+
+        u32 AlphaRef;
+        u32 DispCnt;
+
+        u32 ToonTable[4*34];
+
+        u32 ClearColor, ClearDepth, ClearAttr;
+
+        u32 FogOffset, FogShift, FogColor;
+
+        u32 XScroll;
+    };
+    GLuint MetaUniformMemory;
+
+    static const u32 TexCacheMaxImages = 4096;
+
+    u32 FreeImageDescriptorsCount = 0;
+    u32 FreeImageDescriptors[TexCacheMaxImages];
+
+    std::vector<TexArrayEntry> FreeTextures[8][8];
+    std::vector<TexArray> TexArrays[8][8];
+
+    u32 TextureDecodingBuffer[1024*1024];
+
+    TexCacheEntry& GetTexture(u32 textureParam, u32 paletteParam);
+
+    void SetupAttrs(SpanSetupY* span, Polygon* poly, int from, int to);
+    void SetupYSpan(int polynum, SpanSetupY* span, Polygon* poly, int from, int to, u32 y, int side);
+    void SetupYSpanDummy(SpanSetupY* span, Polygon* poly, int vertex, int side);
+
+    bool CompileShader(GLuint& shader, const char* source, const std::initializer_list<const char*>& defines);
+};
+
+}
+
+#endif

src/GPU3D_OpenGL.cpp

@@ -49,25 +49,18 @@ bool GLRenderer::BuildRenderShader(u32 flags, const char* vs, const char* fs)
     strcpy(&fsbuf[headerlen], kRenderFSCommon);
     strcpy(&fsbuf[headerlen + fsclen], fs);
 
-    bool ret = OpenGL::BuildShaderProgram(vsbuf, fsbuf, RenderShader[flags], shadername);
+    GLuint prog;
+    bool ret = OpenGL::CompileVertexFragmentProgram(prog,
+        vsbuf, fsbuf,
+        shadername,
+        {{"vPosition", 0}, {"vColor", 1}, {"vTexcoord", 2}, {"vPolygonAttr", 3}},
+        {{"oColor", 0}, {"oAttr", 1}});
 
     delete[] vsbuf;
     delete[] fsbuf;
 
     if (!ret) return false;
 
-    GLuint prog = RenderShader[flags][2];
-
-    glBindAttribLocation(prog, 0, "vPosition");
-    glBindAttribLocation(prog, 1, "vColor");
-    glBindAttribLocation(prog, 2, "vTexcoord");
-    glBindAttribLocation(prog, 3, "vPolygonAttr");
-    glBindFragDataLocation(prog, 0, "oColor");
-    glBindFragDataLocation(prog, 1, "oAttr");
-
-    if (!OpenGL::LinkShaderProgram(RenderShader[flags]))
-        return false;
-
     GLint uni_id = glGetUniformBlockIndex(prog, "uConfig");
     glUniformBlockBinding(prog, uni_id, 0);
 
@@ -78,13 +71,15 @@ bool GLRenderer::BuildRenderShader(u32 flags, const char* vs, const char* fs)
     uni_id = glGetUniformLocation(prog, "TexPalMem");
     glUniform1i(uni_id, 1);
 
+    RenderShader[flags] = prog;
+
     return true;
 }
 
 void GLRenderer::UseRenderShader(u32 flags)
 {
     if (CurShaderID == flags) return;
-    glUseProgram(RenderShader[flags][2]);
+    glUseProgram(RenderShader[flags]);
     CurShaderID = flags;
 }
 
@@ -118,21 +113,17 @@ std::unique_ptr<GLRenderer> GLRenderer::New() noexcept
     glDepthRange(0, 1);
     glClearDepth(1.0);
 
-
+    if (!OpenGL::CompileVertexFragmentProgram(result->ClearShaderPlain,
-    if (!OpenGL::BuildShaderProgram(kClearVS, kClearFS, result->ClearShaderPlain, "ClearShader"))
+            kClearVS, kClearFS,
+            "ClearShader",
+            {{"vPosition", 0}},
+            {{"oColor", 0}, {"oAttr", 1}}))
         return nullptr;
 
-    glBindAttribLocation(result->ClearShaderPlain[2], 0, "vPosition");
+    result->ClearUniformLoc[0] = glGetUniformLocation(result->ClearShaderPlain, "uColor");
-    glBindFragDataLocation(result->ClearShaderPlain[2], 0, "oColor");
+    result->ClearUniformLoc[1] = glGetUniformLocation(result->ClearShaderPlain, "uDepth");
-    glBindFragDataLocation(result->ClearShaderPlain[2], 1, "oAttr");
+    result->ClearUniformLoc[2] = glGetUniformLocation(result->ClearShaderPlain, "uOpaquePolyID");
-
+    result->ClearUniformLoc[3] = glGetUniformLocation(result->ClearShaderPlain, "uFogFlag");
-    if (!OpenGL::LinkShaderProgram(result->ClearShaderPlain))
-        return nullptr;
-
-    result->ClearUniformLoc[0] = glGetUniformLocation(result->ClearShaderPlain[2], "uColor");
-    result->ClearUniformLoc[1] = glGetUniformLocation(result->ClearShaderPlain[2], "uDepth");
-    result->ClearUniformLoc[2] = glGetUniformLocation(result->ClearShaderPlain[2], "uOpaquePolyID");
-    result->ClearUniformLoc[3] = glGetUniformLocation(result->ClearShaderPlain[2], "uFogFlag");
 
     memset(result->RenderShader, 0, sizeof(RenderShader));
 
@@ -160,42 +151,35 @@ std::unique_ptr<GLRenderer> GLRenderer::New() noexcept
     if (!result->BuildRenderShader(RenderFlag_ShadowMask | RenderFlag_WBuffer, kRenderVS_W, kRenderFS_WSM))
         return nullptr;
 
-    if (!OpenGL::BuildShaderProgram(kFinalPassVS, kFinalPassEdgeFS, result->FinalPassEdgeShader, "FinalPassEdgeShader"))
+    if (!OpenGL::CompileVertexFragmentProgram(result->FinalPassEdgeShader,
+            kFinalPassVS, kFinalPassEdgeFS,
+            "FinalPassEdgeShader",
+            {{"vPosition", 0}},
+            {{"oColor", 0}}))
+        return nullptr;
+    if (!OpenGL::CompileVertexFragmentProgram(result->FinalPassFogShader,
+            kFinalPassVS, kFinalPassFogFS,
+            "FinalPassFogShader",
+            {{"vPosition", 0}},
+            {{"oColor", 0}}))
         return nullptr;
 
-    if (!OpenGL::BuildShaderProgram(kFinalPassVS, kFinalPassFogFS, result->FinalPassFogShader, "FinalPassFogShader"))
+    GLuint uni_id = glGetUniformBlockIndex(result->FinalPassEdgeShader, "uConfig");
-        return nullptr;
+    glUniformBlockBinding(result->FinalPassEdgeShader, uni_id, 0);
 
-    glBindAttribLocation(result->FinalPassEdgeShader[2], 0, "vPosition");
+    glUseProgram(result->FinalPassEdgeShader);
-    glBindFragDataLocation(result->FinalPassEdgeShader[2], 0, "oColor");
+    uni_id = glGetUniformLocation(result->FinalPassEdgeShader, "DepthBuffer");
-
-    if (!OpenGL::LinkShaderProgram(result->FinalPassEdgeShader))
-        return nullptr;
-
-    GLint uni_id = glGetUniformBlockIndex(result->FinalPassEdgeShader[2], "uConfig");
-    glUniformBlockBinding(result->FinalPassEdgeShader[2], uni_id, 0);
-
-    glUseProgram(result->FinalPassEdgeShader[2]);
-
-    uni_id = glGetUniformLocation(result->FinalPassEdgeShader[2], "DepthBuffer");
     glUniform1i(uni_id, 0);
-    uni_id = glGetUniformLocation(result->FinalPassEdgeShader[2], "AttrBuffer");
+    uni_id = glGetUniformLocation(result->FinalPassEdgeShader, "AttrBuffer");
     glUniform1i(uni_id, 1);
 
-    glBindAttribLocation(result->FinalPassFogShader[2], 0, "vPosition");
+    uni_id = glGetUniformBlockIndex(result->FinalPassFogShader, "uConfig");
-    glBindFragDataLocation(result->FinalPassFogShader[2], 0, "oColor");
+    glUniformBlockBinding(result->FinalPassFogShader, uni_id, 0);
 
-    if (!OpenGL::LinkShaderProgram(result->FinalPassFogShader))
+    glUseProgram(result->FinalPassFogShader);
-        return nullptr;
+    uni_id = glGetUniformLocation(result->FinalPassFogShader, "DepthBuffer");
-
-    uni_id = glGetUniformBlockIndex(result->FinalPassFogShader[2], "uConfig");
-    glUniformBlockBinding(result->FinalPassFogShader[2], uni_id, 0);
-
-    glUseProgram(result->FinalPassFogShader[2]);
-
-    uni_id = glGetUniformLocation(result->FinalPassFogShader[2], "DepthBuffer");
     glUniform1i(uni_id, 0);
-    uni_id = glGetUniformLocation(result->FinalPassFogShader[2], "AttrBuffer");
+    uni_id = glGetUniformLocation(result->FinalPassFogShader, "AttrBuffer");
     glUniform1i(uni_id, 1);
 
 
@@ -320,8 +304,8 @@ GLRenderer::~GLRenderer()
 
     for (int i = 0; i < 16; i++)
     {
-        if (!RenderShader[i][2]) continue;
+        if (!RenderShader[i]) continue;
-        OpenGL::DeleteShaderProgram(RenderShader[i]);
+        glDeleteProgram(RenderShader[i]);
     }
 }
 
@@ -1093,7 +1077,7 @@ void GLRenderer::RenderSceneChunk(int y, int h)
             // edge marking
             // TODO: depth/polyid values at screen edges
 
-            glUseProgram(FinalPassEdgeShader[2]);
+            glUseProgram(FinalPassEdgeShader);
 
             glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ZERO, GL_ONE);
 
@@ -1104,7 +1088,7 @@ void GLRenderer::RenderSceneChunk(int y, int h)
         {
             // fog
 
-            glUseProgram(FinalPassFogShader[2]);
+            glUseProgram(FinalPassFogShader);
 
             if (RenderDispCnt & (1<<6))
                 glBlendFuncSeparate(GL_ZERO, GL_ONE, GL_CONSTANT_COLOR, GL_ONE_MINUS_SRC_ALPHA);
@@ -1238,7 +1222,7 @@ void GLRenderer::RenderFrame()
     // TODO: check whether 'clear polygon ID' affects translucent polyID
     // (for example when alpha is 1..30)
     {
-        glUseProgram(ClearShaderPlain[2]);
+        glUseProgram(ClearShaderPlain);
         glDepthFunc(GL_ALWAYS);
 
         u32 r = RenderClearAttr1 & 0x1F;

src/GPU3D_OpenGL.h

@@ -84,13 +84,13 @@ private:
     };
 
 
-    GLuint ClearShaderPlain[3] {};
+    GLuint ClearShaderPlain {};
 
-    GLuint RenderShader[16][3] {};
+    GLuint RenderShader[16] {};
     GLuint CurShaderID = -1;
 
-    GLuint FinalPassEdgeShader[3] {};
+    GLuint FinalPassEdgeShader {};
-    GLuint FinalPassFogShader[3] {};
+    GLuint FinalPassFogShader {};
 
     // std140 compliant structure
     struct

src/GPU_OpenGL.cpp

@@ -36,32 +36,27 @@ using namespace OpenGL;
 std::unique_ptr<GLCompositor> GLCompositor::New() noexcept
 {
     assert(glBindAttribLocation != nullptr);
+    GLuint CompShader {};
 
-    std::array<GLuint, 3> CompShader {};
+    if (!OpenGL::CompileVertexFragmentProgram(CompShader,
-    if (!OpenGL::BuildShaderProgram(kCompositorVS, kCompositorFS_Nearest, &CompShader[0], "CompositorShader"))
+            kCompositorVS, kCompositorFS_Nearest, 
-        return nullptr;
+            "CompositorShader",
-
+            {{"vPosition", 0}, {"vTexcoord", 1}},
-    glBindAttribLocation(CompShader[2], 0, "vPosition");
+            {{"oColor", 0}}))
-    glBindAttribLocation(CompShader[2], 1, "vTexcoord");
-    glBindFragDataLocation(CompShader[2], 0, "oColor");
-
-    if (!OpenGL::LinkShaderProgram(CompShader.data()))
-        // OpenGL::LinkShaderProgram already deletes the shader program object
-        // if linking the shaders together failed.
         return nullptr;
 
     return std::unique_ptr<GLCompositor>(new GLCompositor(CompShader));
 }
 
-GLCompositor::GLCompositor(std::array<GLuint, 3> compShader) noexcept : CompShader(compShader)
+GLCompositor::GLCompositor(GLuint compShader) noexcept : CompShader(compShader)
 {
-    CompScaleLoc = glGetUniformLocation(CompShader[2], "u3DScale");
+    CompScaleLoc = glGetUniformLocation(CompShader, "u3DScale");
-    Comp3DXPosLoc = glGetUniformLocation(CompShader[2], "u3DXPos");
+    Comp3DXPosLoc = glGetUniformLocation(CompShader, "u3DXPos");
 
-    glUseProgram(CompShader[2]);
+    glUseProgram(CompShader);
-    GLuint screenTextureUniform = glGetUniformLocation(CompShader[2], "ScreenTex");
+    GLuint screenTextureUniform = glGetUniformLocation(CompShader, "ScreenTex");
     glUniform1i(screenTextureUniform, 0);
-    GLuint _3dTextureUniform = glGetUniformLocation(CompShader[2], "_3DTex");
+    GLuint _3dTextureUniform = glGetUniformLocation(CompShader, "_3DTex");
     glUniform1i(_3dTextureUniform, 1);
 
     // all this mess is to prevent bleeding
@@ -136,7 +131,7 @@ GLCompositor::~GLCompositor()
     glDeleteVertexArrays(1, &CompVertexArrayID);
     glDeleteBuffers(1, &CompVertexBufferID);
 
-    OpenGL::DeleteShaderProgram(CompShader.data());
+    glDeleteProgram(CompShader);
 }
 
 void GLCompositor::Reset()
@@ -200,7 +195,7 @@ void GLCompositor::RenderFrame()
     glClear(GL_COLOR_BUFFER_BIT);
 
     // TODO: select more shaders (filtering, etc)
-    OpenGL::UseShaderProgram(CompShader.data());
+    glUseProgram(CompShader);
     glUniform1ui(CompScaleLoc, Scale);
 
     // TODO: support setting this midframe, if ever needed

src/GPU_OpenGL.h

@@ -44,12 +44,12 @@ public:
     void RenderFrame();
     void BindOutputTexture(int buf);
 private:
-    GLCompositor(std::array<GLuint, 3> CompShader) noexcept;
+    GLCompositor(GLuint CompShader) noexcept;
 
     int Scale;
     int ScreenH, ScreenW;
 
-    std::array<GLuint, 3> CompShader;
+    GLuint CompShader;
     GLuint CompScaleLoc;
     GLuint Comp3DXPosLoc;
 

src/NonStupidBitfield.h

@@ -26,6 +26,32 @@
 #include <initializer_list>
 #include <algorithm>
 
+inline u64 GetRangedBitMask(u32 idx, u32 startBit, u32 bitsCount)
+{
+    u32 startEntry = startBit >> 6;
+    u64 entriesCount = ((startBit + bitsCount + 0x3F) >> 6) - startEntry;
+
+    if (entriesCount > 1)
+    {
+        if (idx == startEntry)
+            return 0xFFFFFFFFFFFFFFFF << (startBit & 0x3F);
+        if (((startBit + bitsCount) & 0x3F) && idx == startEntry + entriesCount - 1)
+            return ~(0xFFFFFFFFFFFFFFFF << ((startBit + bitsCount) & 0x3F));
+        else
+            return 0xFFFFFFFFFFFFFFFF;
+
+        return 0xFFFFFFFFFFFFFFFF;
+    }
+    else if (idx == startEntry)
+    {
+        return ((1ULL << bitsCount) - 1) << (startBit & 0x3F);
+    }
+    else
+    {
+        return 0;
+    }
+}
+
 // like std::bitset but less stupid and optimised for 
 // our use case (keeping track of memory invalidations)
 
@@ -164,6 +190,11 @@ struct NonStupidBitField
         return Ref{*this, idx};
     }
 
+    bool operator[](u32 idx) const
+    {
+        return Data[idx >> 6] & (1ULL << (idx & 0x3F));
+    }
+
     void SetRange(u32 startBit, u32 bitsCount)
     {
         u32 startEntry = startBit >> 6;
@@ -185,6 +216,26 @@ struct NonStupidBitField
         }
     }
 
+    int Min() const
+    {
+        for (int i = 0; i < DataLength; i++)
+        {
+            if (Data[i])
+                return i * 64 + __builtin_ctzll(Data[i]);
+        }
+        return -1;
+    }
+
+    int Max() const
+    {
+        for (int i = DataLength - 1; i >= 0; i--)
+        {
+            if (Data[i])
+                return i * 64 + (63 - __builtin_clzll(Data[i]));
+        }
+        return -1;
+    }
+
     NonStupidBitField& operator|=(const NonStupidBitField<Size>& other)
     {
         for (u32 i = 0; i < DataLength; i++)
@@ -193,6 +244,7 @@ struct NonStupidBitField
         }
         return *this;
     }
+
     NonStupidBitField& operator&=(const NonStupidBitField<Size>& other)
     {
         for (u32 i = 0; i < DataLength; i++)
@@ -201,6 +253,20 @@ struct NonStupidBitField
         }
         return *this;
     }
+
+    operator bool() const
+    {
+        for (int i = 0; i < DataLength - 1; i++)
+        {
+            if (Data[i])
+                return true;
+        }
+        if (Data[DataLength-1] & ((Size&0x3F) ? ~(0xFFFFFFFFFFFFFFFF << (Size&0x3F)) : 0xFFFFFFFFFFFFFFFF))
+        {
+            return true;
+        }
+        return false;
+    }
 };
 
 

src/OpenGLSupport.cpp

@@ -24,7 +24,9 @@ using Platform::LogLevel;
 namespace OpenGL
 {
 
-bool BuildShaderProgram(const char* vs, const char* fs, GLuint* ids, const char* name)
+#define checkGLError() if (glGetError() != GL_NO_ERROR) printf("error %d\n", __LINE__)
+
+bool CompilerShader(GLuint& id, const char* source, const char* name, const char* type)
 {
     int len;
     int res;
@@ -35,61 +37,31 @@ bool BuildShaderProgram(const char* vs, const char* fs, GLuint* ids, const char*
         return false;
     }
 
-    ids[0] = glCreateShader(GL_VERTEX_SHADER);
+    len = strlen(source);
-    len = strlen(vs);
+    glShaderSource(id, 1, &source, &len);
-    glShaderSource(ids[0], 1, &vs, &len);
+    checkGLError();
-    glCompileShader(ids[0]);
+    glCompileShader(id);
+    checkGLError();
 
-    glGetShaderiv(ids[0], GL_COMPILE_STATUS, &res);
+    glGetShaderiv(id, GL_COMPILE_STATUS, &res);
+    checkGLError();
     if (res != GL_TRUE)
     {
-        glGetShaderiv(ids[0], GL_INFO_LOG_LENGTH, &res);
+        glGetShaderiv(id, GL_INFO_LOG_LENGTH, &res);
         if (res < 1) res = 1024;
         char* log = new char[res+1];
-        glGetShaderInfoLog(ids[0], res+1, NULL, log);
+        glGetShaderInfoLog(id, res+1, NULL, log);
-        Log(LogLevel::Error, "OpenGL: failed to compile vertex shader %s: %s\n", name, log);
+        Log(LogLevel::Error, "OpenGL: failed to compile %s shader %s: %s\n", type, name, log);
-        Log(LogLevel::Debug, "shader source:\n--\n%s\n--\n", vs);
+        Log(LogLevel::Debug, "shader source:\n--\n%s\n--\n", source);
         delete[] log;
 
-        glDeleteShader(ids[0]);
-
         return false;
     }
 
-    ids[1] = glCreateShader(GL_FRAGMENT_SHADER);
-    len = strlen(fs);
-    glShaderSource(ids[1], 1, &fs, &len);
-    glCompileShader(ids[1]);
-
-    glGetShaderiv(ids[1], GL_COMPILE_STATUS, &res);
-    if (res != GL_TRUE)
-    {
-        glGetShaderiv(ids[1], GL_INFO_LOG_LENGTH, &res);
-        if (res < 1) res = 1024;
-        char* log = new char[res+1];
-        glGetShaderInfoLog(ids[1], res+1, NULL, log);
-        Log(LogLevel::Error, "OpenGL: failed to compile fragment shader %s: %s\n", name, log);
-        //printf("shader source:\n--\n%s\n--\n", fs);
-        delete[] log;
-
-        FILE* logf = fopen("shaderfail.log", "w");
-        fwrite(fs, len+1, 1, logf);
-        fclose(logf);
-
-        glDeleteShader(ids[0]);
-        glDeleteShader(ids[1]);
-
-        return false;
-    }
-
-    ids[2] = glCreateProgram();
-    glAttachShader(ids[2], ids[0]);
-    glAttachShader(ids[2], ids[1]);
-
     return true;
 }
 
-bool LinkShaderProgram(GLuint* ids)
+bool LinkProgram(GLuint& result, GLuint* ids, int numIds)
 {
     int res;
 
@@ -99,46 +71,100 @@ bool LinkShaderProgram(GLuint* ids)
         return false;
     }
 
-    glLinkProgram(ids[2]);
+    for (int i = 0; i < numIds; i++)
+    {
+        glAttachShader(result, ids[i]);
+        checkGLError();
+    }
 
-    glDetachShader(ids[2], ids[0]);
+    glLinkProgram(result);
-    glDetachShader(ids[2], ids[1]);
 
-    glDeleteShader(ids[0]);
+    for (int i = 0; i < numIds; i++)
-    glDeleteShader(ids[1]);
+        glDetachShader(result, ids[i]);
 
-    glGetProgramiv(ids[2], GL_LINK_STATUS, &res);
+    glGetProgramiv(result, GL_LINK_STATUS, &res);
     if (res != GL_TRUE)
     {
-        glGetProgramiv(ids[2], GL_INFO_LOG_LENGTH, &res);
+        glGetProgramiv(result, GL_INFO_LOG_LENGTH, &res);
         if (res < 1) res = 1024;
         char* log = new char[res+1];
-        glGetProgramInfoLog(ids[2], res+1, NULL, log);
+        glGetProgramInfoLog(result, res+1, NULL, log);
         Log(LogLevel::Error, "OpenGL: failed to link shader program: %s\n", log);
         delete[] log;
 
-        glDeleteProgram(ids[2]);
-
         return false;
     }
 
     return true;
 }
 
-void DeleteShaderProgram(GLuint* ids)
+bool CompileComputeProgram(GLuint& result, const char* source, const char* name)
 {
+    GLuint shader = glCreateShader(GL_COMPUTE_SHADER);
+    bool linkingSucess = false;
     if (glDeleteProgram)
     { // If OpenGL isn't loaded, then there's no shader program to delete
-        glDeleteProgram(ids[2]);
+        goto error;
     }
+
+    result = glCreateProgram();
+
+    printf("compiling %s", name);
+    if (!CompilerShader(shader, source, name, "compute"))
+        goto error;
+
+    linkingSucess = LinkProgram(result, &shader, 1);
+
+error:
+    glDeleteShader(shader);
+
+    if (!linkingSucess)
+        glDeleteProgram(result);
+
+    return linkingSucess;
 }
 
-void UseShaderProgram(GLuint* ids)
+bool CompileVertexFragmentProgram(GLuint& result,
+    const char* vs, const char* fs,
+    const char* name,
+    const std::initializer_list<AttributeTarget>& vertexInAttrs,
+    const std::initializer_list<AttributeTarget>& fragmentOutAttrs)
 {
-    if (glUseProgram)
+    GLuint shaders[2] =
-    { // If OpenGL isn't loaded, then there's no shader program to use
+    {
-        glUseProgram(ids[2]);
+        glCreateShader(GL_VERTEX_SHADER),
+        glCreateShader(GL_FRAGMENT_SHADER)
+    };
+    result = glCreateProgram();
+
+    bool linkingSucess = false;
+
+    if (!CompilerShader(shaders[0], vs, name, "vertex"))
+        goto error;
+
+    if (!CompilerShader(shaders[1], fs, name, "fragment"))
+        goto error;
+
+
+    for (const AttributeTarget& target : vertexInAttrs)
+    {
+        glBindAttribLocation(result, target.Location, target.Name);
     }
+    for (const AttributeTarget& target : fragmentOutAttrs)
+    {
+        glBindFragDataLocation(result, target.Location, target.Name);
+    }
+
+    linkingSucess = LinkProgram(result, shaders, 2);
+
+error:
+    glDeleteShader(shaders[1]);
+    glDeleteShader(shaders[0]);
+
+    if (!linkingSucess)
+        glDeleteProgram(result);
+
+    return linkingSucess;
 }
 
 }

src/OpenGLSupport.h

@@ -29,10 +29,19 @@
 namespace OpenGL
 {
 
-bool BuildShaderProgram(const char* vs, const char* fs, GLuint* ids, const char* name);
+struct AttributeTarget
-bool LinkShaderProgram(GLuint* ids);
+{
-void DeleteShaderProgram(GLuint* ids);
+    const char* Name;
-void UseShaderProgram(GLuint* ids);
+    u32 Location;
+};
+
+bool CompileVertexFragmentProgram(GLuint& result,
+    const char* vs, const char* fs,
+    const char* name,
+    const std::initializer_list<AttributeTarget>& vertexInAttrs,
+    const std::initializer_list<AttributeTarget>& fragmentOutAttrs);
+
+bool CompileComputeProgram(GLuint& result, const char* source, const char* name);
 
 }
 

src/frontend/qt_sdl/CMakeLists.txt

@@ -24,6 +24,7 @@ set(SOURCES_QT_SDL
     ROMInfoDialog.cpp
     RAMInfoDialog.cpp
     TitleManagerDialog.cpp
+    RTCSettingsDialog.cpp
     Input.cpp
     LAN_PCap.cpp
     LAN_Socket.cpp

src/frontend/qt_sdl/OSD.cpp

@@ -57,7 +57,7 @@ struct Item
 
 std::deque<Item> ItemQueue;
 
-GLuint Shader[3];
+GLuint Shader;
 GLint uScreenSize, uOSDPos, uOSDSize;
 GLfloat uScaleFactor;
 GLuint OSDVertexArray;
@@ -70,20 +70,19 @@ bool Init(bool openGL)
 {
     if (openGL)
     {
-        OpenGL::BuildShaderProgram(kScreenVS_OSD, kScreenFS_OSD, Shader, "OSDShader");
+        OpenGL::CompileVertexFragmentProgram(Shader,
+            kScreenVS_OSD, kScreenFS_OSD,
+            "OSDShader",
+            {{"vPosition", 0}},
+            {{"oColor", 0}});
 
-        GLuint pid = Shader[2];
+        glUseProgram(Shader);
-        glBindAttribLocation(pid, 0, "vPosition");
+        glUniform1i(glGetUniformLocation(Shader, "OSDTex"), 0);
-        glBindFragDataLocation(pid, 0, "oColor");
 
-        OpenGL::LinkShaderProgram(Shader);
+        uScreenSize = glGetUniformLocation(Shader, "uScreenSize");
-        glUseProgram(pid);
+        uOSDPos = glGetUniformLocation(Shader, "uOSDPos");
-        glUniform1i(glGetUniformLocation(pid, "OSDTex"), 0);
+        uOSDSize = glGetUniformLocation(Shader, "uOSDSize");
-
+        uScaleFactor = glGetUniformLocation(Shader, "uScaleFactor");
-        uScreenSize = glGetUniformLocation(pid, "uScreenSize");
-        uOSDPos = glGetUniformLocation(pid, "uOSDPos");
-        uOSDSize = glGetUniformLocation(pid, "uOSDSize");
-        uScaleFactor = glGetUniformLocation(pid, "uScaleFactor");
 
         float vertices[6*2] =
         {
@@ -425,7 +424,7 @@ void DrawGL(float w, float h)
 
     u32 y = kOSDMargin;
 
-    glUseProgram(Shader[2]);
+    glUseProgram(Shader);
 
     glUniform2f(uScreenSize, w, h);
     glUniform1f(uScaleFactor, mainWindow->devicePixelRatioF());

src/frontend/qt_sdl/main.cpp

@@ -228,19 +228,17 @@ void EmuThread::initOpenGL()
     oglContext = windowctx;
     oglContext->MakeCurrent();
 
-    OpenGL::BuildShaderProgram(kScreenVS, kScreenFS, screenShaderProgram, "ScreenShader");
+    OpenGL::CompileVertexFragmentProgram(screenShaderProgram,
-    GLuint pid = screenShaderProgram[2];
+        kScreenVS, kScreenFS,
-    glBindAttribLocation(pid, 0, "vPosition");
+        "ScreenShader",
-    glBindAttribLocation(pid, 1, "vTexcoord");
+        {{"vPosition", 0}, {"vTexcoord", 1}},
-    glBindFragDataLocation(pid, 0, "oColor");
+        {{"oColor", 0}});
 
-    OpenGL::LinkShaderProgram(screenShaderProgram);
+    glUseProgram(screenShaderProgram);
+    glUniform1i(glGetUniformLocation(screenShaderProgram, "ScreenTex"), 0);
 
-    glUseProgram(pid);
+    screenShaderScreenSizeULoc = glGetUniformLocation(screenShaderProgram, "uScreenSize");
-    glUniform1i(glGetUniformLocation(pid, "ScreenTex"), 0);
+    screenShaderTransformULoc = glGetUniformLocation(screenShaderProgram, "uTransform");
-
-    screenShaderScreenSizeULoc = glGetUniformLocation(pid, "uScreenSize");
-    screenShaderTransformULoc = glGetUniformLocation(pid, "uTransform");
 
     // to prevent bleeding between both parts of the screen
     // with bilinear filtering enabled
@@ -300,7 +298,7 @@ void EmuThread::deinitOpenGL()
     glDeleteVertexArrays(1, &screenVertexArray);
     glDeleteBuffers(1, &screenVertexBuffer);
 
-    OpenGL::DeleteShaderProgram(screenShaderProgram);
+    glDeleteProgram(screenShaderProgram);
 
     OSD::DeInit();
 
@@ -749,7 +747,7 @@ void EmuThread::drawScreenGL()
 
     glViewport(0, 0, w, h);
 
-    glUseProgram(screenShaderProgram[2]);
+    glUseProgram(screenShaderProgram);
     glUniform2f(screenShaderScreenSizeULoc, w / factor, h / factor);
 
     int frontbuf = FrontBuffer;

src/frontend/qt_sdl/main.h

@@ -121,7 +121,7 @@ private:
     GL::Context* oglContext = nullptr;
     GLuint screenVertexBuffer, screenVertexArray;
     GLuint screenTexture;
-    GLuint screenShaderProgram[3];
+    GLuint screenShaderProgram;
     GLuint screenShaderTransformULoc, screenShaderScreenSizeULoc;
 
     QMutex screenSettingsLock;