Rendertarget attachments, texture and image changes (#358)

* Add multiple color outputs for fragment shaders

* Add registers and gal enums

* Use textures for framebuffers and split color and zeta framebuffers

* Abstract texture and framebuffer targets as an image

* Share images between framebuffers and textures

* Unstub formats

* Add some formats

* Disable multiple attachments

* Cache framebuffer attachments

* Handle format types

* Add some rendertarget formats

* Code cleanup

* Fixup half float types

* Address feedback

* Disable multiple attachments in shaders

* Add A4B4G4R4 image format

* Add reversed section for image enums

6 files changed, 174 insertions, 81 deletions

diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs
index d2c5f126..7fb5ea8a 100644
--- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs
+++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs
@@ -154,16 +154,12 @@ namespace Ryujinx.HLE.Gpu.Engines
             }
             else if (IsDstFb)
             {
-                //Texture -> Frame Buffer copy.
-                const GalTextureFormat Format = GalTextureFormat.A8B8G8R8;
-
                 byte[] Buffer = TextureReader.Read(Vmm, SrcTexture());
 
                 Gpu.Renderer.FrameBuffer.SetBufferData(
                     DstKey,
                     DstWidth,
                     DstHeight,
-                    Format,
                     Buffer);
             }
             else
diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs
index 6f601244..1d0834dd 100644
--- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs
+++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs
@@ -102,7 +102,9 @@ namespace Ryujinx.HLE.Gpu.Engines
             SetAlphaBlending(State);
             SetPrimitiveRestart(State);
 
+            //Enabling multiple framebuffer attachments cause graphics reggresions
             SetFrameBuffer(Vmm, 0);
+            SetZeta(Vmm);
 
             long[] Keys = UploadShaders(Vmm);
 
@@ -149,9 +151,11 @@ namespace Ryujinx.HLE.Gpu.Engines
             int Stencil = ReadRegister(NvGpuEngine3dReg.ClearStencil);
 
             SetFrameBuffer(Vmm, FbIndex);
+            SetZeta(Vmm);
 
             Gpu.Renderer.Rasterizer.ClearBuffers(
                 Flags,
+                FbIndex,
                 Red, Green, Blue, Alpha,
                 Depth,
                 Stencil);
@@ -161,6 +165,15 @@ namespace Ryujinx.HLE.Gpu.Engines
         {
             long VA = MakeInt64From2xInt32(NvGpuEngine3dReg.FrameBufferNAddress + FbIndex * 0x10);
 
+            int Format = ReadRegister(NvGpuEngine3dReg.FrameBufferNFormat + FbIndex * 0x10);
+
+            if (VA == 0 || Format == 0)
+            {
+                Gpu.Renderer.FrameBuffer.UnbindColor(FbIndex);
+
+                return;
+            }
+
             long Key = Vmm.GetPhysicalAddress(VA);
 
             FrameBuffers.Add(Key);
@@ -168,11 +181,11 @@ namespace Ryujinx.HLE.Gpu.Engines
             int Width  = ReadRegister(NvGpuEngine3dReg.FrameBufferNWidth  + FbIndex * 0x10);
             int Height = ReadRegister(NvGpuEngine3dReg.FrameBufferNHeight + FbIndex * 0x10);
 
-            float TX = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNTranslateX + FbIndex * 4);
-            float TY = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNTranslateY + FbIndex * 4);
+            float TX = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNTranslateX + FbIndex * 8);
+            float TY = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNTranslateY + FbIndex * 8);
 
-            float SX = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNScaleX + FbIndex * 4);
-            float SY = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNScaleY + FbIndex * 4);
+            float SX = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNScaleX + FbIndex * 8);
+            float SY = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNScaleY + FbIndex * 8);
 
             int VpX = (int)MathF.Max(0, TX - MathF.Abs(SX));
             int VpY = (int)MathF.Max(0, TY - MathF.Abs(SY));
@@ -180,12 +193,48 @@ namespace Ryujinx.HLE.Gpu.Engines
             int VpW = (int)(TX + MathF.Abs(SX)) - VpX;
             int VpH = (int)(TY + MathF.Abs(SY)) - VpY;
 
-            Gpu.Renderer.FrameBuffer.Create(Key, Width, Height);
-            Gpu.Renderer.FrameBuffer.Bind(Key);
+            GalImageFormat ImageFormat = ImageFormatConverter.ConvertFrameBuffer((GalFrameBufferFormat)Format);
+
+            GalImage Image = new GalImage(Width, Height, ImageFormat);
+
+            long Size = TextureHelper.GetTextureSize(Image);
+
+            Gpu.Renderer.Texture.CreateFb(Key, Size, Image);
+            Gpu.Renderer.FrameBuffer.BindColor(Key, FbIndex);
 
             Gpu.Renderer.FrameBuffer.SetViewport(VpX, VpY, VpW, VpH);
         }
 
+        private void SetZeta(NvGpuVmm Vmm)
+        {
+            long ZA = MakeInt64From2xInt32(NvGpuEngine3dReg.ZetaAddress);
+
+            int Format = ReadRegister(NvGpuEngine3dReg.ZetaFormat);
+
+            bool ZetaEnable = (ReadRegister(NvGpuEngine3dReg.ZetaEnable) & 1) != 0;
+
+            if (ZA == 0 || Format == 0 || !ZetaEnable)
+            {
+                Gpu.Renderer.FrameBuffer.UnbindZeta();
+
+                return;
+            }
+
+            long Key = Vmm.GetPhysicalAddress(ZA);
+            
+            int Width  = ReadRegister(NvGpuEngine3dReg.ZetaHoriz);
+            int Height = ReadRegister(NvGpuEngine3dReg.ZetaVert);
+
+            GalImageFormat ImageFormat = ImageFormatConverter.ConvertZeta((GalZetaFormat)Format);
+
+            GalImage Image = new GalImage(Width, Height, ImageFormat);
+
+            long Size = TextureHelper.GetTextureSize(Image);
+
+            Gpu.Renderer.Texture.CreateFb(Key, Size, Image);
+            Gpu.Renderer.FrameBuffer.BindZeta(Key);
+        }
+
         private long[] UploadShaders(NvGpuVmm Vmm)
         {
             long[] Keys = new long[5];
@@ -442,15 +491,15 @@ namespace Ryujinx.HLE.Gpu.Engines
             }
             else
             {
-                GalTexture NewTexture = TextureFactory.MakeTexture(Vmm, TicPosition);
+                GalImage NewImage = TextureFactory.MakeTexture(Vmm, TicPosition);
 
-                long Size = (uint)TextureHelper.GetTextureSize(NewTexture);
+                long Size = (uint)TextureHelper.GetTextureSize(NewImage);
 
                 bool HasCachedTexture = false;
 
-                if (Gpu.Renderer.Texture.TryGetCachedTexture(Key, Size, out GalTexture Texture))
+                if (Gpu.Renderer.Texture.TryGetCachedTexture(Key, Size, out GalImage Image))
                 {
-                    if (NewTexture.Equals(Texture) && !QueryKeyUpload(Vmm, Key, Size, NvGpuBufferType.Texture))
+                    if (NewImage.Equals(Image) && !QueryKeyUpload(Vmm, Key, Size, NvGpuBufferType.Texture))
                     {
                         Gpu.Renderer.Texture.Bind(Key, TexIndex);
 
@@ -462,7 +511,7 @@ namespace Ryujinx.HLE.Gpu.Engines
                 {
                     byte[] Data = TextureFactory.GetTextureData(Vmm, TicPosition);
 
-                    Gpu.Renderer.Texture.Create(Key, Data, NewTexture);
+                    Gpu.Renderer.Texture.Create(Key, Data, NewImage);
                 }
 
                 Gpu.Renderer.Texture.Bind(Key, TexIndex);
diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs
index 39a5ee8c..b03aef02 100644
--- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs
+++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs
@@ -22,7 +22,14 @@ namespace Ryujinx.HLE.Gpu.Engines
         StencilBackFuncRef   = 0x3d5,
         StencilBackMask      = 0x3d6,
         StencilBackFuncMask  = 0x3d7,
+        ZetaAddress          = 0x3f8,
+        ZetaFormat           = 0x3fa,
+        ZetaBlockDimensions  = 0x3fb,
+        ZetaLayerStride      = 0x3fc,
         VertexAttribNFormat  = 0x458,
+        ZetaHoriz            = 0x48a,
+        ZetaVert             = 0x48b,
+        ZetaArrayMode        = 0x48c,
         DepthTestEnable      = 0x4b3,
         IBlendEnable         = 0x4b9,
         DepthTestFunction    = 0x4c3,
@@ -44,6 +51,7 @@ namespace Ryujinx.HLE.Gpu.Engines
         StencilFrontFuncMask = 0x4e6,
         StencilFrontMask     = 0x4e7,
         VertexArrayElemBase  = 0x50d,
+        ZetaEnable           = 0x54e,
         TexHeaderPoolOffset  = 0x55d,
         TexSamplerPoolOffset = 0x557,
         StencilTwoSideEnable = 0x565,
diff --git a/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs b/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs
index 4db0b6f1..0ef33d3b 100644
--- a/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs
+++ b/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs
@@ -6,11 +6,16 @@ namespace Ryujinx.HLE.Gpu.Texture
 {
     static class TextureFactory
     {
-        public static GalTexture MakeTexture(NvGpuVmm Vmm, long TicPosition)
+        public static GalImage MakeTexture(NvGpuVmm Vmm, long TicPosition)
         {
             int[] Tic = ReadWords(Vmm, TicPosition, 8);
 
-            GalTextureFormat Format = (GalTextureFormat)(Tic[0] & 0x7f);
+            GalTextureType RType = (GalTextureType)((Tic[0] >> 7)  & 7);
+            GalTextureType GType = (GalTextureType)((Tic[0] >> 10) & 7);
+            GalTextureType BType = (GalTextureType)((Tic[0] >> 13) & 7);
+            GalTextureType AType = (GalTextureType)((Tic[0] >> 16) & 7);
+
+            GalImageFormat Format = ImageFormatConverter.ConvertTexture((GalTextureFormat)(Tic[0] & 0x7f), RType, GType, BType, AType);
 
             GalTextureSource XSource = (GalTextureSource)((Tic[0] >> 19) & 7);
             GalTextureSource YSource = (GalTextureSource)((Tic[0] >> 22) & 7);
@@ -20,7 +25,7 @@ namespace Ryujinx.HLE.Gpu.Texture
             int Width  = (Tic[4] & 0xffff) + 1;
             int Height = (Tic[5] & 0xffff) + 1;
 
-            return new GalTexture(
+            return new GalImage(
                 Width,
                 Height,
                 Format,
diff --git a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs
index 10a64f36..92b608a9 100644
--- a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs
+++ b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs
@@ -30,117 +30,151 @@ namespace Ryujinx.HLE.Gpu.Texture
             throw new NotImplementedException(Texture.Swizzle.ToString());
         }
 
-        public static int GetTextureSize(GalTexture Texture)
+        public static int GetTextureSize(GalImage Image)
         {
-            switch (Texture.Format)
+            switch (Image.Format)
             {
-                case GalTextureFormat.R32G32B32A32:
-                    return Texture.Width * Texture.Height * 16;
-
-                case GalTextureFormat.R16G16B16A16:
-                    return Texture.Width * Texture.Height * 8;
-
-                case GalTextureFormat.A8B8G8R8:
-                case GalTextureFormat.A2B10G10R10:
-                case GalTextureFormat.R32:
-                case GalTextureFormat.ZF32:
-                case GalTextureFormat.BF10GF11RF11:
-                case GalTextureFormat.Z24S8:
-                    return Texture.Width * Texture.Height * 4;
-
-                case GalTextureFormat.A1B5G5R5:
-                case GalTextureFormat.B5G6R5:
-                case GalTextureFormat.G8R8:
-                case GalTextureFormat.R16:
-                    return Texture.Width * Texture.Height * 2;
-
-                case GalTextureFormat.R8:
-                    return Texture.Width * Texture.Height;
-
-                case GalTextureFormat.BC1:
-                case GalTextureFormat.BC4:
+                case GalImageFormat.R32G32B32A32_SFLOAT:
+                case GalImageFormat.R32G32B32A32_SINT:
+                case GalImageFormat.R32G32B32A32_UINT:
+                    return Image.Width * Image.Height * 16;
+
+                case GalImageFormat.R16G16B16A16_SFLOAT:
+                case GalImageFormat.R16G16B16A16_SINT:
+                case GalImageFormat.R16G16B16A16_SNORM:
+                case GalImageFormat.R16G16B16A16_UINT:
+                case GalImageFormat.R16G16B16A16_UNORM:
+                    return Image.Width * Image.Height * 8;
+
+                case GalImageFormat.A8B8G8R8_SINT_PACK32:
+                case GalImageFormat.A8B8G8R8_SNORM_PACK32:
+                case GalImageFormat.A8B8G8R8_UINT_PACK32:
+                case GalImageFormat.A8B8G8R8_UNORM_PACK32:
+                case GalImageFormat.A8B8G8R8_SRGB_PACK32:
+                case GalImageFormat.A2B10G10R10_SINT_PACK32:
+                case GalImageFormat.A2B10G10R10_SNORM_PACK32:
+                case GalImageFormat.A2B10G10R10_UINT_PACK32:
+                case GalImageFormat.A2B10G10R10_UNORM_PACK32:
+                case GalImageFormat.R16G16_SFLOAT:
+                case GalImageFormat.R16G16_SINT:
+                case GalImageFormat.R16G16_SNORM:
+                case GalImageFormat.R16G16_UINT:
+                case GalImageFormat.R16G16_UNORM:
+                case GalImageFormat.R32_SFLOAT:
+                case GalImageFormat.R32_SINT:
+                case GalImageFormat.R32_UINT:
+                case GalImageFormat.D32_SFLOAT:
+                case GalImageFormat.B10G11R11_UFLOAT_PACK32:
+                case GalImageFormat.D24_UNORM_S8_UINT:
+                    return Image.Width * Image.Height * 4;
+
+                case GalImageFormat.B4G4R4A4_UNORM_PACK16:
+                case GalImageFormat.A1R5G5B5_UNORM_PACK16:
+                case GalImageFormat.B5G6R5_UNORM_PACK16:
+                case GalImageFormat.R8G8_SINT:
+                case GalImageFormat.R8G8_SNORM:
+                case GalImageFormat.R8G8_UINT:
+                case GalImageFormat.R8G8_UNORM:
+                case GalImageFormat.R16_SFLOAT:
+                case GalImageFormat.R16_SINT:
+                case GalImageFormat.R16_SNORM:
+                case GalImageFormat.R16_UINT:
+                case GalImageFormat.R16_UNORM:
+                case GalImageFormat.D16_UNORM:
+                    return Image.Width * Image.Height * 2;
+
+                case GalImageFormat.R8_SINT:
+                case GalImageFormat.R8_SNORM:
+                case GalImageFormat.R8_UINT:
+                case GalImageFormat.R8_UNORM:
+                    return Image.Width * Image.Height;
+
+                case GalImageFormat.BC1_RGBA_UNORM_BLOCK:
+                case GalImageFormat.BC4_SNORM_BLOCK:
+                case GalImageFormat.BC4_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 4, 4, 8);
+                    return CompressedTextureSize(Image.Width, Image.Height, 4, 4, 8);
                 }
 
-                case GalTextureFormat.BC6H_SF16:
-                case GalTextureFormat.BC6H_UF16:
-                case GalTextureFormat.BC7U:
-                case GalTextureFormat.BC2:
-                case GalTextureFormat.BC3:
-                case GalTextureFormat.BC5:
-                case GalTextureFormat.Astc2D4x4:
+                case GalImageFormat.BC6H_SFLOAT_BLOCK:
+                case GalImageFormat.BC6H_UFLOAT_BLOCK:
+                case GalImageFormat.BC7_UNORM_BLOCK:
+                case GalImageFormat.BC2_UNORM_BLOCK:
+                case GalImageFormat.BC3_UNORM_BLOCK:
+                case GalImageFormat.BC5_SNORM_BLOCK:
+                case GalImageFormat.BC5_UNORM_BLOCK:
+                case GalImageFormat.ASTC_4x4_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 4, 4, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 4, 4, 16);
                 }
 
-                case GalTextureFormat.Astc2D5x5:
+                case GalImageFormat.ASTC_5x5_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 5, 5, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 5, 5, 16);
                 }
 
-                case GalTextureFormat.Astc2D6x6:
+                case GalImageFormat.ASTC_6x6_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 6, 6, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 6, 6, 16);
                 }
 
-                case GalTextureFormat.Astc2D8x8:
+                case GalImageFormat.ASTC_8x8_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 8, 8, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 8, 8, 16);
                 }
 
-                case GalTextureFormat.Astc2D10x10:
+                case GalImageFormat.ASTC_10x10_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 10, 10, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 10, 10, 16);
                 }
 
-                case GalTextureFormat.Astc2D12x12:
+                case GalImageFormat.ASTC_12x12_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 12, 12, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 12, 12, 16);
                 }
 
-                case GalTextureFormat.Astc2D5x4:
+                case GalImageFormat.ASTC_5x4_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 5, 4, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 5, 4, 16);
                 }
 
-                case GalTextureFormat.Astc2D6x5:
+                case GalImageFormat.ASTC_6x5_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 6, 5, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 6, 5, 16);
                 }
 
-                case GalTextureFormat.Astc2D8x6:
+                case GalImageFormat.ASTC_8x6_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 8, 6, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 8, 6, 16);
                 }
 
-                case GalTextureFormat.Astc2D10x8:
+                case GalImageFormat.ASTC_10x8_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 10, 8, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 10, 8, 16);
                 }
 
-                case GalTextureFormat.Astc2D12x10:
+                case GalImageFormat.ASTC_12x10_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 12, 10, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 12, 10, 16);
                 }
 
-                case GalTextureFormat.Astc2D8x5:
+                case GalImageFormat.ASTC_8x5_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 8, 5, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 8, 5, 16);
                 }
 
-                case GalTextureFormat.Astc2D10x5:
+                case GalImageFormat.ASTC_10x5_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 10, 5, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 10, 5, 16);
                 }
 
-                case GalTextureFormat.Astc2D10x6:
+                case GalImageFormat.ASTC_10x6_UNORM_BLOCK:
                 {
-                    return CompressedTextureSize(Texture.Width, Texture.Height, 10, 6, 16);
+                    return CompressedTextureSize(Image.Width, Image.Height, 10, 6, 16);
                 }
             }
 
-            throw new NotImplementedException("0x" + Texture.Format.ToString("x2"));
+            throw new NotImplementedException("0x" + Image.Format.ToString("x2"));
         }
 
         public static int CompressedTextureSize(int TextureWidth, int TextureHeight, int BlockWidth, int BlockHeight, int Bpb)
diff --git a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs
index 0cf055db..19aa25d7 100644
--- a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs
+++ b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs
@@ -19,6 +19,7 @@ namespace Ryujinx.HLE.Gpu.Texture
                 case GalTextureFormat.Z24S8:        return Read4Bpp                  (Memory, Texture);
                 case GalTextureFormat.A1B5G5R5:     return Read5551                  (Memory, Texture);
                 case GalTextureFormat.B5G6R5:       return Read565                   (Memory, Texture);
+                case GalTextureFormat.A4B4G4R4:     return Read2Bpp                  (Memory, Texture);
                 case GalTextureFormat.G8R8:         return Read2Bpp                  (Memory, Texture);
                 case GalTextureFormat.R16:          return Read2Bpp                  (Memory, Texture);
                 case GalTextureFormat.R8:           return Read1Bpp                  (Memory, Texture);