From 25dd5f4238d898120f2f65c4d5d5b9c192ce1e10 Mon Sep 17 00:00:00 2001
From: ReinUsesLisp <reinuseslisp@airmail.cc>
Date: Fri, 10 Aug 2018 01:09:40 -0300
Subject: [PATCH] Low level graphics API prerequisites (#319)

* Add GalPipelineState and IGalPipeline

* Separate UploadVertex call

* Add ConstBuffer cache

* Move Vertex Assembly into GalPipelineState

* Move Uniform binds to GalPipelineState

* Move framebuffer flip into a buffer

* Rebase

* Fix regression

* Move clear values from VertexEndGl to ClearBuffers

* Rename obscure names O->Old S->New
---
 Ryujinx.Graphics/Gal/GalPipelineState.cs      |  73 ++++
 Ryujinx.Graphics/Gal/IGalBlend.cs             |  22 -
 Ryujinx.Graphics/Gal/IGalConstBuffer.cs       |  16 +
 Ryujinx.Graphics/Gal/IGalPipeline.cs          |   7 +
 Ryujinx.Graphics/Gal/IGalRasterizer.cs        |  42 +-
 Ryujinx.Graphics/Gal/IGalRenderer.cs          |   4 +-
 Ryujinx.Graphics/Gal/IGalShader.cs            |   4 -
 Ryujinx.Graphics/Gal/OpenGL/OGLBlend.cs       |  49 ---
 Ryujinx.Graphics/Gal/OpenGL/OGLConstBuffer.cs |  66 +++
 Ryujinx.Graphics/Gal/OpenGL/OGLPipeline.cs    | 379 ++++++++++++++++++
 Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs  | 213 +---------
 Ryujinx.Graphics/Gal/OpenGL/OGLRenderer.cs    |  10 +-
 Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs      | 224 +++--------
 .../Gal/OpenGL/OGLShaderProgram.cs            |  86 ++++
 .../Gal/OpenGL/OGLStreamBuffer.cs             |  10 +-
 Ryujinx.Graphics/Gal/Shader/GlslDecl.cs       |   1 +
 Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs |   6 +-
 Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs      | 342 ++++++----------
 Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs   |   1 +
 Ryujinx.HLE/Gpu/Memory/NvGpuBufferType.cs     |   1 +
 20 files changed, 854 insertions(+), 702 deletions(-)
 create mode 100644 Ryujinx.Graphics/Gal/GalPipelineState.cs
 delete mode 100644 Ryujinx.Graphics/Gal/IGalBlend.cs
 create mode 100644 Ryujinx.Graphics/Gal/IGalConstBuffer.cs
 create mode 100644 Ryujinx.Graphics/Gal/IGalPipeline.cs
 delete mode 100644 Ryujinx.Graphics/Gal/OpenGL/OGLBlend.cs
 create mode 100644 Ryujinx.Graphics/Gal/OpenGL/OGLConstBuffer.cs
 create mode 100644 Ryujinx.Graphics/Gal/OpenGL/OGLPipeline.cs
 create mode 100644 Ryujinx.Graphics/Gal/OpenGL/OGLShaderProgram.cs

diff --git a/Ryujinx.Graphics/Gal/GalPipelineState.cs b/Ryujinx.Graphics/Gal/GalPipelineState.cs
new file mode 100644
index 00000000..d1ffbe76
--- /dev/null
+++ b/Ryujinx.Graphics/Gal/GalPipelineState.cs
@@ -0,0 +1,73 @@
+namespace Ryujinx.Graphics.Gal
+{
+    public struct GalVertexBinding
+    {
+        //VboKey shouldn't be here, but ARB_vertex_attrib_binding is core since 4.3
+
+        public bool Enabled;
+        public int Stride;
+        public long VboKey;
+        public GalVertexAttrib[] Attribs;
+    }
+
+    public class GalPipelineState
+    {
+        public const int Stages = 5;
+        public const int ConstBuffersPerStage = 18;
+
+        public long[][] ConstBufferKeys;
+
+        public GalVertexBinding[] VertexBindings;
+
+        public float FlipX;
+        public float FlipY;
+
+        public GalFrontFace FrontFace;
+
+        public bool CullFaceEnabled;
+        public GalCullFace CullFace;
+
+        public bool DepthTestEnabled;
+        public GalComparisonOp DepthFunc;
+
+        public bool StencilTestEnabled;
+
+        public GalComparisonOp StencilBackFuncFunc;
+        public int StencilBackFuncRef;
+        public uint StencilBackFuncMask;
+        public GalStencilOp StencilBackOpFail;
+        public GalStencilOp StencilBackOpZFail;
+        public GalStencilOp StencilBackOpZPass;
+        public uint StencilBackMask;
+
+        public GalComparisonOp StencilFrontFuncFunc;
+        public int StencilFrontFuncRef;
+        public uint StencilFrontFuncMask;
+        public GalStencilOp StencilFrontOpFail;
+        public GalStencilOp StencilFrontOpZFail;
+        public GalStencilOp StencilFrontOpZPass;
+        public uint StencilFrontMask;
+
+        public bool BlendEnabled;
+        public bool BlendSeparateAlpha;
+        public GalBlendEquation BlendEquationRgb;
+        public GalBlendFactor BlendFuncSrcRgb;
+        public GalBlendFactor BlendFuncDstRgb;
+        public GalBlendEquation BlendEquationAlpha;
+        public GalBlendFactor BlendFuncSrcAlpha;
+        public GalBlendFactor BlendFuncDstAlpha;
+
+        public bool PrimitiveRestartEnabled;
+        public uint PrimitiveRestartIndex;
+
+        public GalPipelineState()
+        {
+            ConstBufferKeys = new long[Stages][];
+
+            for (int Stage = 0; Stage < Stages; Stage++)
+            {
+                ConstBufferKeys[Stage] = new long[ConstBuffersPerStage];
+            }
+        }
+    }
+}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/IGalBlend.cs b/Ryujinx.Graphics/Gal/IGalBlend.cs
deleted file mode 100644
index 5c96a492..00000000
--- a/Ryujinx.Graphics/Gal/IGalBlend.cs
+++ /dev/null
@@ -1,22 +0,0 @@
-namespace Ryujinx.Graphics.Gal
-{
-    public interface IGalBlend
-    {
-        void Enable();
-
-        void Disable();
-
-        void Set(
-            GalBlendEquation Equation,
-            GalBlendFactor   FuncSrc,
-            GalBlendFactor   FuncDst);
-
-        void SetSeparate(
-            GalBlendEquation EquationRgb,
-            GalBlendEquation EquationAlpha,
-            GalBlendFactor   FuncSrcRgb,
-            GalBlendFactor   FuncDstRgb,
-            GalBlendFactor   FuncSrcAlpha,
-            GalBlendFactor   FuncDstAlpha);
-    }
-}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/IGalConstBuffer.cs b/Ryujinx.Graphics/Gal/IGalConstBuffer.cs
new file mode 100644
index 00000000..37545b2a
--- /dev/null
+++ b/Ryujinx.Graphics/Gal/IGalConstBuffer.cs
@@ -0,0 +1,16 @@
+using System;
+
+namespace Ryujinx.Graphics.Gal
+{
+    public interface IGalConstBuffer
+    {
+        void LockCache();
+        void UnlockCache();
+
+        void Create(long Key, long Size);
+
+        bool IsCached(long Key, long Size);
+
+        void SetData(long Key, long Size, IntPtr HostAddress);
+    }
+}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/IGalPipeline.cs b/Ryujinx.Graphics/Gal/IGalPipeline.cs
new file mode 100644
index 00000000..d8cf266a
--- /dev/null
+++ b/Ryujinx.Graphics/Gal/IGalPipeline.cs
@@ -0,0 +1,7 @@
+namespace Ryujinx.Graphics.Gal
+{
+    public interface IGalPipeline
+    {
+        void Bind(GalPipelineState State);
+    }
+}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/IGalRasterizer.cs b/Ryujinx.Graphics/Gal/IGalRasterizer.cs
index a87d36c3..89e50b1f 100644
--- a/Ryujinx.Graphics/Gal/IGalRasterizer.cs
+++ b/Ryujinx.Graphics/Gal/IGalRasterizer.cs
@@ -7,52 +7,20 @@ namespace Ryujinx.Graphics.Gal
         void LockCaches();
         void UnlockCaches();
 
-        void ClearBuffers(GalClearBufferFlags Flags);
+        void ClearBuffers(
+            GalClearBufferFlags Flags,
+            float Red, float Green, float Blue, float Alpha,
+            float Depth,
+            int Stencil);
 
         bool IsVboCached(long Key, long DataSize);
 
         bool IsIboCached(long Key, long DataSize);
 
-        void SetFrontFace(GalFrontFace FrontFace);
-
-        void EnableCullFace();
-
-        void DisableCullFace();
-
-        void SetCullFace(GalCullFace CullFace);
-
-        void EnableDepthTest();
-
-        void DisableDepthTest();
-
-        void SetDepthFunction(GalComparisonOp Func);
-
-        void SetClearDepth(float Depth);
-
-        void EnableStencilTest();
-
-        void DisableStencilTest();
-
-        void SetStencilFunction(bool IsFrontFace, GalComparisonOp Func, int Ref, int Mask);
-
-        void SetStencilOp(bool IsFrontFace, GalStencilOp Fail, GalStencilOp ZFail, GalStencilOp ZPass);
-
-        void SetStencilMask(bool IsFrontFace, int Mask);
-
-        void SetClearStencil(int Stencil);
-
-        void EnablePrimitiveRestart();
-
-        void DisablePrimitiveRestart();
-
-        void SetPrimitiveRestartIndex(uint Index);
-
         void CreateVbo(long Key, int DataSize, IntPtr HostAddress);
 
         void CreateIbo(long Key, int DataSize, IntPtr HostAddress);
 
-        void SetVertexArray(int Stride, long VboKey, GalVertexAttrib[] Attribs);
-
         void SetIndexArray(int Size, GalIndexFormat Format);
 
         void DrawArrays(int First, int PrimCount, GalPrimitiveType PrimType);
diff --git a/Ryujinx.Graphics/Gal/IGalRenderer.cs b/Ryujinx.Graphics/Gal/IGalRenderer.cs
index c6324c4a..b47ac71d 100644
--- a/Ryujinx.Graphics/Gal/IGalRenderer.cs
+++ b/Ryujinx.Graphics/Gal/IGalRenderer.cs
@@ -8,7 +8,7 @@ namespace Ryujinx.Graphics.Gal
 
         void RunActions();
 
-        IGalBlend Blend { get; }
+        IGalConstBuffer Buffer { get; }
 
         IGalFrameBuffer FrameBuffer { get; }
 
@@ -16,6 +16,8 @@ namespace Ryujinx.Graphics.Gal
 
         IGalShader Shader { get; }
 
+        IGalPipeline Pipeline { get; }
+
         IGalTexture Texture { get; }
     }
 }
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/IGalShader.cs b/Ryujinx.Graphics/Gal/IGalShader.cs
index 56235a07..a9bd1381 100644
--- a/Ryujinx.Graphics/Gal/IGalShader.cs
+++ b/Ryujinx.Graphics/Gal/IGalShader.cs
@@ -11,12 +11,8 @@ namespace Ryujinx.Graphics.Gal
 
         IEnumerable<ShaderDeclInfo> GetTextureUsage(long Key);
 
-        void SetConstBuffer(long Key, int Cbuf, int DataSize, IntPtr HostAddress);
-
         void EnsureTextureBinding(string UniformName, int Value);
 
-        void SetFlip(float X, float Y);
-
         void Bind(long Key);
 
         void Unbind(GalShaderType Type);
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLBlend.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLBlend.cs
deleted file mode 100644
index 7175e3a0..00000000
--- a/Ryujinx.Graphics/Gal/OpenGL/OGLBlend.cs
+++ /dev/null
@@ -1,49 +0,0 @@
-using OpenTK.Graphics.OpenGL;
-
-namespace Ryujinx.Graphics.Gal.OpenGL
-{
-    public class OGLBlend : IGalBlend
-    {
-        public void Enable()
-        {
-            GL.Enable(EnableCap.Blend);
-        }
-
-        public void Disable()
-        {
-            GL.Disable(EnableCap.Blend);
-        }
-
-        public void Set(
-            GalBlendEquation Equation,
-            GalBlendFactor   FuncSrc,
-            GalBlendFactor   FuncDst)
-        {
-            GL.BlendEquation(
-                OGLEnumConverter.GetBlendEquation(Equation));
-
-            GL.BlendFunc(
-                OGLEnumConverter.GetBlendFactor(FuncSrc),
-                OGLEnumConverter.GetBlendFactor(FuncDst));
-        }
-
-        public void SetSeparate(
-            GalBlendEquation EquationRgb,
-            GalBlendEquation EquationAlpha,
-            GalBlendFactor   FuncSrcRgb,
-            GalBlendFactor   FuncDstRgb,
-            GalBlendFactor   FuncSrcAlpha,
-            GalBlendFactor   FuncDstAlpha)
-        {
-            GL.BlendEquationSeparate(
-                OGLEnumConverter.GetBlendEquation(EquationRgb),
-                OGLEnumConverter.GetBlendEquation(EquationAlpha));
-
-            GL.BlendFuncSeparate(
-                (BlendingFactorSrc)OGLEnumConverter.GetBlendFactor(FuncSrcRgb),
-                (BlendingFactorDest)OGLEnumConverter.GetBlendFactor(FuncDstRgb),
-                (BlendingFactorSrc)OGLEnumConverter.GetBlendFactor(FuncSrcAlpha),
-                (BlendingFactorDest)OGLEnumConverter.GetBlendFactor(FuncDstAlpha));
-        }
-    }
-}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLConstBuffer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLConstBuffer.cs
new file mode 100644
index 00000000..50825541
--- /dev/null
+++ b/Ryujinx.Graphics/Gal/OpenGL/OGLConstBuffer.cs
@@ -0,0 +1,66 @@
+using OpenTK.Graphics.OpenGL;
+using System;
+
+namespace Ryujinx.Graphics.Gal.OpenGL
+{
+    class OGLConstBuffer : IGalConstBuffer
+    {
+        private OGLCachedResource<OGLStreamBuffer> Cache;
+
+        public OGLConstBuffer()
+        {
+            Cache = new OGLCachedResource<OGLStreamBuffer>(DeleteBuffer);
+        }
+
+        public void LockCache()
+        {
+            Cache.Lock();
+        }
+
+        public void UnlockCache()
+        {
+            Cache.Unlock();
+        }
+
+        public void Create(long Key, long Size)
+        {
+            OGLStreamBuffer Buffer = new OGLStreamBuffer(BufferTarget.UniformBuffer, Size);
+
+            Cache.AddOrUpdate(Key, Buffer, Size);
+        }
+
+        public bool IsCached(long Key, long Size)
+        {
+            return Cache.TryGetSize(Key, out long CachedSize) && CachedSize == Size;
+        }
+
+        public void SetData(long Key, long Size, IntPtr HostAddress)
+        {
+            if (!Cache.TryGetValue(Key, out OGLStreamBuffer Buffer))
+            {
+                throw new InvalidOperationException();
+            }
+
+            Buffer.SetData(Size, HostAddress);
+        }
+
+        public bool TryGetUbo(long Key, out int UboHandle)
+        {
+            if (Cache.TryGetValue(Key, out OGLStreamBuffer Buffer))
+            {
+                UboHandle = Buffer.Handle;
+
+                return true;
+            }
+
+            UboHandle = 0;
+
+            return false;
+        }
+
+        private static void DeleteBuffer(OGLStreamBuffer Buffer)
+        {
+            Buffer.Dispose();
+        }
+    }
+}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLPipeline.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLPipeline.cs
new file mode 100644
index 00000000..54f984cd
--- /dev/null
+++ b/Ryujinx.Graphics/Gal/OpenGL/OGLPipeline.cs
@@ -0,0 +1,379 @@
+using OpenTK.Graphics.OpenGL;
+using System;
+using System.Collections.Generic;
+
+namespace Ryujinx.Graphics.Gal.OpenGL
+{
+    class OGLPipeline : IGalPipeline
+    {
+        private static Dictionary<GalVertexAttribSize, int> AttribElements =
+                   new Dictionary<GalVertexAttribSize, int>()
+        {
+            { GalVertexAttribSize._32_32_32_32, 4 },
+            { GalVertexAttribSize._32_32_32,    3 },
+            { GalVertexAttribSize._16_16_16_16, 4 },
+            { GalVertexAttribSize._32_32,       2 },
+            { GalVertexAttribSize._16_16_16,    3 },
+            { GalVertexAttribSize._8_8_8_8,     4 },
+            { GalVertexAttribSize._16_16,       2 },
+            { GalVertexAttribSize._32,          1 },
+            { GalVertexAttribSize._8_8_8,       3 },
+            { GalVertexAttribSize._8_8,         2 },
+            { GalVertexAttribSize._16,          1 },
+            { GalVertexAttribSize._8,           1 },
+            { GalVertexAttribSize._10_10_10_2,  4 },
+            { GalVertexAttribSize._11_11_10,    3 }
+        };
+
+        private static Dictionary<GalVertexAttribSize, VertexAttribPointerType> AttribTypes =
+                   new Dictionary<GalVertexAttribSize, VertexAttribPointerType>()
+        {
+            { GalVertexAttribSize._32_32_32_32, VertexAttribPointerType.Int   },
+            { GalVertexAttribSize._32_32_32,    VertexAttribPointerType.Int   },
+            { GalVertexAttribSize._16_16_16_16, VertexAttribPointerType.Short },
+            { GalVertexAttribSize._32_32,       VertexAttribPointerType.Int   },
+            { GalVertexAttribSize._16_16_16,    VertexAttribPointerType.Short },
+            { GalVertexAttribSize._8_8_8_8,     VertexAttribPointerType.Byte  },
+            { GalVertexAttribSize._16_16,       VertexAttribPointerType.Short },
+            { GalVertexAttribSize._32,          VertexAttribPointerType.Int   },
+            { GalVertexAttribSize._8_8_8,       VertexAttribPointerType.Byte  },
+            { GalVertexAttribSize._8_8,         VertexAttribPointerType.Byte  },
+            { GalVertexAttribSize._16,          VertexAttribPointerType.Short },
+            { GalVertexAttribSize._8,           VertexAttribPointerType.Byte  },
+            { GalVertexAttribSize._10_10_10_2,  VertexAttribPointerType.Int   }, //?
+            { GalVertexAttribSize._11_11_10,    VertexAttribPointerType.Int   }  //?
+        };
+
+        private GalPipelineState Old;
+
+        private OGLConstBuffer Buffer;
+        private OGLRasterizer Rasterizer;
+        private OGLShader Shader;
+
+        private int VaoHandle;
+
+        public OGLPipeline(OGLConstBuffer Buffer, OGLRasterizer Rasterizer, OGLShader Shader)
+        {
+            this.Buffer     = Buffer;
+            this.Rasterizer = Rasterizer;
+            this.Shader     = Shader;
+
+            //These values match OpenGL's defaults
+            Old = new GalPipelineState
+            {
+                FrontFace = GalFrontFace.CCW,
+
+                CullFaceEnabled = false,
+                CullFace = GalCullFace.Back,
+
+                DepthTestEnabled = false,
+                DepthFunc = GalComparisonOp.Less,
+
+                StencilTestEnabled = false,
+
+                StencilBackFuncFunc = GalComparisonOp.Always,
+                StencilBackFuncRef = 0,
+                StencilBackFuncMask = UInt32.MaxValue,
+                StencilBackOpFail = GalStencilOp.Keep,
+                StencilBackOpZFail = GalStencilOp.Keep,
+                StencilBackOpZPass = GalStencilOp.Keep,
+                StencilBackMask = UInt32.MaxValue,
+
+                StencilFrontFuncFunc = GalComparisonOp.Always,
+                StencilFrontFuncRef = 0,
+                StencilFrontFuncMask = UInt32.MaxValue,
+                StencilFrontOpFail = GalStencilOp.Keep,
+                StencilFrontOpZFail = GalStencilOp.Keep,
+                StencilFrontOpZPass = GalStencilOp.Keep,
+                StencilFrontMask = UInt32.MaxValue,
+
+                BlendEnabled = false,
+                BlendSeparateAlpha = false,
+
+                BlendEquationRgb = 0,
+                BlendFuncSrcRgb = GalBlendFactor.One,
+                BlendFuncDstRgb = GalBlendFactor.Zero,
+                BlendEquationAlpha = 0,
+                BlendFuncSrcAlpha = GalBlendFactor.One,
+                BlendFuncDstAlpha = GalBlendFactor.Zero,
+
+                PrimitiveRestartEnabled = false,
+                PrimitiveRestartIndex = 0
+            };
+        }
+
+        public void Bind(GalPipelineState New)
+        {
+            BindConstBuffers(New);
+
+            BindVertexLayout(New);
+
+            if (New.FlipX != Old.FlipX || New.FlipY != Old.FlipY)
+            {
+                Shader.SetFlip(New.FlipX, New.FlipY);
+            }
+
+            //Note: Uncomment SetFrontFace and SetCullFace when flipping issues are solved
+
+            //if (New.FrontFace != O.FrontFace)
+            //{
+            //    GL.FrontFace(OGLEnumConverter.GetFrontFace(New.FrontFace));
+            //}
+
+            //if (New.CullFaceEnabled != O.CullFaceEnabled)
+            //{
+            //    Enable(EnableCap.CullFace, New.CullFaceEnabled);
+            //}
+
+            //if (New.CullFaceEnabled)
+            //{
+            //    if (New.CullFace != O.CullFace)
+            //    {
+            //        GL.CullFace(OGLEnumConverter.GetCullFace(New.CullFace));
+            //    }
+            //}
+
+            if (New.DepthTestEnabled != Old.DepthTestEnabled)
+            {
+                Enable(EnableCap.DepthTest, New.DepthTestEnabled);
+            }
+
+            if (New.DepthTestEnabled)
+            {
+                if (New.DepthFunc != Old.DepthFunc)
+                {
+                    GL.DepthFunc(OGLEnumConverter.GetDepthFunc(New.DepthFunc));
+                }
+            }
+
+            if (New.StencilTestEnabled != Old.StencilTestEnabled)
+            {
+                Enable(EnableCap.StencilTest, New.StencilTestEnabled);
+            }
+
+            if (New.StencilTestEnabled)
+            {
+                if (New.StencilBackFuncFunc != Old.StencilBackFuncFunc ||
+                    New.StencilBackFuncRef  != Old.StencilBackFuncRef  ||
+                    New.StencilBackFuncMask != Old.StencilBackFuncMask)
+                {
+                    GL.StencilFuncSeparate(
+                        StencilFace.Back,
+                        OGLEnumConverter.GetStencilFunc(New.StencilBackFuncFunc),
+                        New.StencilBackFuncRef,
+                        New.StencilBackFuncMask);
+                }
+
+                if (New.StencilBackOpFail  != Old.StencilBackOpFail  ||
+                    New.StencilBackOpZFail != Old.StencilBackOpZFail ||
+                    New.StencilBackOpZPass != Old.StencilBackOpZPass)
+                {
+                    GL.StencilOpSeparate(
+                        StencilFace.Back,
+                        OGLEnumConverter.GetStencilOp(New.StencilBackOpFail),
+                        OGLEnumConverter.GetStencilOp(New.StencilBackOpZFail),
+                        OGLEnumConverter.GetStencilOp(New.StencilBackOpZPass));
+                }
+
+                if (New.StencilBackMask != Old.StencilBackMask)
+                {
+                    GL.StencilMaskSeparate(StencilFace.Back, New.StencilBackMask);
+                }
+
+                if (New.StencilFrontFuncFunc != Old.StencilFrontFuncFunc ||
+                    New.StencilFrontFuncRef  != Old.StencilFrontFuncRef  ||
+                    New.StencilFrontFuncMask != Old.StencilFrontFuncMask)
+                {
+                    GL.StencilFuncSeparate(
+                        StencilFace.Front,
+                        OGLEnumConverter.GetStencilFunc(New.StencilFrontFuncFunc),
+                        New.StencilFrontFuncRef,
+                        New.StencilFrontFuncMask);
+                }
+
+                if (New.StencilFrontOpFail  != Old.StencilFrontOpFail  ||
+                    New.StencilFrontOpZFail != Old.StencilFrontOpZFail ||
+                    New.StencilFrontOpZPass != Old.StencilFrontOpZPass)
+                {
+                    GL.StencilOpSeparate(
+                        StencilFace.Front,
+                        OGLEnumConverter.GetStencilOp(New.StencilFrontOpFail),
+                        OGLEnumConverter.GetStencilOp(New.StencilFrontOpZFail),
+                        OGLEnumConverter.GetStencilOp(New.StencilFrontOpZPass));
+                }
+
+                if (New.StencilFrontMask != Old.StencilFrontMask)
+                {
+                    GL.StencilMaskSeparate(StencilFace.Front, New.StencilFrontMask);
+                }
+            }
+
+            if (New.BlendEnabled != Old.BlendEnabled)
+            {
+                Enable(EnableCap.Blend, New.BlendEnabled);
+            }
+
+            if (New.BlendEnabled)
+            {
+                if (New.BlendSeparateAlpha)
+                {
+                    if (New.BlendEquationRgb   != Old.BlendEquationRgb ||
+                        New.BlendEquationAlpha != Old.BlendEquationAlpha)
+                    {
+                        GL.BlendEquationSeparate(
+                            OGLEnumConverter.GetBlendEquation(New.BlendEquationRgb),
+                            OGLEnumConverter.GetBlendEquation(New.BlendEquationAlpha));
+                    }
+
+                    if (New.BlendFuncSrcRgb   != Old.BlendFuncSrcRgb   ||
+                        New.BlendFuncDstRgb   != Old.BlendFuncDstRgb   ||
+                        New.BlendFuncSrcAlpha != Old.BlendFuncSrcAlpha ||
+                        New.BlendFuncDstAlpha != Old.BlendFuncDstAlpha)
+                    {
+                        GL.BlendFuncSeparate(
+                            (BlendingFactorSrc) OGLEnumConverter.GetBlendFactor(New.BlendFuncSrcRgb),
+                            (BlendingFactorDest)OGLEnumConverter.GetBlendFactor(New.BlendFuncDstRgb),
+                            (BlendingFactorSrc) OGLEnumConverter.GetBlendFactor(New.BlendFuncSrcAlpha),
+                            (BlendingFactorDest)OGLEnumConverter.GetBlendFactor(New.BlendFuncDstAlpha));
+                    }
+                }
+                else
+                {
+                    if (New.BlendEquationRgb != Old.BlendEquationRgb)
+                    {
+                        GL.BlendEquation(OGLEnumConverter.GetBlendEquation(New.BlendEquationRgb));
+                    }
+
+                    if (New.BlendFuncSrcRgb != Old.BlendFuncSrcRgb ||
+                        New.BlendFuncDstRgb != Old.BlendFuncDstRgb)
+                    {
+                        GL.BlendFunc(
+                            OGLEnumConverter.GetBlendFactor(New.BlendFuncSrcRgb),
+                            OGLEnumConverter.GetBlendFactor(New.BlendFuncDstRgb));
+                    }
+                }
+            }
+
+            if (New.PrimitiveRestartEnabled != Old.PrimitiveRestartEnabled)
+            {
+                Enable(EnableCap.PrimitiveRestart, New.PrimitiveRestartEnabled);
+            }
+
+            if (New.PrimitiveRestartEnabled)
+            {
+                if (New.PrimitiveRestartIndex != Old.PrimitiveRestartIndex)
+                {
+                    GL.PrimitiveRestartIndex(New.PrimitiveRestartIndex);
+                }
+            }
+
+            Old = New;
+        }
+
+        private void BindConstBuffers(GalPipelineState New)
+        {
+            //Index 0 is reserved
+            int FreeBinding = 1;
+
+            void BindIfNotNull(OGLShaderStage Stage)
+            {
+                if (Stage != null)
+                {
+                    foreach (ShaderDeclInfo DeclInfo in Stage.UniformUsage)
+                    {
+                        long Key = New.ConstBufferKeys[(int)Stage.Type][DeclInfo.Cbuf];
+
+                        if (Key != 0 && Buffer.TryGetUbo(Key, out int UboHandle))
+                        {
+                            GL.BindBufferBase(BufferRangeTarget.UniformBuffer, FreeBinding, UboHandle);
+                        }
+
+                        FreeBinding++;
+                    }
+                }
+            }
+
+            BindIfNotNull(Shader.Current.Vertex);
+            BindIfNotNull(Shader.Current.TessControl);
+            BindIfNotNull(Shader.Current.TessEvaluation);
+            BindIfNotNull(Shader.Current.Geometry);
+            BindIfNotNull(Shader.Current.Fragment);
+        }
+
+        private void BindVertexLayout(GalPipelineState New)
+        {
+            foreach (GalVertexBinding Binding in New.VertexBindings)
+            {
+                if (!Binding.Enabled || !Rasterizer.TryGetVbo(Binding.VboKey, out int VboHandle))
+                {
+                    continue;
+                }
+
+                if (VaoHandle == 0)
+                {
+                    VaoHandle = GL.GenVertexArray();
+
+                    //Vertex arrays shouldn't be used anywhere else in OpenGL's backend
+                    //if you want to use it, move this line out of the if
+                    GL.BindVertexArray(VaoHandle);
+                }
+
+                foreach (GalVertexAttrib Attrib in Binding.Attribs)
+                {
+                    GL.EnableVertexAttribArray(Attrib.Index);
+
+                    GL.BindBuffer(BufferTarget.ArrayBuffer, VboHandle);
+
+                    bool Unsigned =
+                        Attrib.Type == GalVertexAttribType.Unorm ||
+                        Attrib.Type == GalVertexAttribType.Uint ||
+                        Attrib.Type == GalVertexAttribType.Uscaled;
+
+                    bool Normalize =
+                        Attrib.Type == GalVertexAttribType.Snorm ||
+                        Attrib.Type == GalVertexAttribType.Unorm;
+
+                    VertexAttribPointerType Type = 0;
+
+                    if (Attrib.Type == GalVertexAttribType.Float)
+                    {
+                        Type = VertexAttribPointerType.Float;
+                    }
+                    else
+                    {
+                        Type = AttribTypes[Attrib.Size] + (Unsigned ? 1 : 0);
+                    }
+
+                    int Size = AttribElements[Attrib.Size];
+                    int Offset = Attrib.Offset;
+
+                    if (Attrib.Type == GalVertexAttribType.Sint ||
+                        Attrib.Type == GalVertexAttribType.Uint)
+                    {
+                        IntPtr Pointer = new IntPtr(Offset);
+
+                        VertexAttribIntegerType IType = (VertexAttribIntegerType)Type;
+
+                        GL.VertexAttribIPointer(Attrib.Index, Size, IType, Binding.Stride, Pointer);
+                    }
+                    else
+                    {
+                        GL.VertexAttribPointer(Attrib.Index, Size, Type, Normalize, Binding.Stride, Offset);
+                    }
+                }
+            }
+        }
+
+        private void Enable(EnableCap Cap, bool Enabled)
+        {
+            if (Enabled)
+            {
+                GL.Enable(Cap);
+            }
+            else
+            {
+                GL.Disable(Cap);
+            }
+        }
+    }
+}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs
index c5166b51..08021478 100644
--- a/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs
+++ b/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs
@@ -6,46 +6,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 {
     public class OGLRasterizer : IGalRasterizer
     {
-        private static Dictionary<GalVertexAttribSize, int> AttribElements =
-                   new Dictionary<GalVertexAttribSize, int>()
-        {
-            { GalVertexAttribSize._32_32_32_32, 4 },
-            { GalVertexAttribSize._32_32_32,    3 },
-            { GalVertexAttribSize._16_16_16_16, 4 },
-            { GalVertexAttribSize._32_32,       2 },
-            { GalVertexAttribSize._16_16_16,    3 },
-            { GalVertexAttribSize._8_8_8_8,     4 },
-            { GalVertexAttribSize._16_16,       2 },
-            { GalVertexAttribSize._32,          1 },
-            { GalVertexAttribSize._8_8_8,       3 },
-            { GalVertexAttribSize._8_8,         2 },
-            { GalVertexAttribSize._16,          1 },
-            { GalVertexAttribSize._8,           1 },
-            { GalVertexAttribSize._10_10_10_2,  4 },
-            { GalVertexAttribSize._11_11_10,    3 }
-        };
-
-        private static Dictionary<GalVertexAttribSize, VertexAttribPointerType> AttribTypes =
-                   new Dictionary<GalVertexAttribSize, VertexAttribPointerType>()
-        {
-            { GalVertexAttribSize._32_32_32_32, VertexAttribPointerType.Int   },
-            { GalVertexAttribSize._32_32_32,    VertexAttribPointerType.Int   },
-            { GalVertexAttribSize._16_16_16_16, VertexAttribPointerType.Short },
-            { GalVertexAttribSize._32_32,       VertexAttribPointerType.Int   },
-            { GalVertexAttribSize._16_16_16,    VertexAttribPointerType.Short },
-            { GalVertexAttribSize._8_8_8_8,     VertexAttribPointerType.Byte  },
-            { GalVertexAttribSize._16_16,       VertexAttribPointerType.Short },
-            { GalVertexAttribSize._32,          VertexAttribPointerType.Int   },
-            { GalVertexAttribSize._8_8_8,       VertexAttribPointerType.Byte  },
-            { GalVertexAttribSize._8_8,         VertexAttribPointerType.Byte  },
-            { GalVertexAttribSize._16,          VertexAttribPointerType.Short },
-            { GalVertexAttribSize._8,           VertexAttribPointerType.Byte  },
-            { GalVertexAttribSize._10_10_10_2,  VertexAttribPointerType.Int   }, //?
-            { GalVertexAttribSize._11_11_10,    VertexAttribPointerType.Int   }  //?
-        };
-
-        private int VaoHandle;
-
         private int[] VertexBuffers;
 
         private OGLCachedResource<int> VboCache;
@@ -83,7 +43,11 @@ namespace Ryujinx.Graphics.Gal.OpenGL
             IboCache.Unlock();
         }
 
-        public void ClearBuffers(GalClearBufferFlags Flags)
+        public void ClearBuffers(
+            GalClearBufferFlags Flags,
+            float Red, float Green, float Blue, float Alpha,
+            float Depth,
+            int Stencil)
         {
             ClearBufferMask Mask = ClearBufferMask.ColorBufferBit;
 
@@ -103,6 +67,12 @@ namespace Ryujinx.Graphics.Gal.OpenGL
                 Mask |= ClearBufferMask.StencilBufferBit;
             }
 
+            GL.ClearColor(Red, Green, Blue, Alpha);
+
+            GL.ClearDepth(Depth);
+
+            GL.ClearStencil(Stencil);
+
             GL.Clear(Mask);
 
             GL.ColorMask(true, true, true, true);
@@ -118,99 +88,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL
             return IboCache.TryGetSize(Key, out long Size) && Size == DataSize;
         }
 
-        public void SetFrontFace(GalFrontFace FrontFace)
-        {
-            GL.FrontFace(OGLEnumConverter.GetFrontFace(FrontFace));
-        }
-
-        public void EnableCullFace()
-        {
-            GL.Enable(EnableCap.CullFace);
-        }
-
-        public void DisableCullFace()
-        {
-            GL.Disable(EnableCap.CullFace);
-        }
-
-        public void SetCullFace(GalCullFace CullFace)
-        {
-            GL.CullFace(OGLEnumConverter.GetCullFace(CullFace));
-        }
-
-        public void EnableDepthTest()
-        {
-            GL.Enable(EnableCap.DepthTest);
-        }
-
-        public void DisableDepthTest()
-        {
-            GL.Disable(EnableCap.DepthTest);
-        }
-
-        public void SetDepthFunction(GalComparisonOp Func)
-        {
-            GL.DepthFunc(OGLEnumConverter.GetDepthFunc(Func));
-        }
-
-        public void SetClearDepth(float Depth)
-        {
-            GL.ClearDepth(Depth);
-        }
-
-        public void EnableStencilTest()
-        {
-            GL.Enable(EnableCap.StencilTest);
-        }
-
-        public void DisableStencilTest()
-        {
-            GL.Disable(EnableCap.StencilTest);
-        }
-
-        public void SetStencilFunction(bool IsFrontFace, GalComparisonOp Func, int Ref, int Mask)
-        {
-            GL.StencilFuncSeparate(
-                IsFrontFace ? StencilFace.Front : StencilFace.Back,
-                OGLEnumConverter.GetStencilFunc(Func),
-                Ref,
-                Mask);
-        }
-
-        public void SetStencilOp(bool IsFrontFace, GalStencilOp Fail, GalStencilOp ZFail, GalStencilOp ZPass)
-        {
-            GL.StencilOpSeparate(
-                IsFrontFace ? StencilFace.Front : StencilFace.Back,
-                OGLEnumConverter.GetStencilOp(Fail),
-                OGLEnumConverter.GetStencilOp(ZFail),
-                OGLEnumConverter.GetStencilOp(ZPass));
-        }
-
-        public void SetStencilMask(bool IsFrontFace, int Mask)
-        {
-            GL.StencilMaskSeparate(IsFrontFace ? StencilFace.Front : StencilFace.Back, Mask);
-        }
-
-        public void SetClearStencil(int Stencil)
-        {
-            GL.ClearStencil(Stencil);
-        }
-
-        public void EnablePrimitiveRestart()
-        {
-            GL.Enable(EnableCap.PrimitiveRestart);
-        }
-
-        public void DisablePrimitiveRestart()
-        {
-            GL.Disable(EnableCap.PrimitiveRestart);
-        }
-
-        public void SetPrimitiveRestartIndex(uint Index)
-        {
-            GL.PrimitiveRestartIndex(Index);
-        }
-
         public void CreateVbo(long Key, int DataSize, IntPtr HostAddress)
         {
             int Handle = GL.GenBuffer();
@@ -235,65 +112,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL
             GL.BufferData(BufferTarget.ElementArrayBuffer, Length, HostAddress, BufferUsageHint.StreamDraw);
         }
 
-        public void SetVertexArray(int Stride, long VboKey, GalVertexAttrib[] Attribs)
-        {
-            if (!VboCache.TryGetValue(VboKey, out int VboHandle))
-            {
-                return;
-            }
-
-            if (VaoHandle == 0)
-            {
-                VaoHandle = GL.GenVertexArray();
-            }
-
-            GL.BindVertexArray(VaoHandle);
-
-            foreach (GalVertexAttrib Attrib in Attribs)
-            {
-                GL.EnableVertexAttribArray(Attrib.Index);
-
-                GL.BindBuffer(BufferTarget.ArrayBuffer, VboHandle);
-
-                bool Unsigned =
-                    Attrib.Type == GalVertexAttribType.Unorm ||
-                    Attrib.Type == GalVertexAttribType.Uint  ||
-                    Attrib.Type == GalVertexAttribType.Uscaled;
-
-                bool Normalize =
-                    Attrib.Type == GalVertexAttribType.Snorm ||
-                    Attrib.Type == GalVertexAttribType.Unorm;
-
-                VertexAttribPointerType Type = 0;
-
-                if (Attrib.Type == GalVertexAttribType.Float)
-                {
-                    Type = VertexAttribPointerType.Float;
-                }
-                else
-                {
-                    Type = AttribTypes[Attrib.Size] + (Unsigned ? 1 : 0);
-                }
-
-                int Size   = AttribElements[Attrib.Size];
-                int Offset = Attrib.Offset;
-
-                if (Attrib.Type == GalVertexAttribType.Sint ||
-                    Attrib.Type == GalVertexAttribType.Uint)
-                {
-                    IntPtr Pointer = new IntPtr(Offset);
-
-                    VertexAttribIntegerType IType = (VertexAttribIntegerType)Type;
-
-                    GL.VertexAttribIPointer(Attrib.Index, Size, IType, Stride, Pointer);
-                }
-                else
-                {
-                    GL.VertexAttribPointer(Attrib.Index, Size, Type, Normalize, Stride, Offset);
-                }
-            }
-        }
-
         public void SetIndexArray(int Size, GalIndexFormat Format)
         {
             IndexBuffer.Type = OGLEnumConverter.GetDrawElementsType(Format);
@@ -310,8 +128,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL
                 return;
             }
 
-            GL.BindVertexArray(VaoHandle);
-
             GL.DrawArrays(OGLEnumConverter.GetPrimitiveType(PrimType), First, PrimCount);
         }
 
@@ -324,8 +140,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 
             PrimitiveType Mode = OGLEnumConverter.GetPrimitiveType(PrimType);
 
-            GL.BindVertexArray(VaoHandle);
-
             GL.BindBuffer(BufferTarget.ElementArrayBuffer, IboHandle);
 
             First <<= IndexBuffer.ElemSizeLog2;
@@ -341,5 +155,10 @@ namespace Ryujinx.Graphics.Gal.OpenGL
                 GL.DrawElements(Mode, IndexBuffer.Count, IndexBuffer.Type, First);
             }
         }
+
+        public bool TryGetVbo(long VboKey, out int VboHandle)
+        {
+            return VboCache.TryGetValue(VboKey, out VboHandle);
+        }
     }
 }
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLRenderer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLRenderer.cs
index ca70d4f6..b0f6da45 100644
--- a/Ryujinx.Graphics/Gal/OpenGL/OGLRenderer.cs
+++ b/Ryujinx.Graphics/Gal/OpenGL/OGLRenderer.cs
@@ -5,7 +5,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 {
     public class OGLRenderer : IGalRenderer
     {
-        public IGalBlend Blend { get; private set; }
+        public IGalConstBuffer Buffer { get; private set; }
 
         public IGalFrameBuffer FrameBuffer { get; private set; }
 
@@ -13,19 +13,23 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 
         public IGalShader Shader { get; private set; }
 
+        public IGalPipeline Pipeline { get; private set; }
+
         public IGalTexture Texture { get; private set; }
 
         private ConcurrentQueue<Action> ActionsQueue;
 
         public OGLRenderer()
         {
-            Blend = new OGLBlend();
+            Buffer = new OGLConstBuffer();
 
             FrameBuffer = new OGLFrameBuffer();
 
             Rasterizer = new OGLRasterizer();
 
-            Shader = new OGLShader();
+            Shader = new OGLShader(Buffer as OGLConstBuffer);
+
+            Pipeline = new OGLPipeline(Buffer as OGLConstBuffer, Rasterizer as OGLRasterizer, Shader as OGLShader);
 
             Texture = new OGLTexture();
 
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs
index 3f3f23b8..4792bc5e 100644
--- a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs
+++ b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs
@@ -5,96 +5,29 @@ using System.Collections.Concurrent;
 using System.Collections.Generic;
 using System.Linq;
 
-using Buffer = System.Buffer;
-
 namespace Ryujinx.Graphics.Gal.OpenGL
 {
-    public class OGLShader : IGalShader
+    class OGLShader : IGalShader
     {
-        private class ShaderStage : IDisposable
-        {
-            public int Handle { get; private set; }
+        public OGLShaderProgram Current;
 
-            public bool IsCompiled { get; private set; }
+        private ConcurrentDictionary<long, OGLShaderStage> Stages;
 
-            public GalShaderType Type { get; private set; }
-
-            public string Code { get; private set; }
-
-            public IEnumerable<ShaderDeclInfo> TextureUsage { get; private set; }
-            public IEnumerable<ShaderDeclInfo> UniformUsage { get; private set; }
-
-            public ShaderStage(
-                GalShaderType               Type,
-                string                      Code,
-                IEnumerable<ShaderDeclInfo> TextureUsage,
-                IEnumerable<ShaderDeclInfo> UniformUsage)
-            {
-                this.Type         = Type;
-                this.Code         = Code;
-                this.TextureUsage = TextureUsage;
-                this.UniformUsage = UniformUsage;
-            }
-
-            public void Compile()
-            {
-                if (Handle == 0)
-                {
-                    Handle = GL.CreateShader(OGLEnumConverter.GetShaderType(Type));
-
-                    CompileAndCheck(Handle, Code);
-                }
-            }
-
-            public void Dispose()
-            {
-                Dispose(true);
-            }
-
-            protected virtual void Dispose(bool Disposing)
-            {
-                if (Disposing && Handle != 0)
-                {
-                    GL.DeleteShader(Handle);
-
-                    Handle = 0;
-                }
-            }
-        }
-
-        private struct ShaderProgram
-        {
-            public ShaderStage Vertex;
-            public ShaderStage TessControl;
-            public ShaderStage TessEvaluation;
-            public ShaderStage Geometry;
-            public ShaderStage Fragment;
-        }
-
-        const int ConstBuffersPerStage = 18;
-
-        private ShaderProgram Current;
-
-        private ConcurrentDictionary<long, ShaderStage> Stages;
-
-        private Dictionary<ShaderProgram, int> Programs;
+        private Dictionary<OGLShaderProgram, int> Programs;
 
         public int CurrentProgramHandle { get; private set; }
 
-        private OGLStreamBuffer[][] ConstBuffers;
+        private OGLConstBuffer Buffer;
 
-        public OGLShader()
+        private int ExtraUboHandle;
+
+        public OGLShader(OGLConstBuffer Buffer)
         {
-            Stages = new ConcurrentDictionary<long, ShaderStage>();
+            this.Buffer = Buffer;
 
-            Programs = new Dictionary<ShaderProgram, int>();
+            Stages = new ConcurrentDictionary<long, OGLShaderStage>();
 
-            ConstBuffers = new OGLStreamBuffer[5][];
-
-            for (int i = 0; i < 5; i++)
-            {
-                ConstBuffers[i] = new OGLStreamBuffer[ConstBuffersPerStage];
-            }
+            Programs = new Dictionary<OGLShaderProgram, int>();
         }
 
         public void Create(IGalMemory Memory, long Key, GalShaderType Type)
@@ -107,7 +40,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL
             Stages.GetOrAdd(Key, (Stage) => ShaderStageFactory(Memory, VpAPos, Key, true, Type));
         }
 
-        private ShaderStage ShaderStageFactory(
+        private OGLShaderStage ShaderStageFactory(
             IGalMemory    Memory,
             long          Position,
             long          PositionB,
@@ -136,7 +69,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL
                 Program = Decompiler.Decompile(Memory, Position, Type);
             }
 
-            return new ShaderStage(
+            return new OGLShaderStage(
                 Type,
                 Program.Code,
                 Program.Textures,
@@ -145,7 +78,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 
         public IEnumerable<ShaderDeclInfo> GetTextureUsage(long Key)
         {
-            if (Stages.TryGetValue(Key, out ShaderStage Stage))
+            if (Stages.TryGetValue(Key, out OGLShaderStage Stage))
             {
                 return Stage.TextureUsage;
             }
@@ -153,21 +86,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL
             return Enumerable.Empty<ShaderDeclInfo>();
         }
 
-        public void SetConstBuffer(long Key, int Cbuf, int DataSize, IntPtr HostAddress)
-        {
-            if (Stages.TryGetValue(Key, out ShaderStage Stage))
-            {
-                foreach (ShaderDeclInfo DeclInfo in Stage.UniformUsage.Where(x => x.Cbuf == Cbuf))
-                {
-                    OGLStreamBuffer Buffer = GetConstBuffer(Stage.Type, Cbuf);
-
-                    int Size = Math.Min(DataSize, Buffer.Size);
-
-                    Buffer.SetData(Size, HostAddress);
-                }
-            }
-        }
-
         public void EnsureTextureBinding(string UniformName, int Value)
         {
             BindProgram();
@@ -177,24 +95,33 @@ namespace Ryujinx.Graphics.Gal.OpenGL
             GL.Uniform1(Location, Value);
         }
 
-        public void SetFlip(float X, float Y)
+        public unsafe void SetFlip(float X, float Y)
         {
             BindProgram();
 
-            int Location = GL.GetUniformLocation(CurrentProgramHandle, GlslDecl.FlipUniformName);
+            EnsureExtraBlock();
 
-            GL.Uniform2(Location, X, Y);
+            GL.BindBuffer(BufferTarget.UniformBuffer, ExtraUboHandle);
+
+            float* Data = stackalloc float[4];
+            Data[0] = X;
+            Data[1] = Y;
+
+            //Invalidate buffer
+            GL.BufferData(BufferTarget.UniformBuffer, 4 * sizeof(float), IntPtr.Zero, BufferUsageHint.StreamDraw);
+
+            GL.BufferSubData(BufferTarget.UniformBuffer, IntPtr.Zero, 4 * sizeof(float), (IntPtr)Data);
         }
 
         public void Bind(long Key)
         {
-            if (Stages.TryGetValue(Key, out ShaderStage Stage))
+            if (Stages.TryGetValue(Key, out OGLShaderStage Stage))
             {
                 Bind(Stage);
             }
         }
 
-        private void Bind(ShaderStage Stage)
+        private void Bind(OGLShaderStage Stage)
         {
             if (Stage.Type == GalShaderType.Geometry)
             {
@@ -257,15 +184,24 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 
             GL.UseProgram(Handle);
 
-            if (CurrentProgramHandle != Handle)
-            {
-                BindUniformBuffers(Handle);
-            }
-
             CurrentProgramHandle = Handle;
         }
 
-        private void AttachIfNotNull(int ProgramHandle, ShaderStage Stage)
+        private void EnsureExtraBlock()
+        {
+            if (ExtraUboHandle == 0)
+            {
+                ExtraUboHandle = GL.GenBuffer();
+
+                GL.BindBuffer(BufferTarget.UniformBuffer, ExtraUboHandle);
+
+                GL.BufferData(BufferTarget.UniformBuffer, 4 * sizeof(float), IntPtr.Zero, BufferUsageHint.StreamDraw);
+
+                GL.BindBufferBase(BufferRangeTarget.UniformBuffer, 0, ExtraUboHandle);
+            }
+        }
+
+        private void AttachIfNotNull(int ProgramHandle, OGLShaderStage Stage)
         {
             if (Stage != null)
             {
@@ -277,9 +213,14 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 
         private void BindUniformBlocks(int ProgramHandle)
         {
-            int FreeBinding = 0;
+            int ExtraBlockindex = GL.GetUniformBlockIndex(ProgramHandle, GlslDecl.ExtraUniformBlockName);
 
-            void BindUniformBlocksIfNotNull(ShaderStage Stage)
+            GL.UniformBlockBinding(ProgramHandle, ExtraBlockindex, 0);
+
+            //First index is reserved
+            int FreeBinding = 1;
+
+            void BindUniformBlocksIfNotNull(OGLShaderStage Stage)
             {
                 if (Stage != null)
                 {
@@ -307,71 +248,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL
             BindUniformBlocksIfNotNull(Current.Fragment);
         }
 
-        private void BindUniformBuffers(int ProgramHandle)
-        {
-            int FreeBinding = 0;
-
-            void BindUniformBuffersIfNotNull(ShaderStage Stage)
-            {
-                if (Stage != null)
-                {
-                    foreach (ShaderDeclInfo DeclInfo in Stage.UniformUsage)
-                    {
-                        OGLStreamBuffer Buffer = GetConstBuffer(Stage.Type, DeclInfo.Cbuf);
-
-                        GL.BindBufferBase(BufferRangeTarget.UniformBuffer, FreeBinding, Buffer.Handle);
-
-                        FreeBinding++;
-                    }
-                }
-            }
-
-            BindUniformBuffersIfNotNull(Current.Vertex);
-            BindUniformBuffersIfNotNull(Current.TessControl);
-            BindUniformBuffersIfNotNull(Current.TessEvaluation);
-            BindUniformBuffersIfNotNull(Current.Geometry);
-            BindUniformBuffersIfNotNull(Current.Fragment);
-        }
-
-        private OGLStreamBuffer GetConstBuffer(GalShaderType StageType, int Cbuf)
-        {
-            int StageIndex = (int)StageType;
-
-            OGLStreamBuffer Buffer = ConstBuffers[StageIndex][Cbuf];
-
-            if (Buffer == null)
-            {
-                //Allocate a maximum of 64 KiB
-                int Size = Math.Min(GL.GetInteger(GetPName.MaxUniformBlockSize), 64 * 1024);
-
-                Buffer = new OGLStreamBuffer(BufferTarget.UniformBuffer, Size);
-
-                ConstBuffers[StageIndex][Cbuf] = Buffer;
-            }
-
-            return Buffer;
-        }
-
-        public static void CompileAndCheck(int Handle, string Code)
-        {
-            GL.ShaderSource(Handle, Code);
-            GL.CompileShader(Handle);
-
-            CheckCompilation(Handle);
-        }
-
-        private static void CheckCompilation(int Handle)
-        {
-            int Status = 0;
-
-            GL.GetShader(Handle, ShaderParameter.CompileStatus, out Status);
-
-            if (Status == 0)
-            {
-                throw new ShaderException(GL.GetShaderInfoLog(Handle));
-            }
-        }
-
         private static void CheckProgramLink(int Handle)
         {
             int Status = 0;
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLShaderProgram.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLShaderProgram.cs
new file mode 100644
index 00000000..731994ce
--- /dev/null
+++ b/Ryujinx.Graphics/Gal/OpenGL/OGLShaderProgram.cs
@@ -0,0 +1,86 @@
+using OpenTK.Graphics.OpenGL;
+using System;
+using System.Collections.Generic;
+
+namespace Ryujinx.Graphics.Gal.OpenGL
+{
+    struct OGLShaderProgram
+    {
+        public OGLShaderStage Vertex;
+        public OGLShaderStage TessControl;
+        public OGLShaderStage TessEvaluation;
+        public OGLShaderStage Geometry;
+        public OGLShaderStage Fragment;
+    }
+
+    class OGLShaderStage : IDisposable
+    {
+        public int Handle { get; private set; }
+
+        public bool IsCompiled { get; private set; }
+
+        public GalShaderType Type { get; private set; }
+
+        public string Code { get; private set; }
+
+        public IEnumerable<ShaderDeclInfo> TextureUsage { get; private set; }
+        public IEnumerable<ShaderDeclInfo> UniformUsage { get; private set; }
+
+        public OGLShaderStage(
+            GalShaderType Type,
+            string Code,
+            IEnumerable<ShaderDeclInfo> TextureUsage,
+            IEnumerable<ShaderDeclInfo> UniformUsage)
+        {
+            this.Type = Type;
+            this.Code = Code;
+            this.TextureUsage = TextureUsage;
+            this.UniformUsage = UniformUsage;
+        }
+
+        public void Compile()
+        {
+            if (Handle == 0)
+            {
+                Handle = GL.CreateShader(OGLEnumConverter.GetShaderType(Type));
+
+                CompileAndCheck(Handle, Code);
+            }
+        }
+
+        public void Dispose()
+        {
+            Dispose(true);
+        }
+
+        protected virtual void Dispose(bool Disposing)
+        {
+            if (Disposing && Handle != 0)
+            {
+                GL.DeleteShader(Handle);
+
+                Handle = 0;
+            }
+        }
+
+        public static void CompileAndCheck(int Handle, string Code)
+        {
+            GL.ShaderSource(Handle, Code);
+            GL.CompileShader(Handle);
+
+            CheckCompilation(Handle);
+        }
+
+        private static void CheckCompilation(int Handle)
+        {
+            int Status = 0;
+
+            GL.GetShader(Handle, ShaderParameter.CompileStatus, out Status);
+
+            if (Status == 0)
+            {
+                throw new ShaderException(GL.GetShaderInfoLog(Handle));
+            }
+        }
+    }
+}
\ No newline at end of file
diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs
index 0d5dee93..94639405 100644
--- a/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs
+++ b/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs
@@ -7,11 +7,11 @@ namespace Ryujinx.Graphics.Gal.OpenGL
     {
         public int Handle { get; protected set; }
 
-        public int Size { get; protected set; }
+        public long Size { get; protected set; }
 
         protected BufferTarget Target { get; private set; }
 
-        public OGLStreamBuffer(BufferTarget Target, int Size)
+        public OGLStreamBuffer(BufferTarget Target, long Size)
         {
             this.Target = Target;
             this.Size   = Size;
@@ -20,14 +20,14 @@ namespace Ryujinx.Graphics.Gal.OpenGL
 
             GL.BindBuffer(Target, Handle);
 
-            GL.BufferData(Target, Size, IntPtr.Zero, BufferUsageHint.StreamDraw);
+            GL.BufferData(Target, (IntPtr)Size, IntPtr.Zero, BufferUsageHint.StreamDraw);
         }
 
-        public void SetData(int Size, IntPtr HostAddress)
+        public void SetData(long Size, IntPtr HostAddress)
         {
             GL.BindBuffer(Target, Handle);
 
-            GL.BufferSubData(Target, IntPtr.Zero, Size, HostAddress);
+            GL.BufferSubData(Target, IntPtr.Zero, (IntPtr)Size, HostAddress);
         }
 
         public void Dispose()
diff --git a/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs b/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs
index 7688545c..a0c747ba 100644
--- a/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs
+++ b/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs
@@ -35,6 +35,7 @@ namespace Ryujinx.Graphics.Gal.Shader
 
         public const string FragmentOutputName = "FragColor";
 
+        public const string ExtraUniformBlockName = "Extra";
         public const string FlipUniformName = "flip";
 
         public const string ProgramName  = "program";
diff --git a/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs b/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs
index 5261d677..aa1803a5 100644
--- a/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs
+++ b/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs
@@ -220,7 +220,11 @@ namespace Ryujinx.Graphics.Gal.Shader
         {
             if (Decl.ShaderType == GalShaderType.Vertex)
             {
-                SB.AppendLine("uniform vec2 " + GlslDecl.FlipUniformName + ";");
+                SB.AppendLine("layout (std140) uniform " + GlslDecl.ExtraUniformBlockName + "{");
+
+                SB.AppendLine(IdentationStr + "vec2 " + GlslDecl.FlipUniformName + ";");
+
+                SB.AppendLine("};");
             }
 
             SB.AppendLine();
diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs
index 3dd0e98e..0576601f 100644
--- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs
+++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs
@@ -84,35 +84,41 @@ namespace Ryujinx.HLE.Gpu.Engines
         {
             LockCaches();
 
+            GalPipelineState State = new GalPipelineState();
+
+            SetFlip(State);
+            SetFrontFace(State);
+            SetCullFace(State);
+            SetDepth(State);
+            SetStencil(State);
+            SetAlphaBlending(State);
+            SetPrimitiveRestart(State);
+
             SetFrameBuffer(Vmm, 0);
 
             long[] Keys = UploadShaders(Vmm);
 
             Gpu.Renderer.Shader.BindProgram();
 
-            //Note: Uncomment SetFrontFace SetCullFace when flipping issues are solved
-            //SetFrontFace();
-            //SetCullFace();
-            SetDepth();
-            SetStencil();
-            SetAlphaBlending();
-            SetPrimitiveRestart();
+            UploadTextures(Vmm, State, Keys);
+            UploadConstBuffers(Vmm, State);
+            UploadVertexArrays(Vmm, State);
 
-            UploadTextures(Vmm, Keys);
-            UploadUniforms(Vmm);
-            UploadVertexArrays(Vmm);
+            DispatchRender(Vmm, State);
 
             UnlockCaches();
         }
 
         private void LockCaches()
         {
+            Gpu.Renderer.Buffer.LockCache();
             Gpu.Renderer.Rasterizer.LockCaches();
             Gpu.Renderer.Texture.LockCache();
         }
 
         private void UnlockCaches()
         {
+            Gpu.Renderer.Buffer.UnlockCache();
             Gpu.Renderer.Rasterizer.UnlockCaches();
             Gpu.Renderer.Texture.UnlockCache();
         }
@@ -125,9 +131,22 @@ namespace Ryujinx.HLE.Gpu.Engines
 
             GalClearBufferFlags Flags = (GalClearBufferFlags)(Arg0 & 0x3f);
 
+            float Red   = ReadRegisterFloat(NvGpuEngine3dReg.ClearNColor + 0);
+            float Green = ReadRegisterFloat(NvGpuEngine3dReg.ClearNColor + 1);
+            float Blue  = ReadRegisterFloat(NvGpuEngine3dReg.ClearNColor + 2);
+            float Alpha = ReadRegisterFloat(NvGpuEngine3dReg.ClearNColor + 3);
+
+            float Depth = ReadRegisterFloat(NvGpuEngine3dReg.ClearDepth);
+
+            int Stencil = ReadRegister(NvGpuEngine3dReg.ClearStencil);
+
             SetFrameBuffer(Vmm, FbIndex);
 
-            Gpu.Renderer.Rasterizer.ClearBuffers(Flags);
+            Gpu.Renderer.Rasterizer.ClearBuffers(
+                Flags,
+                Red, Green, Blue, Alpha,
+                Depth,
+                Stencil);
         }
 
         private void SetFrameBuffer(NvGpuVmm Vmm, int FbIndex)
@@ -185,6 +204,8 @@ namespace Ryujinx.HLE.Gpu.Engines
                 long VpAPos = BasePosition + (uint)VpAOffset;
                 long VpBPos = BasePosition + (uint)VpBOffset;
 
+                Keys[(int)GalShaderType.Vertex] = VpBPos;
+
                 Gpu.Renderer.Shader.Create(Vmm, VpAPos, VpBPos, GalShaderType.Vertex);
                 Gpu.Renderer.Shader.Bind(VpBPos);
 
@@ -216,11 +237,6 @@ namespace Ryujinx.HLE.Gpu.Engines
                 Gpu.Renderer.Shader.Bind(Key);
             }
 
-            float SignX = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleX);
-            float SignY = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleY);
-
-            Gpu.Renderer.Shader.SetFlip(SignX, SignY);
-
             return Keys;
         }
 
@@ -239,7 +255,13 @@ namespace Ryujinx.HLE.Gpu.Engines
             throw new ArgumentOutOfRangeException(nameof(Program));
         }
 
-        private void SetFrontFace()
+        private void SetFlip(GalPipelineState State)
+        {
+            State.FlipX = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleX);
+            State.FlipY = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleY);
+        }
+
+        private void SetFrontFace(GalPipelineState State)
         {
             float SignX = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleX);
             float SignY = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleY);
@@ -261,198 +283,82 @@ namespace Ryujinx.HLE.Gpu.Engines
                 }
             }
 
-            Gpu.Renderer.Rasterizer.SetFrontFace(FrontFace);
+            State.FrontFace = FrontFace;
         }
 
-        private void SetCullFace()
+        private void SetCullFace(GalPipelineState State)
         {
-            bool Enable = (ReadRegister(NvGpuEngine3dReg.CullFaceEnable) & 1) != 0;
+            State.CullFaceEnabled = (ReadRegister(NvGpuEngine3dReg.CullFaceEnable) & 1) != 0;
 
-            if (Enable)
+            if (State.CullFaceEnabled)
             {
-                Gpu.Renderer.Rasterizer.EnableCullFace();
+                State.CullFace = (GalCullFace)ReadRegister(NvGpuEngine3dReg.CullFace);
             }
-            else
-            {
-                Gpu.Renderer.Rasterizer.DisableCullFace();
-            }
-
-            if (!Enable)
-            {
-                return;
-            }
-
-            GalCullFace CullFace = (GalCullFace)ReadRegister(NvGpuEngine3dReg.CullFace);
-
-            Gpu.Renderer.Rasterizer.SetCullFace(CullFace);
         }
 
-        private void SetDepth()
+        private void SetDepth(GalPipelineState State)
         {
-            float ClearDepth = ReadRegisterFloat(NvGpuEngine3dReg.ClearDepth);
+            State.DepthTestEnabled = (ReadRegister(NvGpuEngine3dReg.DepthTestEnable) & 1) != 0;
 
-            Gpu.Renderer.Rasterizer.SetClearDepth(ClearDepth);
-
-            bool Enable = (ReadRegister(NvGpuEngine3dReg.DepthTestEnable) & 1) != 0;
-
-            if (Enable)
+            if (State.DepthTestEnabled)
             {
-                Gpu.Renderer.Rasterizer.EnableDepthTest();
+                State.DepthFunc = (GalComparisonOp)ReadRegister(NvGpuEngine3dReg.DepthTestFunction);
             }
-            else
-            {
-                Gpu.Renderer.Rasterizer.DisableDepthTest();
-            }
-
-            if (!Enable)
-            {
-                return;
-            }
-
-            GalComparisonOp Func = (GalComparisonOp)ReadRegister(NvGpuEngine3dReg.DepthTestFunction);
-
-            Gpu.Renderer.Rasterizer.SetDepthFunction(Func);
         }
 
-        private void SetStencil()
+        private void SetStencil(GalPipelineState State)
         {
-            int ClearStencil = ReadRegister(NvGpuEngine3dReg.ClearStencil);
-
-            Gpu.Renderer.Rasterizer.SetClearStencil(ClearStencil);
-
-            bool Enable = (ReadRegister(NvGpuEngine3dReg.StencilEnable) & 1) != 0;
-
-            if (Enable)
+            State.StencilTestEnabled = (ReadRegister(NvGpuEngine3dReg.StencilEnable) & 1) != 0;
+            
+            if (State.StencilTestEnabled)
             {
-                Gpu.Renderer.Rasterizer.EnableStencilTest();
+                State.StencilBackFuncFunc = (GalComparisonOp)ReadRegister(NvGpuEngine3dReg.StencilBackFuncFunc);
+                State.StencilBackFuncRef  =                  ReadRegister(NvGpuEngine3dReg.StencilBackFuncRef);
+                State.StencilBackFuncMask =            (uint)ReadRegister(NvGpuEngine3dReg.StencilBackFuncMask);
+                State.StencilBackOpFail   =    (GalStencilOp)ReadRegister(NvGpuEngine3dReg.StencilBackOpFail);
+                State.StencilBackOpZFail  =    (GalStencilOp)ReadRegister(NvGpuEngine3dReg.StencilBackOpZFail);
+                State.StencilBackOpZPass  =    (GalStencilOp)ReadRegister(NvGpuEngine3dReg.StencilBackOpZPass);
+                State.StencilBackMask     =            (uint)ReadRegister(NvGpuEngine3dReg.StencilBackMask);
+
+                State.StencilFrontFuncFunc = (GalComparisonOp)ReadRegister(NvGpuEngine3dReg.StencilFrontFuncFunc);
+                State.StencilFrontFuncRef  =                  ReadRegister(NvGpuEngine3dReg.StencilFrontFuncRef);
+                State.StencilFrontFuncMask =            (uint)ReadRegister(NvGpuEngine3dReg.StencilFrontFuncMask);
+                State.StencilFrontOpFail   =    (GalStencilOp)ReadRegister(NvGpuEngine3dReg.StencilFrontOpFail);
+                State.StencilFrontOpZFail  =    (GalStencilOp)ReadRegister(NvGpuEngine3dReg.StencilFrontOpZFail);
+                State.StencilFrontOpZPass  =    (GalStencilOp)ReadRegister(NvGpuEngine3dReg.StencilFrontOpZPass);
+                State.StencilFrontMask     =            (uint)ReadRegister(NvGpuEngine3dReg.StencilFrontMask);
             }
-            else
-            {
-                Gpu.Renderer.Rasterizer.DisableStencilTest();
-            }
-
-            if (!Enable)
-            {
-                return;
-            }
-
-            void SetFaceStencil(
-                bool IsFrontFace,
-                NvGpuEngine3dReg Func,
-                NvGpuEngine3dReg FuncRef,
-                NvGpuEngine3dReg FuncMask,
-                NvGpuEngine3dReg OpFail,
-                NvGpuEngine3dReg OpZFail,
-                NvGpuEngine3dReg OpZPass,
-                NvGpuEngine3dReg Mask)
-            {
-                Gpu.Renderer.Rasterizer.SetStencilFunction(
-                    IsFrontFace,
-                    (GalComparisonOp)ReadRegister(Func),
-                    ReadRegister(FuncRef),
-                    ReadRegister(FuncMask));
-
-                Gpu.Renderer.Rasterizer.SetStencilOp(
-                    IsFrontFace,
-                    (GalStencilOp)ReadRegister(OpFail),
-                    (GalStencilOp)ReadRegister(OpZFail),
-                    (GalStencilOp)ReadRegister(OpZPass));
-
-                Gpu.Renderer.Rasterizer.SetStencilMask(IsFrontFace, ReadRegister(Mask));
-            }
-
-            SetFaceStencil(false,
-                NvGpuEngine3dReg.StencilBackFuncFunc,
-                NvGpuEngine3dReg.StencilBackFuncRef,
-                NvGpuEngine3dReg.StencilBackFuncMask,
-                NvGpuEngine3dReg.StencilBackOpFail,
-                NvGpuEngine3dReg.StencilBackOpZFail,
-                NvGpuEngine3dReg.StencilBackOpZPass,
-                NvGpuEngine3dReg.StencilBackMask);
-
-            SetFaceStencil(true,
-                NvGpuEngine3dReg.StencilFrontFuncFunc,
-                NvGpuEngine3dReg.StencilFrontFuncRef,
-                NvGpuEngine3dReg.StencilFrontFuncMask,
-                NvGpuEngine3dReg.StencilFrontOpFail,
-                NvGpuEngine3dReg.StencilFrontOpZFail,
-                NvGpuEngine3dReg.StencilFrontOpZPass,
-                NvGpuEngine3dReg.StencilFrontMask);
         }
 
-        private void SetAlphaBlending()
+        private void SetAlphaBlending(GalPipelineState State)
         {
             //TODO: Support independent blend properly.
-            bool Enable = (ReadRegister(NvGpuEngine3dReg.IBlendNEnable) & 1) != 0;
+            State.BlendEnabled = (ReadRegister(NvGpuEngine3dReg.IBlendNEnable) & 1) != 0;
 
-            if (Enable)
+            if (State.BlendEnabled)
             {
-                Gpu.Renderer.Blend.Enable();
-            }
-            else
-            {
-                Gpu.Renderer.Blend.Disable();
-            }
+                State.BlendSeparateAlpha = (ReadRegister(NvGpuEngine3dReg.IBlendNSeparateAlpha) & 1) != 0;
 
-            if (!Enable)
-            {
-                //If blend is not enabled, then the other values have no effect.
-                //Note that if it is disabled, the register may contain invalid values.
-                return;
-            }
-
-            bool BlendSeparateAlpha = (ReadRegister(NvGpuEngine3dReg.IBlendNSeparateAlpha) & 1) != 0;
-
-            GalBlendEquation EquationRgb = (GalBlendEquation)ReadRegister(NvGpuEngine3dReg.IBlendNEquationRgb);
-
-            GalBlendFactor FuncSrcRgb = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncSrcRgb);
-            GalBlendFactor FuncDstRgb = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncDstRgb);
-
-            if (BlendSeparateAlpha)
-            {
-                GalBlendEquation EquationAlpha = (GalBlendEquation)ReadRegister(NvGpuEngine3dReg.IBlendNEquationAlpha);
-
-                GalBlendFactor FuncSrcAlpha = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncSrcAlpha);
-                GalBlendFactor FuncDstAlpha = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncDstAlpha);
-
-                Gpu.Renderer.Blend.SetSeparate(
-                    EquationRgb,
-                    EquationAlpha,
-                    FuncSrcRgb,
-                    FuncDstRgb,
-                    FuncSrcAlpha,
-                    FuncDstAlpha);
-            }
-            else
-            {
-                Gpu.Renderer.Blend.Set(EquationRgb, FuncSrcRgb, FuncDstRgb);
+                State.BlendEquationRgb   = (GalBlendEquation)ReadRegister(NvGpuEngine3dReg.IBlendNEquationRgb);
+                State.BlendFuncSrcRgb    =   (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncSrcRgb);
+                State.BlendFuncDstRgb    =   (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncDstRgb);
+                State.BlendEquationAlpha = (GalBlendEquation)ReadRegister(NvGpuEngine3dReg.IBlendNEquationAlpha);
+                State.BlendFuncSrcAlpha  =   (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncSrcAlpha);
+                State.BlendFuncDstAlpha  =   (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.IBlendNFuncDstAlpha);
             }
         }
 
-        private void SetPrimitiveRestart()
+        private void SetPrimitiveRestart(GalPipelineState State)
         {
-            bool Enable = (ReadRegister(NvGpuEngine3dReg.PrimRestartEnable) & 1) != 0;
+            State.PrimitiveRestartEnabled = (ReadRegister(NvGpuEngine3dReg.PrimRestartEnable) & 1) != 0;
 
-            if (Enable)
+            if (State.PrimitiveRestartEnabled)
             {
-                Gpu.Renderer.Rasterizer.EnablePrimitiveRestart();
+                State.PrimitiveRestartIndex = (uint)ReadRegister(NvGpuEngine3dReg.PrimRestartIndex);
             }
-            else
-            {
-                Gpu.Renderer.Rasterizer.DisablePrimitiveRestart();
-            }
-
-            if (!Enable)
-            {
-                return;
-            }
-
-            uint Index = (uint)ReadRegister(NvGpuEngine3dReg.PrimRestartIndex);
-
-            Gpu.Renderer.Rasterizer.SetPrimitiveRestartIndex(Index);
         }
 
-        private void UploadTextures(NvGpuVmm Vmm, long[] Keys)
+        private void UploadTextures(NvGpuVmm Vmm, GalPipelineState State, long[] Keys)
         {
             long BaseShPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.ShaderAddress);
 
@@ -546,45 +452,35 @@ namespace Ryujinx.HLE.Gpu.Engines
             Gpu.Renderer.Texture.SetSampler(Sampler);
         }
 
-        private void UploadUniforms(NvGpuVmm Vmm)
+        private void UploadConstBuffers(NvGpuVmm Vmm, GalPipelineState State)
         {
-            long BasePosition = MakeInt64From2xInt32(NvGpuEngine3dReg.ShaderAddress);
-
-            for (int Index = 0; Index < 5; Index++)
+            for (int Stage = 0; Stage < State.ConstBufferKeys.Length; Stage++)
             {
-                int Control = ReadRegister(NvGpuEngine3dReg.ShaderNControl + (Index + 1) * 0x10);
-                int Offset  = ReadRegister(NvGpuEngine3dReg.ShaderNOffset  + (Index + 1) * 0x10);
-
-                //Note: Vertex Program (B) is always enabled.
-                bool Enable = (Control & 1) != 0 || Index == 0;
-
-                if (!Enable)
+                for (int Index = 0; Index < State.ConstBufferKeys[Stage].Length; Index++)
                 {
-                    continue;
-                }
+                    ConstBuffer Cb = ConstBuffers[Stage][Index];
 
-                for (int Cbuf = 0; Cbuf < ConstBuffers[Index].Length; Cbuf++)
-                {
-                    ConstBuffer Cb = ConstBuffers[Index][Cbuf];
+                    long Key = Cb.Position;
 
-                    if (Cb.Enabled)
+                    if (Cb.Enabled && QueryKeyUpload(Vmm, Key, Cb.Size, NvGpuBufferType.ConstBuffer))
                     {
-                        IntPtr DataAddress = Vmm.GetHostAddress(Cb.Position, Cb.Size);
+                        IntPtr Source = Vmm.GetHostAddress(Key, Cb.Size);
 
-                        Gpu.Renderer.Shader.SetConstBuffer(BasePosition + (uint)Offset, Cbuf, Cb.Size, DataAddress);
+                        Gpu.Renderer.Buffer.SetData(Key, Cb.Size, Source);
                     }
+
+                    State.ConstBufferKeys[Stage][Index] = Key;
                 }
             }
         }
 
-        private void UploadVertexArrays(NvGpuVmm Vmm)
+        private void UploadVertexArrays(NvGpuVmm Vmm, GalPipelineState State)
         {
             long IndexPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.IndexArrayAddress);
 
             long IboKey = Vmm.GetPhysicalAddress(IndexPosition);
 
             int IndexEntryFmt = ReadRegister(NvGpuEngine3dReg.IndexArrayFormat);
-            int IndexFirst    = ReadRegister(NvGpuEngine3dReg.IndexBatchFirst);
             int IndexCount    = ReadRegister(NvGpuEngine3dReg.IndexBatchCount);
 
             GalIndexFormat IndexFormat = (GalIndexFormat)IndexEntryFmt;
@@ -634,10 +530,7 @@ namespace Ryujinx.HLE.Gpu.Engines
                                          ((Packed >> 31) & 0x1) != 0));
             }
 
-            int VertexFirst = ReadRegister(NvGpuEngine3dReg.VertexArrayFirst);
-            int VertexCount = ReadRegister(NvGpuEngine3dReg.VertexArrayCount);
-
-            int PrimCtrl = ReadRegister(NvGpuEngine3dReg.VertexBeginGl);
+            State.VertexBindings = new GalVertexBinding[32];
 
             for (int Index = 0; Index < 32; Index++)
             {
@@ -650,14 +543,14 @@ namespace Ryujinx.HLE.Gpu.Engines
 
                 bool Enable = (Control & 0x1000) != 0;
 
-                long VertexPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.VertexArrayNAddress + Index * 4);
-                long VertexEndPos   = MakeInt64From2xInt32(NvGpuEngine3dReg.VertexArrayNEndAddr + Index * 2);
-
                 if (!Enable)
                 {
                     continue;
                 }
 
+                long VertexPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.VertexArrayNAddress + Index * 4);
+                long VertexEndPos   = MakeInt64From2xInt32(NvGpuEngine3dReg.VertexArrayNEndAddr + Index * 2);
+
                 long VboKey = Vmm.GetPhysicalAddress(VertexPosition);
 
                 int Stride = Control & 0xfff;
@@ -673,19 +566,39 @@ namespace Ryujinx.HLE.Gpu.Engines
                     Gpu.Renderer.Rasterizer.CreateVbo(VboKey, (int)VbSize, DataAddress);
                 }
 
-                Gpu.Renderer.Rasterizer.SetVertexArray(Stride, VboKey, Attribs[Index].ToArray());
+                State.VertexBindings[Index].Enabled = true;
+                State.VertexBindings[Index].Stride  = Stride;
+                State.VertexBindings[Index].VboKey  = VboKey;
+                State.VertexBindings[Index].Attribs = Attribs[Index].ToArray();
             }
+        }
+
+        private void DispatchRender(NvGpuVmm Vmm, GalPipelineState State)
+        {
+            int IndexCount = ReadRegister(NvGpuEngine3dReg.IndexBatchCount);
+            int PrimCtrl   = ReadRegister(NvGpuEngine3dReg.VertexBeginGl);
 
             GalPrimitiveType PrimType = (GalPrimitiveType)(PrimCtrl & 0xffff);
 
+            Gpu.Renderer.Pipeline.Bind(State);
+
             if (IndexCount != 0)
             {
-                int VertexBase = ReadRegister(NvGpuEngine3dReg.VertexArrayElemBase);
+                int IndexEntryFmt = ReadRegister(NvGpuEngine3dReg.IndexArrayFormat);
+                int IndexFirst    = ReadRegister(NvGpuEngine3dReg.IndexBatchFirst);
+                int VertexBase    = ReadRegister(NvGpuEngine3dReg.VertexArrayElemBase);
+
+                long IndexPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.IndexArrayAddress);
+
+                long IboKey = Vmm.GetPhysicalAddress(IndexPosition);
 
                 Gpu.Renderer.Rasterizer.DrawElements(IboKey, IndexFirst, VertexBase, PrimType);
             }
             else
             {
+                int VertexFirst = ReadRegister(NvGpuEngine3dReg.VertexArrayFirst);
+                int VertexCount = ReadRegister(NvGpuEngine3dReg.VertexArrayCount);
+
                 Gpu.Renderer.Rasterizer.DrawArrays(VertexFirst, VertexCount, PrimType);
             }
         }
@@ -741,10 +654,21 @@ namespace Ryujinx.HLE.Gpu.Engines
 
             long Position = MakeInt64From2xInt32(NvGpuEngine3dReg.ConstBufferAddress);
 
-            ConstBuffers[Stage][Index].Position = Position;
-            ConstBuffers[Stage][Index].Enabled  = Enabled;
+            int Size = ReadRegister(NvGpuEngine3dReg.ConstBufferSize);
 
-            ConstBuffers[Stage][Index].Size = ReadRegister(NvGpuEngine3dReg.ConstBufferSize);
+            if (!Gpu.Renderer.Buffer.IsCached(Position, Size))
+            {
+                Gpu.Renderer.Buffer.Create(Position, Size);
+            }
+
+            ConstBuffer Cb = ConstBuffers[Stage][Index];
+
+            if (Cb.Position != Position || Cb.Enabled != Enabled || Cb.Size != Size)
+            {
+                ConstBuffers[Stage][Index].Position = Position;
+                ConstBuffers[Stage][Index].Enabled = Enabled;
+                ConstBuffers[Stage][Index].Size = Size;
+            }
         }
 
         private float GetFlipSign(NvGpuEngine3dReg Reg)
diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs
index e7dabe44..39a5ee8c 100644
--- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs
+++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs
@@ -16,6 +16,7 @@ namespace Ryujinx.HLE.Gpu.Engines
         ViewportNVert        = 0x301,
         VertexArrayFirst     = 0x35d,
         VertexArrayCount     = 0x35e,
+        ClearNColor          = 0x360,
         ClearDepth           = 0x364,
         ClearStencil         = 0x368,
         StencilBackFuncRef   = 0x3d5,
diff --git a/Ryujinx.HLE/Gpu/Memory/NvGpuBufferType.cs b/Ryujinx.HLE/Gpu/Memory/NvGpuBufferType.cs
index 469cd6cd..a6c03f42 100644
--- a/Ryujinx.HLE/Gpu/Memory/NvGpuBufferType.cs
+++ b/Ryujinx.HLE/Gpu/Memory/NvGpuBufferType.cs
@@ -5,6 +5,7 @@ namespace Ryujinx.HLE.Gpu.Memory
         Index,
         Vertex,
         Texture,
+        ConstBuffer,
         Count
     }
 }
\ No newline at end of file