From 611bec6e44effa90554c95ed1fe4dd4812893947 Mon Sep 17 00:00:00 2001
From: gdkchan <gab.dark.100@gmail.com>
Date: Wed, 10 Nov 2021 15:37:49 -0300
Subject: [PATCH] Implement DrawTexture functionality (#2747)

* Implement DrawTexture functionality

* Non-NVIDIA support

* Disable some features that should not affect draw texture (slow path)

* Remove space from shader source

* Match 2D engine names

* Fix resolution scale and add missing XML docs

* Disable transform feedback for draw texture fallback
---
 Ryujinx.Graphics.GAL/Extents2DF.cs            |  18 +++
 Ryujinx.Graphics.GAL/IPipeline.cs             |   1 +
 .../Multithreading/CommandHelper.cs           |   2 +
 .../Multithreading/CommandType.cs             |   1 +
 .../Commands/DrawTextureCommand.cs            |  31 ++++
 .../Multithreading/ThreadedPipeline.cs        |   6 +
 .../Engine/Threed/DrawManager.cs              |  59 ++++++++
 .../Engine/Threed/ThreedClass.cs              |  10 ++
 .../Engine/Threed/ThreedClassState.cs         |  13 +-
 .../Image/TextureBindingsManager.cs           |  16 ++
 Ryujinx.Graphics.Gpu/Image/TextureManager.cs  |  10 ++
 .../DrawTextureEmulation.cs                   | 138 ++++++++++++++++++
 Ryujinx.Graphics.OpenGL/HwCapabilities.cs     |   2 +
 Ryujinx.Graphics.OpenGL/Pipeline.cs           | 128 ++++++++++++++--
 14 files changed, 421 insertions(+), 14 deletions(-)
 create mode 100644 Ryujinx.Graphics.GAL/Extents2DF.cs
 create mode 100644 Ryujinx.Graphics.GAL/Multithreading/Commands/DrawTextureCommand.cs
 create mode 100644 Ryujinx.Graphics.OpenGL/DrawTextureEmulation.cs

diff --git a/Ryujinx.Graphics.GAL/Extents2DF.cs b/Ryujinx.Graphics.GAL/Extents2DF.cs
new file mode 100644
index 0000000000..8fb263c479
--- /dev/null
+++ b/Ryujinx.Graphics.GAL/Extents2DF.cs
@@ -0,0 +1,18 @@
+namespace Ryujinx.Graphics.GAL
+{
+    public struct Extents2DF
+    {
+        public float X1 { get; }
+        public float Y1 { get; }
+        public float X2 { get; }
+        public float Y2 { get; }
+
+        public Extents2DF(float x1, float y1, float x2, float y2)
+        {
+            X1 = x1;
+            Y1 = y1;
+            X2 = x2;
+            Y2 = y2;
+        }
+    }
+}
\ No newline at end of file
diff --git a/Ryujinx.Graphics.GAL/IPipeline.cs b/Ryujinx.Graphics.GAL/IPipeline.cs
index bfc432b1cf..b7da2c2167 100644
--- a/Ryujinx.Graphics.GAL/IPipeline.cs
+++ b/Ryujinx.Graphics.GAL/IPipeline.cs
@@ -32,6 +32,7 @@ namespace Ryujinx.Graphics.GAL
             int firstIndex,
             int firstVertex,
             int firstInstance);
+        void DrawTexture(ITexture texture, ISampler sampler, Extents2DF srcRegion, Extents2DF dstRegion);
 
         void EndTransformFeedback();
 
diff --git a/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs b/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs
index 47ceeb7d55..6111e32ceb 100644
--- a/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs
+++ b/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs
@@ -143,6 +143,8 @@ namespace Ryujinx.Graphics.GAL.Multithreading
                 DrawCommand.Run(ref GetCommand<DrawCommand>(memory), threaded, renderer);
             _lookup[(int)CommandType.DrawIndexed] = (Span<byte> memory, ThreadedRenderer threaded, IRenderer renderer) =>
                 DrawIndexedCommand.Run(ref GetCommand<DrawIndexedCommand>(memory), threaded, renderer);
+            _lookup[(int)CommandType.DrawTexture] = (Span<byte> memory, ThreadedRenderer threaded, IRenderer renderer) =>
+                DrawTextureCommand.Run(ref GetCommand<DrawTextureCommand>(memory), threaded, renderer);
             _lookup[(int)CommandType.EndHostConditionalRendering] = (Span<byte> memory, ThreadedRenderer threaded, IRenderer renderer) =>
                 EndHostConditionalRenderingCommand.Run(renderer);
             _lookup[(int)CommandType.EndTransformFeedback] = (Span<byte> memory, ThreadedRenderer threaded, IRenderer renderer) =>
diff --git a/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs b/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs
index ac73a3febe..4bceaa1edc 100644
--- a/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs
+++ b/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs
@@ -53,6 +53,7 @@
         DispatchCompute,
         Draw,
         DrawIndexed,
+        DrawTexture,
         EndHostConditionalRendering,
         EndTransformFeedback,
         MultiDrawIndirectCount,
diff --git a/Ryujinx.Graphics.GAL/Multithreading/Commands/DrawTextureCommand.cs b/Ryujinx.Graphics.GAL/Multithreading/Commands/DrawTextureCommand.cs
new file mode 100644
index 0000000000..41a852bb33
--- /dev/null
+++ b/Ryujinx.Graphics.GAL/Multithreading/Commands/DrawTextureCommand.cs
@@ -0,0 +1,31 @@
+using Ryujinx.Graphics.GAL.Multithreading.Model;
+using Ryujinx.Graphics.GAL.Multithreading.Resources;
+
+namespace Ryujinx.Graphics.GAL.Multithreading.Commands
+{
+    struct DrawTextureCommand : IGALCommand
+    {
+        public CommandType CommandType => CommandType.DrawTexture;
+        private TableRef<ITexture> _texture;
+        private TableRef<ISampler> _sampler;
+        private Extents2DF _srcRegion;
+        private Extents2DF _dstRegion;
+
+        public void Set(TableRef<ITexture> texture, TableRef<ISampler> sampler, Extents2DF srcRegion, Extents2DF dstRegion)
+        {
+            _texture = texture;
+            _sampler = sampler;
+            _srcRegion = srcRegion;
+            _dstRegion = dstRegion;
+        }
+
+        public static void Run(ref DrawTextureCommand command, ThreadedRenderer threaded, IRenderer renderer)
+        {
+            renderer.Pipeline.DrawTexture(
+                command._texture.GetAs<ThreadedTexture>(threaded)?.Base,
+                command._sampler.GetAs<ThreadedSampler>(threaded)?.Base,
+                command._srcRegion,
+                command._dstRegion);
+        }
+    }
+}
diff --git a/Ryujinx.Graphics.GAL/Multithreading/ThreadedPipeline.cs b/Ryujinx.Graphics.GAL/Multithreading/ThreadedPipeline.cs
index 3c39a77f94..63a29b1bed 100644
--- a/Ryujinx.Graphics.GAL/Multithreading/ThreadedPipeline.cs
+++ b/Ryujinx.Graphics.GAL/Multithreading/ThreadedPipeline.cs
@@ -83,6 +83,12 @@ namespace Ryujinx.Graphics.GAL.Multithreading
             _renderer.QueueCommand();
         }
 
+        public void DrawTexture(ITexture texture, ISampler sampler, Extents2DF srcRegion, Extents2DF dstRegion)
+        {
+            _renderer.New<DrawTextureCommand>().Set(Ref(texture), Ref(sampler), srcRegion, dstRegion);
+            _renderer.QueueCommand();
+        }
+
         public void EndHostConditionalRendering()
         {
             _renderer.New<EndHostConditionalRenderingCommand>();
diff --git a/Ryujinx.Graphics.Gpu/Engine/Threed/DrawManager.cs b/Ryujinx.Graphics.Gpu/Engine/Threed/DrawManager.cs
index a060c6c961..518e71ade5 100644
--- a/Ryujinx.Graphics.Gpu/Engine/Threed/DrawManager.cs
+++ b/Ryujinx.Graphics.Gpu/Engine/Threed/DrawManager.cs
@@ -319,6 +319,65 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
             _drawState.DrawIndexed = oldDrawIndexed;
         }
 
+        /// <summary>
+        /// Performs a texture draw with a source texture and sampler ID, along with source
+        /// and destination coordinates and sizes.
+        /// </summary>
+        /// <param name="engine">3D engine where this method is being called</param>
+        /// <param name="argument">Method call argument</param>
+        public void DrawTexture(ThreedClass engine, int argument)
+        {
+            static float FixedToFloat(int fixedValue)
+            {
+                return fixedValue * (1f / 4096);
+            }
+
+            float dstX0 = FixedToFloat(_state.State.DrawTextureDstX);
+            float dstY0 = FixedToFloat(_state.State.DrawTextureDstY);
+            float dstWidth = FixedToFloat(_state.State.DrawTextureDstWidth);
+            float dstHeight = FixedToFloat(_state.State.DrawTextureDstHeight);
+
+            // TODO: Confirm behaviour on hardware.
+            // When this is active, the origin appears to be on the bottom.
+            if (_state.State.YControl.HasFlag(YControl.NegateY))
+            {
+                dstY0 -= dstHeight;
+            }
+
+            float dstX1 = dstX0 + dstWidth;
+            float dstY1 = dstY0 + dstHeight;
+
+            float srcX0 = FixedToFloat(_state.State.DrawTextureSrcX);
+            float srcY0 = FixedToFloat(_state.State.DrawTextureSrcY);
+            float srcX1 = ((float)_state.State.DrawTextureDuDx / (1UL << 32)) * dstWidth + srcX0;
+            float srcY1 = ((float)_state.State.DrawTextureDvDy / (1UL << 32)) * dstHeight + srcY0;
+
+            engine.UpdateState();
+
+            int textureId = _state.State.DrawTextureTextureId;
+            int samplerId = _state.State.DrawTextureSamplerId;
+
+            (var texture, var sampler) = _channel.TextureManager.GetGraphicsTextureAndSampler(textureId, samplerId);
+
+            srcX0 *= texture.ScaleFactor;
+            srcY0 *= texture.ScaleFactor;
+            srcX1 *= texture.ScaleFactor;
+            srcY1 *= texture.ScaleFactor;
+
+            float dstScale = _channel.TextureManager.RenderTargetScale;
+
+            dstX0 *= dstScale;
+            dstY0 *= dstScale;
+            dstX1 *= dstScale;
+            dstY1 *= dstScale;
+
+            _context.Renderer.Pipeline.DrawTexture(
+                texture?.HostTexture,
+                sampler?.HostSampler,
+                new Extents2DF(srcX0, srcY0, srcX1, srcY1),
+                new Extents2DF(dstX0, dstY0, dstX1, dstY1));
+        }
+
         /// <summary>
         /// Performs a indirect multi-draw, with parameters from a GPU buffer.
         /// </summary>
diff --git a/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClass.cs b/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClass.cs
index 5478704acb..f3061c73ad 100644
--- a/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClass.cs
+++ b/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClass.cs
@@ -36,6 +36,7 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
                 { nameof(ThreedClassState.SyncpointAction), new RwCallback(IncrementSyncpoint, null) },
                 { nameof(ThreedClassState.TextureBarrier), new RwCallback(TextureBarrier, null) },
                 { nameof(ThreedClassState.TextureBarrierTiled), new RwCallback(TextureBarrierTiled, null) },
+                { nameof(ThreedClassState.DrawTextureSrcY), new RwCallback(DrawTexture, null) },
                 { nameof(ThreedClassState.VbElementU8), new RwCallback(VbElementU8, null) },
                 { nameof(ThreedClassState.VbElementU16), new RwCallback(VbElementU16, null) },
                 { nameof(ThreedClassState.VbElementU32), new RwCallback(VbElementU32, null) },
@@ -251,6 +252,15 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
             _context.Renderer.Pipeline.TextureBarrierTiled();
         }
 
+        /// <summary>
+        /// Draws a texture, without needing to specify shader programs.
+        /// </summary>
+        /// <param name="argument">Method call argument</param>
+        private void DrawTexture(int argument)
+        {
+            _drawManager.DrawTexture(this, argument);
+        }
+
         /// <summary>
         /// Pushes four 8-bit index buffer elements.
         /// </summary>
diff --git a/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClassState.cs b/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClassState.cs
index 58bc0957fb..165f507216 100644
--- a/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClassState.cs
+++ b/Ryujinx.Graphics.Gpu/Engine/Threed/ThreedClassState.cs
@@ -743,7 +743,18 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
         public fixed uint ReservedF94[19];
         public RtDepthStencilState RtDepthStencilState;
         public ScreenScissorState ScreenScissorState;
-        public fixed uint ReservedFFC[89];
+        public fixed uint ReservedFFC[33];
+        public int DrawTextureDstX;
+        public int DrawTextureDstY;
+        public int DrawTextureDstWidth;
+        public int DrawTextureDstHeight;
+        public long DrawTextureDuDx;
+        public long DrawTextureDvDy;
+        public int DrawTextureSamplerId;
+        public int DrawTextureTextureId;
+        public int DrawTextureSrcX;
+        public int DrawTextureSrcY;
+        public fixed uint Reserved10B0[44];
         public Array16<VertexAttribState> VertexAttribState;
         public fixed uint Reserved11A0[31];
         public RtControl RtControl;
diff --git a/Ryujinx.Graphics.Gpu/Image/TextureBindingsManager.cs b/Ryujinx.Graphics.Gpu/Image/TextureBindingsManager.cs
index 5862ea712e..621dc2e740 100644
--- a/Ryujinx.Graphics.Gpu/Image/TextureBindingsManager.cs
+++ b/Ryujinx.Graphics.Gpu/Image/TextureBindingsManager.cs
@@ -201,6 +201,22 @@ namespace Ryujinx.Graphics.Gpu.Image
             }
         }
 
+        /// <summary>
+        /// Gets a texture and a sampler from their respective pools from a texture ID and a sampler ID.
+        /// </summary>
+        /// <param name="textureId">ID of the texture</param>
+        /// <param name="samplerId">ID of the sampler</param>
+        public (Texture, Sampler) GetTextureAndSampler(int textureId, int samplerId)
+        {
+            ulong texturePoolAddress = _texturePoolAddress;
+
+            TexturePool texturePool = texturePoolAddress != 0
+                ? _texturePoolCache.FindOrCreate(_channel, texturePoolAddress, _texturePoolMaximumId)
+                : null;
+
+            return (texturePool.Get(textureId), _samplerPool.Get(samplerId));
+        }
+
         /// <summary>
         /// Updates the texture scale for a given texture or image.
         /// </summary>
diff --git a/Ryujinx.Graphics.Gpu/Image/TextureManager.cs b/Ryujinx.Graphics.Gpu/Image/TextureManager.cs
index a6373872b9..70cb57d09a 100644
--- a/Ryujinx.Graphics.Gpu/Image/TextureManager.cs
+++ b/Ryujinx.Graphics.Gpu/Image/TextureManager.cs
@@ -308,6 +308,16 @@ namespace Ryujinx.Graphics.Gpu.Image
             RenderTargetScale = targetScale;
         }
 
+        /// <summary>
+        /// Gets a texture and a sampler from their respective pools from a texture ID and a sampler ID.
+        /// </summary>
+        /// <param name="textureId">ID of the texture</param>
+        /// <param name="samplerId">ID of the sampler</param>
+        public (Texture, Sampler) GetGraphicsTextureAndSampler(int textureId, int samplerId)
+        {
+            return _gpBindingsManager.GetTextureAndSampler(textureId, samplerId);
+        }
+
         /// <summary>
         /// Commits bindings on the compute pipeline.
         /// </summary>
diff --git a/Ryujinx.Graphics.OpenGL/DrawTextureEmulation.cs b/Ryujinx.Graphics.OpenGL/DrawTextureEmulation.cs
new file mode 100644
index 0000000000..509e20fe25
--- /dev/null
+++ b/Ryujinx.Graphics.OpenGL/DrawTextureEmulation.cs
@@ -0,0 +1,138 @@
+using OpenTK.Graphics.OpenGL;
+using Ryujinx.Graphics.OpenGL.Image;
+using System;
+
+namespace Ryujinx.Graphics.OpenGL
+{
+    class DrawTextureEmulation
+    {
+        private const string VertexShader = @"#version 430 core
+
+uniform float srcX0;
+uniform float srcY0;
+uniform float srcX1;
+uniform float srcY1;
+
+layout (location = 0) out vec2 texcoord;
+
+void main()
+{
+    bool x1 = (gl_VertexID & 1) != 0;
+    bool y1 = (gl_VertexID & 2) != 0;
+    gl_Position = vec4(x1 ? 1 : -1, y1 ? -1 : 1, 0, 1);
+    texcoord = vec2(x1 ? srcX1 : srcX0, y1 ? srcY1 : srcY0);
+}";
+
+        private const string FragmentShader = @"#version 430 core
+
+layout (location = 0) uniform sampler2D tex;
+
+layout (location = 0) in vec2 texcoord;
+layout (location = 0) out vec4 colour;
+
+void main()
+{
+    colour = texture(tex, texcoord);
+}";
+
+        private int _vsHandle;
+        private int _fsHandle;
+        private int _programHandle;
+        private int _uniformSrcX0Location;
+        private int _uniformSrcY0Location;
+        private int _uniformSrcX1Location;
+        private int _uniformSrcY1Location;
+        private bool _initialized;
+
+        public void Draw(
+            TextureView texture,
+            Sampler sampler,
+            float x0,
+            float y0,
+            float x1,
+            float y1,
+            float s0,
+            float t0,
+            float s1,
+            float t1)
+        {
+            EnsureInitialized();
+
+            GL.UseProgram(_programHandle);
+
+            texture.Bind(0);
+            sampler.Bind(0);
+
+            if (x0 > x1)
+            {
+                float temp = s0;
+                s0 = s1;
+                s1 = temp;
+            }
+
+            if (y0 > y1)
+            {
+                float temp = t0;
+                t0 = t1;
+                t1 = temp;
+            }
+
+            GL.Uniform1(_uniformSrcX0Location, s0);
+            GL.Uniform1(_uniformSrcY0Location, t0);
+            GL.Uniform1(_uniformSrcX1Location, s1);
+            GL.Uniform1(_uniformSrcY1Location, t1);
+
+            GL.ViewportIndexed(0, MathF.Min(x0, x1), MathF.Min(y0, y1), MathF.Abs(x1 - x0), MathF.Abs(y1 - y0));
+
+            GL.DrawArrays(PrimitiveType.TriangleStrip, 0, 4);
+        }
+
+        private void EnsureInitialized()
+        {
+            if (_initialized)
+            {
+                return;
+            }
+
+            _initialized = true;
+
+            _vsHandle = GL.CreateShader(ShaderType.VertexShader);
+            _fsHandle = GL.CreateShader(ShaderType.FragmentShader);
+
+            GL.ShaderSource(_vsHandle, VertexShader);
+            GL.ShaderSource(_fsHandle, FragmentShader);
+
+            GL.CompileShader(_vsHandle);
+            GL.CompileShader(_fsHandle);
+
+            _programHandle = GL.CreateProgram();
+
+            GL.AttachShader(_programHandle, _vsHandle);
+            GL.AttachShader(_programHandle, _fsHandle);
+
+            GL.LinkProgram(_programHandle);
+
+            GL.DetachShader(_programHandle, _vsHandle);
+            GL.DetachShader(_programHandle, _fsHandle);
+
+            _uniformSrcX0Location = GL.GetUniformLocation(_programHandle, "srcX0");
+            _uniformSrcY0Location = GL.GetUniformLocation(_programHandle, "srcY0");
+            _uniformSrcX1Location = GL.GetUniformLocation(_programHandle, "srcX1");
+            _uniformSrcY1Location = GL.GetUniformLocation(_programHandle, "srcY1");
+        }
+
+        public void Dispose()
+        {
+            if (!_initialized)
+            {
+                return;
+            }
+
+            GL.DeleteShader(_vsHandle);
+            GL.DeleteShader(_fsHandle);
+            GL.DeleteProgram(_programHandle);
+
+            _initialized = false;
+        }
+    }
+}
diff --git a/Ryujinx.Graphics.OpenGL/HwCapabilities.cs b/Ryujinx.Graphics.OpenGL/HwCapabilities.cs
index ec9bd4a201..773c9f634e 100644
--- a/Ryujinx.Graphics.OpenGL/HwCapabilities.cs
+++ b/Ryujinx.Graphics.OpenGL/HwCapabilities.cs
@@ -6,6 +6,7 @@ namespace Ryujinx.Graphics.OpenGL
     static class HwCapabilities
     {
         private static readonly Lazy<bool> _supportsAstcCompression           = new Lazy<bool>(() => HasExtension("GL_KHR_texture_compression_astc_ldr"));
+        private static readonly Lazy<bool> _supportsDrawTexture               = new Lazy<bool>(() => HasExtension("GL_NV_draw_texture"));
         private static readonly Lazy<bool> _supportsFragmentShaderInterlock   = new Lazy<bool>(() => HasExtension("GL_ARB_fragment_shader_interlock"));
         private static readonly Lazy<bool> _supportsFragmentShaderOrdering    = new Lazy<bool>(() => HasExtension("GL_INTEL_fragment_shader_ordering"));
         private static readonly Lazy<bool> _supportsImageLoadFormatted        = new Lazy<bool>(() => HasExtension("GL_EXT_shader_image_load_formatted"));
@@ -43,6 +44,7 @@ namespace Ryujinx.Graphics.OpenGL
         public static bool UsePersistentBufferForFlush       => _gpuVendor.Value == GpuVendor.AmdWindows || _gpuVendor.Value == GpuVendor.Nvidia;
 
         public static bool SupportsAstcCompression           => _supportsAstcCompression.Value;
+        public static bool SupportsDrawTexture               => _supportsDrawTexture.Value;
         public static bool SupportsFragmentShaderInterlock   => _supportsFragmentShaderInterlock.Value;
         public static bool SupportsFragmentShaderOrdering    => _supportsFragmentShaderOrdering.Value;
         public static bool SupportsImageLoadFormatted        => _supportsImageLoadFormatted.Value;
diff --git a/Ryujinx.Graphics.OpenGL/Pipeline.cs b/Ryujinx.Graphics.OpenGL/Pipeline.cs
index d0a509b474..aafc4db80d 100644
--- a/Ryujinx.Graphics.OpenGL/Pipeline.cs
+++ b/Ryujinx.Graphics.OpenGL/Pipeline.cs
@@ -12,6 +12,8 @@ namespace Ryujinx.Graphics.OpenGL
 {
     class Pipeline : IPipeline, IDisposable
     {
+        private readonly DrawTextureEmulation _drawTexture;
+
         internal ulong DrawCount { get; private set; }
 
         private Program _program;
@@ -29,6 +31,12 @@ namespace Ryujinx.Graphics.OpenGL
 
         private int _stencilFrontMask;
         private bool _depthMask;
+        private bool _depthTestEnable;
+        private bool _stencilTestEnable;
+        private bool _cullEnable;
+
+        private float[] _viewportArray = Array.Empty<float>();
+        private double[] _depthRangeArray = Array.Empty<double>();
 
         private int _boundDrawFramebuffer;
         private int _boundReadFramebuffer;
@@ -47,6 +55,7 @@ namespace Ryujinx.Graphics.OpenGL
         private Vector4<float>[] _renderScale = new Vector4<float>[65];
 
         private TextureBase _unit0Texture;
+        private Sampler _unit0Sampler;
 
         private FrontFaceDirection _frontFace;
         private ClipOrigin _clipOrigin;
@@ -67,6 +76,7 @@ namespace Ryujinx.Graphics.OpenGL
 
         internal Pipeline()
         {
+            _drawTexture = new DrawTextureEmulation();
             _rasterizerDiscard = false;
             _clipOrigin = ClipOrigin.LowerLeft;
             _clipDepthMode = ClipDepthMode.NegativeOneToOne;
@@ -544,6 +554,91 @@ namespace Ryujinx.Graphics.OpenGL
             }
         }
 
+        public void DrawTexture(ITexture texture, ISampler sampler, Extents2DF srcRegion, Extents2DF dstRegion)
+        {
+            if (texture is TextureView view && sampler is Sampler samp)
+            {
+                if (HwCapabilities.SupportsDrawTexture)
+                {
+                    GL.NV.DrawTexture(
+                        view.Handle,
+                        samp.Handle,
+                        dstRegion.X1,
+                        dstRegion.Y1,
+                        dstRegion.X2,
+                        dstRegion.Y2,
+                        0,
+                        srcRegion.X1 / view.Width,
+                        srcRegion.Y1 / view.Height,
+                        srcRegion.X2 / view.Width,
+                        srcRegion.Y2 / view.Height);
+                }
+                else
+                {
+                    static void Disable(EnableCap cap, bool enabled)
+                    {
+                        if (enabled)
+                        {
+                            GL.Disable(cap);
+                        }
+                    }
+
+                    static void Enable(EnableCap cap, bool enabled)
+                    {
+                        if (enabled)
+                        {
+                            GL.Enable(cap);
+                        }
+                    }
+
+                    Disable(EnableCap.CullFace, _cullEnable);
+                    Disable(EnableCap.StencilTest, _stencilTestEnable);
+                    Disable(EnableCap.DepthTest, _depthTestEnable);
+
+                    if (_depthMask)
+                    {
+                        GL.DepthMask(false);
+                    }
+
+                    if (_tfEnabled)
+                    {
+                        GL.EndTransformFeedback();
+                    }
+
+                    _drawTexture.Draw(
+                        view,
+                        samp,
+                        dstRegion.X1,
+                        dstRegion.Y1,
+                        dstRegion.X2,
+                        dstRegion.Y2,
+                        srcRegion.X1 / view.Width,
+                        srcRegion.Y1 / view.Height,
+                        srcRegion.X2 / view.Width,
+                        srcRegion.Y2 / view.Height);
+
+                    _program?.Bind();
+                    _unit0Sampler?.Bind(0);
+
+                    GL.ViewportArray(0, 1, _viewportArray);
+
+                    Enable(EnableCap.CullFace, _cullEnable);
+                    Enable(EnableCap.StencilTest, _stencilTestEnable);
+                    Enable(EnableCap.DepthTest, _depthTestEnable);
+
+                    if (_depthMask)
+                    {
+                        GL.DepthMask(true);
+                    }
+
+                    if (_tfEnabled)
+                    {
+                        GL.BeginTransformFeedback(_tfTopology);
+                    }
+                }
+            }
+        }
+
         public void EndTransformFeedback()
         {
             GL.EndTransformFeedback();
@@ -754,10 +849,13 @@ namespace Ryujinx.Graphics.OpenGL
 
             GL.DepthMask(depthTest.WriteEnable);
             _depthMask = depthTest.WriteEnable;
+            _depthTestEnable = depthTest.TestEnable;
         }
 
         public void SetFaceCulling(bool enable, Face face)
         {
+            _cullEnable = enable;
+
             if (!enable)
             {
                 GL.Disable(EnableCap.CullFace);
@@ -994,7 +1092,14 @@ namespace Ryujinx.Graphics.OpenGL
                 return;
             }
 
-            ((Sampler)sampler).Bind(binding);
+            Sampler samp = (Sampler)sampler;
+
+            if (binding == 0)
+            {
+                _unit0Sampler = samp;
+            }
+
+            samp.Bind(binding);
         }
 
         public void SetScissor(int index, bool enable, int x, int y, int width, int height)
@@ -1023,6 +1128,8 @@ namespace Ryujinx.Graphics.OpenGL
 
         public void SetStencilTest(StencilTestDescriptor stencilTest)
         {
+            _stencilTestEnable = stencilTest.TestEnable;
+
             if (!stencilTest.TestEnable)
             {
                 GL.Disable(EnableCap.StencilTest);
@@ -1152,9 +1259,11 @@ namespace Ryujinx.Graphics.OpenGL
 
         public void SetViewports(int first, ReadOnlySpan<Viewport> viewports)
         {
-            float[] viewportArray = new float[viewports.Length * 4];
+            Array.Resize(ref _viewportArray, viewports.Length * 4);
+            Array.Resize(ref _depthRangeArray, viewports.Length * 2);
 
-            double[] depthRangeArray = new double[viewports.Length * 2];
+            float[] viewportArray = _viewportArray;
+            double[] depthRangeArray = _depthRangeArray;
 
             for (int index = 0; index < viewports.Length; index++)
             {
@@ -1186,7 +1295,6 @@ namespace Ryujinx.Graphics.OpenGL
             SetOrigin(flipY ? ClipOrigin.UpperLeft : ClipOrigin.LowerLeft);
 
             GL.ViewportArray(first, viewports.Length, viewportArray);
-
             GL.DepthRangeArray(first, viewports.Length, depthRangeArray);
         }
 
@@ -1307,10 +1415,7 @@ namespace Ryujinx.Graphics.OpenGL
 
         private void PrepareForDispatch()
         {
-            if (_unit0Texture != null)
-            {
-                _unit0Texture.Bind(0);
-            }
+            _unit0Texture?.Bind(0);
         }
 
         private void PreDraw()
@@ -1318,11 +1423,7 @@ namespace Ryujinx.Graphics.OpenGL
             DrawCount++;
 
             _vertexArray.Validate();
-
-            if (_unit0Texture != null)
-            {
-                _unit0Texture.Bind(0);
-            }
+            _unit0Texture?.Bind(0);
         }
 
         private void PostDraw()
@@ -1438,6 +1539,7 @@ namespace Ryujinx.Graphics.OpenGL
             _activeConditionalRender?.ReleaseHostAccess();
             _framebuffer?.Dispose();
             _vertexArray?.Dispose();
+            _drawTexture.Dispose();
         }
     }
 }