New NVDEC and VIC implementation (#1384)

* Initial NVDEC and VIC implementation

* Update FFmpeg.AutoGen to 4.3.0

* Add nvdec dependencies for Windows

* Unify some VP9 structures

* Rename VP9 structure fields

* Improvements to Video API

* XML docs for Common.Memory

* Remove now unused or redundant overloads from MemoryAccessor

* NVDEC UV surface read/write scalar paths

* Add FIXME comments about hacky things/stuff that will need to be fixed in the future

* Cleaned up VP9 memory allocation

* Remove some debug logs

* Rename some VP9 structs

* Remove unused struct

* No need to compile Ryujinx.Graphics.Host1x with unsafe anymore

* Name AsyncWorkQueue threads to make debugging easier

* Make Vp9PictureInfo a ref struct

* LayoutConverter no longer needs the depth argument (broken by rebase)

* Pooling of VP9 buffers, plus fix a memory leak on VP9

* Really wish VS could rename projects properly...

* Address feedback

* Remove using

* Catch OperationCanceledException

* Add licensing informations

* Add THIRDPARTY.md to release too

Co-authored-by: Thog <me@thog.eu>

8 files changed, 81 insertions, 45 deletions

diff --git a/Ryujinx.Graphics.Gpu/Engine/Compute.cs b/Ryujinx.Graphics.Gpu/Engine/Compute.cs
index e40984af..60fba006 100644
--- a/Ryujinx.Graphics.Gpu/Engine/Compute.cs
+++ b/Ryujinx.Graphics.Gpu/Engine/Compute.cs
@@ -67,7 +67,7 @@ namespace Ryujinx.Graphics.Gpu.Engine
 
             TextureManager.SetComputeTextureBufferIndex(state.Get<int>(MethodOffset.TextureBufferIndex));
 
-            ShaderProgramInfo info = cs.Shaders[0].Program.Info;            
+            ShaderProgramInfo info = cs.Shaders[0].Program.Info;
 
             for (int index = 0; index < info.CBuffers.Count; index++)
             {
diff --git a/Ryujinx.Graphics.Gpu/Engine/MethodConditionalRendering.cs b/Ryujinx.Graphics.Gpu/Engine/MethodConditionalRendering.cs
index c8d47b9f..225c732e 100644
--- a/Ryujinx.Graphics.Gpu/Engine/MethodConditionalRendering.cs
+++ b/Ryujinx.Graphics.Gpu/Engine/MethodConditionalRendering.cs
@@ -63,7 +63,7 @@ namespace Ryujinx.Graphics.Gpu.Engine
             else
             {
                 evt.Flush();
-                return (_context.MemoryAccessor.ReadUInt64(gpuVa) != 0) ? ConditionalRenderEnabled.True : ConditionalRenderEnabled.False;
+                return (_context.MemoryAccessor.Read<ulong>(gpuVa) != 0) ? ConditionalRenderEnabled.True : ConditionalRenderEnabled.False;
             }
         }
 
@@ -87,11 +87,11 @@ namespace Ryujinx.Graphics.Gpu.Engine
 
             if (evt != null && evt2 == null)
             {
-                useHost = _context.Renderer.Pipeline.TryHostConditionalRendering(evt, _context.MemoryAccessor.ReadUInt64(gpuVa + 16), isEqual);
+                useHost = _context.Renderer.Pipeline.TryHostConditionalRendering(evt, _context.MemoryAccessor.Read<ulong>(gpuVa + 16), isEqual);
             }
             else if (evt == null && evt2 != null)
             {
-                useHost = _context.Renderer.Pipeline.TryHostConditionalRendering(evt2, _context.MemoryAccessor.ReadUInt64(gpuVa), isEqual);
+                useHost = _context.Renderer.Pipeline.TryHostConditionalRendering(evt2, _context.MemoryAccessor.Read<ulong>(gpuVa), isEqual);
             }
             else
             {
@@ -107,8 +107,8 @@ namespace Ryujinx.Graphics.Gpu.Engine
                 evt?.Flush();
                 evt2?.Flush();
 
-                ulong x = _context.MemoryAccessor.ReadUInt64(gpuVa);
-                ulong y = _context.MemoryAccessor.ReadUInt64(gpuVa + 16);
+                ulong x = _context.MemoryAccessor.Read<ulong>(gpuVa);
+                ulong y = _context.MemoryAccessor.Read<ulong>(gpuVa + 16);
 
                 return (isEqual ? x == y : x != y) ? ConditionalRenderEnabled.True : ConditionalRenderEnabled.False;
             }
diff --git a/Ryujinx.Graphics.Gpu/Engine/Methods.cs b/Ryujinx.Graphics.Gpu/Engine/Methods.cs
index 06298cdf..d5b11c2c 100644
--- a/Ryujinx.Graphics.Gpu/Engine/Methods.cs
+++ b/Ryujinx.Graphics.Gpu/Engine/Methods.cs
@@ -466,7 +466,7 @@ namespace Ryujinx.Graphics.Gpu.Engine
 
             bool   flipY  = yControl.HasFlag(YControl.NegateY);
             Origin origin = yControl.HasFlag(YControl.TriangleRastFlip) ? Origin.LowerLeft : Origin.UpperLeft;
-            
+
             _context.Renderer.Pipeline.SetOrigin(origin);
 
             // The triangle rast flip flag only affects rasterization, the viewport is not flipped.
diff --git a/Ryujinx.Graphics.Gpu/GpuContext.cs b/Ryujinx.Graphics.Gpu/GpuContext.cs
index 5e117831..b07694b9 100644
--- a/Ryujinx.Graphics.Gpu/GpuContext.cs
+++ b/Ryujinx.Graphics.Gpu/GpuContext.cs
@@ -77,7 +77,7 @@ namespace Ryujinx.Graphics.Gpu
         {
             Renderer = renderer;
 
-            MemoryManager = new MemoryManager();
+            MemoryManager = new MemoryManager(this);
 
             MemoryAccessor = new MemoryAccessor(this);
 
diff --git a/Ryujinx.Graphics.Gpu/Image/TextureManager.cs b/Ryujinx.Graphics.Gpu/Image/TextureManager.cs
index ccd56ae2..69bee541 100644
--- a/Ryujinx.Graphics.Gpu/Image/TextureManager.cs
+++ b/Ryujinx.Graphics.Gpu/Image/TextureManager.cs
@@ -643,6 +643,8 @@ namespace Ryujinx.Graphics.Gpu.Image
                         overlap.ChangeSize(info.Width, info.Height, info.DepthOrLayers);
                     }
 
+                    overlap.SynchronizeMemory();
+
                     return overlap;
                 }
             }
diff --git a/Ryujinx.Graphics.Gpu/Memory/MemoryAccessor.cs b/Ryujinx.Graphics.Gpu/Memory/MemoryAccessor.cs
index 38f448d9..5cc8ec24 100644
--- a/Ryujinx.Graphics.Gpu/Memory/MemoryAccessor.cs
+++ b/Ryujinx.Graphics.Gpu/Memory/MemoryAccessor.cs
@@ -59,42 +59,6 @@ namespace Ryujinx.Graphics.Gpu.Memory
         }
 
         /// <summary>
-        /// Reads a 32-bits signed integer from GPU mapped memory.
-        /// </summary>
-        /// <param name="gpuVa">GPU virtual address where the value is located</param>
-        /// <returns>The value at the specified memory location</returns>
-        public int ReadInt32(ulong gpuVa)
-        {
-            ulong processVa = _context.MemoryManager.Translate(gpuVa);
-
-            return _context.PhysicalMemory.Read<int>(processVa);
-        }
-
-        /// <summary>
-        /// Reads a 64-bits unsigned integer from GPU mapped memory.
-        /// </summary>
-        /// <param name="gpuVa">GPU virtual address where the value is located</param>
-        /// <returns>The value at the specified memory location</returns>
-        public ulong ReadUInt64(ulong gpuVa)
-        {
-            ulong processVa = _context.MemoryManager.Translate(gpuVa);
-
-            return _context.PhysicalMemory.Read<ulong>(processVa);
-        }
-
-        /// <summary>
-        /// Reads a 8-bits unsigned integer from GPU mapped memory.
-        /// </summary>
-        /// <param name="gpuVa">GPU virtual address where the value is located</param>
-        /// <param name="value">The value to be written</param>
-        public void WriteByte(ulong gpuVa, byte value)
-        {
-            ulong processVa = _context.MemoryManager.Translate(gpuVa);
-
-            _context.PhysicalMemory.Write(processVa, MemoryMarshal.CreateSpan(ref value, 1));
-        }
-
-        /// <summary>
         /// Writes a 32-bits signed integer to GPU mapped memory.
         /// </summary>
         /// <param name="gpuVa">GPU virtual address to write the value into</param>
diff --git a/Ryujinx.Graphics.Gpu/Memory/MemoryManager.cs b/Ryujinx.Graphics.Gpu/Memory/MemoryManager.cs
index a9a8fbac..2d988f8d 100644
--- a/Ryujinx.Graphics.Gpu/Memory/MemoryManager.cs
+++ b/Ryujinx.Graphics.Gpu/Memory/MemoryManager.cs
@@ -1,4 +1,7 @@
+using Ryujinx.Cpu;
 using System;
+using System.Runtime.CompilerServices;
+using System.Runtime.InteropServices;
 
 namespace Ryujinx.Graphics.Gpu.Memory
 {
@@ -33,15 +36,70 @@ namespace Ryujinx.Graphics.Gpu.Memory
 
         public event EventHandler<UnmapEventArgs> MemoryUnmapped;
 
+        private GpuContext _context;
+
         /// <summary>
         /// Creates a new instance of the GPU memory manager.
         /// </summary>
-        public MemoryManager()
+        public MemoryManager(GpuContext context)
         {
+            _context = context;
             _pageTable = new ulong[PtLvl0Size][];
         }
 
         /// <summary>
+        /// Reads data from GPU mapped memory.
+        /// </summary>
+        /// <typeparam name="T">Type of the data</typeparam>
+        /// <param name="gpuVa">GPU virtual address where the data is located</param>
+        /// <returns>The data at the specified memory location</returns>
+        public T Read<T>(ulong gpuVa) where T : unmanaged
+        {
+            ulong processVa = Translate(gpuVa);
+
+            return MemoryMarshal.Cast<byte, T>(_context.PhysicalMemory.GetSpan(processVa, Unsafe.SizeOf<T>()))[0];
+        }
+
+        /// <summary>
+        /// Gets a read-only span of data from GPU mapped memory.
+        /// This reads as much data as possible, up to the specified maximum size.
+        /// </summary>
+        /// <param name="gpuVa">GPU virtual address where the data is located</param>
+        /// <param name="size">Size of the data</param>
+        /// <returns>The span of the data at the specified memory location</returns>
+        public ReadOnlySpan<byte> GetSpan(ulong gpuVa, int size)
+        {
+            ulong processVa = Translate(gpuVa);
+
+            return _context.PhysicalMemory.GetSpan(processVa, size);
+        }
+
+        /// <summary>
+        /// Gets a writable region from GPU mapped memory.
+        /// </summary>
+        /// <param name="address">Start address of the range</param>
+        /// <param name="size">Size in bytes to be range</param>
+        /// <returns>A writable region with the data at the specified memory location</returns>
+        public WritableRegion GetWritableRegion(ulong gpuVa, int size)
+        {
+            ulong processVa = Translate(gpuVa);
+
+            return _context.PhysicalMemory.GetWritableRegion(processVa, size);
+        }
+
+        /// <summary>
+        /// Writes data to GPU mapped memory.
+        /// </summary>
+        /// <param name="gpuVa">GPU virtual address to write the data into</param>
+        /// <param name="data">The data to be written</param>
+        public void Write(ulong gpuVa, ReadOnlySpan<byte> data)
+        {
+            ulong processVa = Translate(gpuVa);
+
+            _context.PhysicalMemory.Write(processVa, data);
+        }
+
+        /// <summary>
         /// Maps a given range of pages to the specified CPU virtual address.
         /// </summary>
         /// <remarks>
diff --git a/Ryujinx.Graphics.Gpu/Memory/PhysicalMemory.cs b/Ryujinx.Graphics.Gpu/Memory/PhysicalMemory.cs
index 4a80aa1a..88beab8f 100644
--- a/Ryujinx.Graphics.Gpu/Memory/PhysicalMemory.cs
+++ b/Ryujinx.Graphics.Gpu/Memory/PhysicalMemory.cs
@@ -1,3 +1,4 @@
+using Ryujinx.Cpu;
 using System;
 using System.Runtime.CompilerServices;
 using System.Runtime.InteropServices;
@@ -35,6 +36,17 @@ namespace Ryujinx.Graphics.Gpu.Memory
         }
 
         /// <summary>
+        /// Gets a writable region from the application process.
+        /// </summary>
+        /// <param name="address">Start address of the range</param>
+        /// <param name="size">Size in bytes to be range</param>
+        /// <returns>A writable region with the data at the specified memory location</returns>
+        public WritableRegion GetWritableRegion(ulong address, int size)
+        {
+            return _cpuMemory.GetWritableRegion(address, size);
+        }
+
+        /// <summary>
         /// Reads data from the application process.
         /// </summary>
         /// <typeparam name="T">Type of the structure</typeparam>