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>

1 files changed, 126 insertions, 0 deletions

diff --git a/Ryujinx.Graphics.Nvdec/Image/SurfaceWriter.cs b/Ryujinx.Graphics.Nvdec/Image/SurfaceWriter.cs
new file mode 100644
index 00000000..5c294621
--- /dev/null
+++ b/Ryujinx.Graphics.Nvdec/Image/SurfaceWriter.cs
@@ -0,0 +1,126 @@
+using Ryujinx.Common;
+using Ryujinx.Graphics.Gpu.Memory;
+using Ryujinx.Graphics.Texture;
+using Ryujinx.Graphics.Video;
+using System;
+using System.Runtime.Intrinsics;
+using System.Runtime.Intrinsics.X86;
+using static Ryujinx.Graphics.Nvdec.Image.SurfaceCommon;
+using static Ryujinx.Graphics.Nvdec.MemoryExtensions;
+
+namespace Ryujinx.Graphics.Nvdec.Image
+{
+    static class SurfaceWriter
+    {
+        public static void Write(MemoryManager gmm, ISurface surface, uint lumaOffset, uint chromaOffset)
+        {
+            int lumaSize = GetBlockLinearSize(surface.Width, surface.Height, 1);
+
+            using var luma = gmm.GetWritableRegion(ExtendOffset(lumaOffset), lumaSize);
+
+            WriteLuma(
+                luma.Memory.Span,
+                surface.YPlane.AsSpan(),
+                surface.Stride,
+                surface.Width,
+                surface.Height);
+
+            int chromaSize = GetBlockLinearSize(surface.UvWidth, surface.UvHeight, 2);
+
+            using var chroma = gmm.GetWritableRegion(ExtendOffset(chromaOffset), chromaSize);
+
+            WriteChroma(
+                chroma.Memory.Span,
+                surface.UPlane.AsSpan(),
+                surface.VPlane.AsSpan(),
+                surface.UvStride,
+                surface.UvWidth,
+                surface.UvHeight);
+        }
+
+        private static void WriteLuma(Span<byte> dst, ReadOnlySpan<byte> src, int srcStride, int width, int height)
+        {
+            LayoutConverter.ConvertLinearToBlockLinear(dst, width, height, srcStride, 1, 2, src);
+        }
+
+        private unsafe static void WriteChroma(
+            Span<byte> dst,
+            ReadOnlySpan<byte> srcU,
+            ReadOnlySpan<byte> srcV,
+            int srcStride,
+            int width,
+            int height)
+        {
+            OffsetCalculator calc = new OffsetCalculator(width, height, 0, false, 2, 2);
+
+            if (Sse2.IsSupported)
+            {
+                int strideTrunc64 = BitUtils.AlignDown(width * 2, 64);
+
+                int inStrideGap = srcStride - width;
+
+                fixed (byte* outputPtr = dst, srcUPtr = srcU, srcVPtr = srcV)
+                {
+                    byte* inUPtr = srcUPtr;
+                    byte* inVPtr = srcVPtr;
+
+                    for (int y = 0; y < height; y++)
+                    {
+                        calc.SetY(y);
+
+                        for (int x = 0; x < strideTrunc64; x += 64, inUPtr += 32, inVPtr += 32)
+                        {
+                            byte* offset = outputPtr + calc.GetOffsetWithLineOffset64(x);
+                            byte* offset2 = offset + 0x20;
+                            byte* offset3 = offset + 0x100;
+                            byte* offset4 = offset + 0x120;
+
+                            Vector128<byte> value = *(Vector128<byte>*)inUPtr;
+                            Vector128<byte> value2 = *(Vector128<byte>*)inVPtr;
+                            Vector128<byte> value3 = *(Vector128<byte>*)(inUPtr + 16);
+                            Vector128<byte> value4 = *(Vector128<byte>*)(inVPtr + 16);
+
+                            Vector128<byte> uv0 = Sse2.UnpackLow(value, value2);
+                            Vector128<byte> uv1 = Sse2.UnpackHigh(value, value2);
+                            Vector128<byte> uv2 = Sse2.UnpackLow(value3, value4);
+                            Vector128<byte> uv3 = Sse2.UnpackHigh(value3, value4);
+
+                            *(Vector128<byte>*)offset = uv0;
+                            *(Vector128<byte>*)offset2 = uv1;
+                            *(Vector128<byte>*)offset3 = uv2;
+                            *(Vector128<byte>*)offset4 = uv3;
+                        }
+
+                        for (int x = strideTrunc64 / 2; x < width; x++, inUPtr++, inVPtr++)
+                        {
+                            byte* offset = outputPtr + calc.GetOffset(x);
+
+                            *offset = *inUPtr;
+                            *(offset + 1) = *inVPtr;
+                        }
+
+                        inUPtr += inStrideGap;
+                        inVPtr += inStrideGap;
+                    }
+                }
+            }
+            else
+            {
+                for (int y = 0; y < height; y++)
+                {
+                    int srcBaseOffset = y * srcStride;
+
+                    calc.SetY(y);
+
+                    for (int x = 0; x < width; x++)
+                    {
+                        int dstOffset = calc.GetOffset(x);
+
+                        dst[dstOffset + 0] = srcU[srcBaseOffset + x];
+                        dst[dstOffset + 1] = srcV[srcBaseOffset + x];
+                    }
+                }
+            }
+        }
+    }
+}