mirror of
https://git.743378673.xyz/MeloNX/MeloNX.git
synced 2025-07-24 23:47:10 +02:00
Merge shader branch, adding support for GLSL decompilation, a macro
interpreter, and a rewrite of the GPU code.
This commit is contained in:
parent
7acd0e0122
commit
b9aa3966c0
77 changed files with 5301 additions and 766 deletions
|
@ -1,468 +0,0 @@
|
|||
using System;
|
||||
using System.Drawing;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
static class BCn
|
||||
{
|
||||
public static byte[] DecodeBC1(NsGpuTexture Tex, int Offset)
|
||||
{
|
||||
int W = (Tex.Width + 3) / 4;
|
||||
int H = (Tex.Height + 3) / 4;
|
||||
|
||||
byte[] Output = new byte[W * H * 64];
|
||||
|
||||
SwizzleAddr Swizzle = new SwizzleAddr(W, H, 8);
|
||||
|
||||
for (int Y = 0; Y < H; Y++)
|
||||
{
|
||||
for (int X = 0; X < W; X++)
|
||||
{
|
||||
int IOffs = Offset + Swizzle.GetSwizzledAddress64(X, Y) * 8;
|
||||
|
||||
byte[] Tile = BCnDecodeTile(Tex.Data, IOffs, true);
|
||||
|
||||
int TOffset = 0;
|
||||
|
||||
for (int TY = 0; TY < 4; TY++)
|
||||
{
|
||||
for (int TX = 0; TX < 4; TX++)
|
||||
{
|
||||
int OOffset = (X * 4 + TX + (Y * 4 + TY) * W * 4) * 4;
|
||||
|
||||
Output[OOffset + 0] = Tile[TOffset + 0];
|
||||
Output[OOffset + 1] = Tile[TOffset + 1];
|
||||
Output[OOffset + 2] = Tile[TOffset + 2];
|
||||
Output[OOffset + 3] = Tile[TOffset + 3];
|
||||
|
||||
TOffset += 4;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
public static byte[] DecodeBC2(NsGpuTexture Tex, int Offset)
|
||||
{
|
||||
int W = (Tex.Width + 3) / 4;
|
||||
int H = (Tex.Height + 3) / 4;
|
||||
|
||||
byte[] Output = new byte[W * H * 64];
|
||||
|
||||
SwizzleAddr Swizzle = new SwizzleAddr(W, H, 4);
|
||||
|
||||
for (int Y = 0; Y < H; Y++)
|
||||
{
|
||||
for (int X = 0; X < W; X++)
|
||||
{
|
||||
int IOffs = Offset + Swizzle.GetSwizzledAddress128(X, Y) * 16;
|
||||
|
||||
byte[] Tile = BCnDecodeTile(Tex.Data, IOffs + 8, false);
|
||||
|
||||
int AlphaLow = Get32(Tex.Data, IOffs + 0);
|
||||
int AlphaHigh = Get32(Tex.Data, IOffs + 4);
|
||||
|
||||
ulong AlphaCh = (uint)AlphaLow | (ulong)AlphaHigh << 32;
|
||||
|
||||
int TOffset = 0;
|
||||
|
||||
for (int TY = 0; TY < 4; TY++)
|
||||
{
|
||||
for (int TX = 0; TX < 4; TX++)
|
||||
{
|
||||
ulong Alpha = (AlphaCh >> (TY * 16 + TX * 4)) & 0xf;
|
||||
|
||||
int OOffset = (X * 4 + TX + (Y * 4 + TY) * W * 4) * 4;
|
||||
|
||||
Output[OOffset + 0] = Tile[TOffset + 0];
|
||||
Output[OOffset + 1] = Tile[TOffset + 1];
|
||||
Output[OOffset + 2] = Tile[TOffset + 2];
|
||||
Output[OOffset + 3] = (byte)(Alpha | (Alpha << 4));
|
||||
|
||||
TOffset += 4;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
public static byte[] DecodeBC3(NsGpuTexture Tex, int Offset)
|
||||
{
|
||||
int W = (Tex.Width + 3) / 4;
|
||||
int H = (Tex.Height + 3) / 4;
|
||||
|
||||
byte[] Output = new byte[W * H * 64];
|
||||
|
||||
SwizzleAddr Swizzle = new SwizzleAddr(W, H, 4);
|
||||
|
||||
for (int Y = 0; Y < H; Y++)
|
||||
{
|
||||
for (int X = 0; X < W; X++)
|
||||
{
|
||||
int IOffs = Offset + Swizzle.GetSwizzledAddress128(X, Y) * 16;
|
||||
|
||||
byte[] Tile = BCnDecodeTile(Tex.Data, IOffs + 8, false);
|
||||
|
||||
byte[] Alpha = new byte[8];
|
||||
|
||||
Alpha[0] = Tex.Data[IOffs + 0];
|
||||
Alpha[1] = Tex.Data[IOffs + 1];
|
||||
|
||||
CalculateBC3Alpha(Alpha);
|
||||
|
||||
int AlphaLow = Get32(Tex.Data, IOffs + 2);
|
||||
int AlphaHigh = Get16(Tex.Data, IOffs + 6);
|
||||
|
||||
ulong AlphaCh = (uint)AlphaLow | (ulong)AlphaHigh << 32;
|
||||
|
||||
int TOffset = 0;
|
||||
|
||||
for (int TY = 0; TY < 4; TY++)
|
||||
{
|
||||
for (int TX = 0; TX < 4; TX++)
|
||||
{
|
||||
int OOffset = (X * 4 + TX + (Y * 4 + TY) * W * 4) * 4;
|
||||
|
||||
byte AlphaPx = Alpha[(AlphaCh >> (TY * 12 + TX * 3)) & 7];
|
||||
|
||||
Output[OOffset + 0] = Tile[TOffset + 0];
|
||||
Output[OOffset + 1] = Tile[TOffset + 1];
|
||||
Output[OOffset + 2] = Tile[TOffset + 2];
|
||||
Output[OOffset + 3] = AlphaPx;
|
||||
|
||||
TOffset += 4;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
public static byte[] DecodeBC4(NsGpuTexture Tex, int Offset)
|
||||
{
|
||||
int W = (Tex.Width + 3) / 4;
|
||||
int H = (Tex.Height + 3) / 4;
|
||||
|
||||
byte[] Output = new byte[W * H * 64];
|
||||
|
||||
SwizzleAddr Swizzle = new SwizzleAddr(W, H, 8);
|
||||
|
||||
for (int Y = 0; Y < H; Y++)
|
||||
{
|
||||
for (int X = 0; X < W; X++)
|
||||
{
|
||||
int IOffs = Swizzle.GetSwizzledAddress64(X, Y) * 8;
|
||||
|
||||
byte[] Red = new byte[8];
|
||||
|
||||
Red[0] = Tex.Data[IOffs + 0];
|
||||
Red[1] = Tex.Data[IOffs + 1];
|
||||
|
||||
CalculateBC3Alpha(Red);
|
||||
|
||||
int RedLow = Get32(Tex.Data, IOffs + 2);
|
||||
int RedHigh = Get16(Tex.Data, IOffs + 6);
|
||||
|
||||
ulong RedCh = (uint)RedLow | (ulong)RedHigh << 32;
|
||||
|
||||
int TOffset = 0;
|
||||
|
||||
for (int TY = 0; TY < 4; TY++)
|
||||
{
|
||||
for (int TX = 0; TX < 4; TX++)
|
||||
{
|
||||
int OOffset = (X * 4 + TX + (Y * 4 + TY) * W * 4) * 4;
|
||||
|
||||
byte RedPx = Red[(RedCh >> (TY * 12 + TX * 3)) & 7];
|
||||
|
||||
Output[OOffset + 0] = RedPx;
|
||||
Output[OOffset + 1] = RedPx;
|
||||
Output[OOffset + 2] = RedPx;
|
||||
Output[OOffset + 3] = 0xff;
|
||||
|
||||
TOffset += 4;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
public static byte[] DecodeBC5(NsGpuTexture Tex, int Offset, bool SNorm)
|
||||
{
|
||||
int W = (Tex.Width + 3) / 4;
|
||||
int H = (Tex.Height + 3) / 4;
|
||||
|
||||
byte[] Output = new byte[W * H * 64];
|
||||
|
||||
SwizzleAddr Swizzle = new SwizzleAddr(W, H, 4);
|
||||
|
||||
for (int Y = 0; Y < H; Y++)
|
||||
{
|
||||
for (int X = 0; X < W; X++)
|
||||
{
|
||||
int IOffs = Swizzle.GetSwizzledAddress128(X, Y) * 16;
|
||||
|
||||
byte[] Red = new byte[8];
|
||||
byte[] Green = new byte[8];
|
||||
|
||||
Red[0] = Tex.Data[IOffs + 0];
|
||||
Red[1] = Tex.Data[IOffs + 1];
|
||||
|
||||
Green[0] = Tex.Data[IOffs + 8];
|
||||
Green[1] = Tex.Data[IOffs + 9];
|
||||
|
||||
if (SNorm)
|
||||
{
|
||||
CalculateBC3AlphaS(Red);
|
||||
CalculateBC3AlphaS(Green);
|
||||
}
|
||||
else
|
||||
{
|
||||
CalculateBC3Alpha(Red);
|
||||
CalculateBC3Alpha(Green);
|
||||
}
|
||||
|
||||
int RedLow = Get32(Tex.Data, IOffs + 2);
|
||||
int RedHigh = Get16(Tex.Data, IOffs + 6);
|
||||
|
||||
int GreenLow = Get32(Tex.Data, IOffs + 10);
|
||||
int GreenHigh = Get16(Tex.Data, IOffs + 14);
|
||||
|
||||
ulong RedCh = (uint)RedLow | (ulong)RedHigh << 32;
|
||||
ulong GreenCh = (uint)GreenLow | (ulong)GreenHigh << 32;
|
||||
|
||||
int TOffset = 0;
|
||||
|
||||
if (SNorm)
|
||||
{
|
||||
for (int TY = 0; TY < 4; TY++)
|
||||
{
|
||||
for (int TX = 0; TX < 4; TX++)
|
||||
{
|
||||
int Shift = TY * 12 + TX * 3;
|
||||
|
||||
int OOffset = (X * 4 + TX + (Y * 4 + TY) * W * 4) * 4;
|
||||
|
||||
byte RedPx = Red [(RedCh >> Shift) & 7];
|
||||
byte GreenPx = Green[(GreenCh >> Shift) & 7];
|
||||
|
||||
RedPx += 0x80;
|
||||
GreenPx += 0x80;
|
||||
|
||||
float NX = (RedPx / 255f) * 2 - 1;
|
||||
float NY = (GreenPx / 255f) * 2 - 1;
|
||||
|
||||
float NZ = (float)Math.Sqrt(1 - (NX * NX + NY * NY));
|
||||
|
||||
Output[OOffset + 0] = Clamp((NZ + 1) * 0.5f);
|
||||
Output[OOffset + 1] = Clamp((NY + 1) * 0.5f);
|
||||
Output[OOffset + 2] = Clamp((NX + 1) * 0.5f);
|
||||
Output[OOffset + 3] = 0xff;
|
||||
|
||||
TOffset += 4;
|
||||
}
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
for (int TY = 0; TY < 4; TY++)
|
||||
{
|
||||
for (int TX = 0; TX < 4; TX++)
|
||||
{
|
||||
int Shift = TY * 12 + TX * 3;
|
||||
|
||||
int OOffset = (X * 4 + TX + (Y * 4 + TY) * W * 4) * 4;
|
||||
|
||||
byte RedPx = Red [(RedCh >> Shift) & 7];
|
||||
byte GreenPx = Green[(GreenCh >> Shift) & 7];
|
||||
|
||||
Output[OOffset + 0] = RedPx;
|
||||
Output[OOffset + 1] = RedPx;
|
||||
Output[OOffset + 2] = RedPx;
|
||||
Output[OOffset + 3] = GreenPx;
|
||||
|
||||
TOffset += 4;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
private static byte Clamp(float Value)
|
||||
{
|
||||
if (Value > 1)
|
||||
{
|
||||
return 0xff;
|
||||
}
|
||||
else if (Value < 0)
|
||||
{
|
||||
return 0;
|
||||
}
|
||||
else
|
||||
{
|
||||
return (byte)(Value * 0xff);
|
||||
}
|
||||
}
|
||||
|
||||
private static void CalculateBC3Alpha(byte[] Alpha)
|
||||
{
|
||||
for (int i = 2; i < 8; i++)
|
||||
{
|
||||
if (Alpha[0] > Alpha[1])
|
||||
{
|
||||
Alpha[i] = (byte)(((8 - i) * Alpha[0] + (i - 1) * Alpha[1]) / 7);
|
||||
}
|
||||
else if (i < 6)
|
||||
{
|
||||
Alpha[i] = (byte)(((6 - i) * Alpha[0] + (i - 1) * Alpha[1]) / 7);
|
||||
}
|
||||
else if (i == 6)
|
||||
{
|
||||
Alpha[i] = 0;
|
||||
}
|
||||
else /* i == 7 */
|
||||
{
|
||||
Alpha[i] = 0xff;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private static void CalculateBC3AlphaS(byte[] Alpha)
|
||||
{
|
||||
for (int i = 2; i < 8; i++)
|
||||
{
|
||||
if ((sbyte)Alpha[0] > (sbyte)Alpha[1])
|
||||
{
|
||||
Alpha[i] = (byte)(((8 - i) * (sbyte)Alpha[0] + (i - 1) * (sbyte)Alpha[1]) / 7);
|
||||
}
|
||||
else if (i < 6)
|
||||
{
|
||||
Alpha[i] = (byte)(((6 - i) * (sbyte)Alpha[0] + (i - 1) * (sbyte)Alpha[1]) / 7);
|
||||
}
|
||||
else if (i == 6)
|
||||
{
|
||||
Alpha[i] = 0x80;
|
||||
}
|
||||
else /* i == 7 */
|
||||
{
|
||||
Alpha[i] = 0x7f;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private static byte[] BCnDecodeTile(
|
||||
byte[] Input,
|
||||
int Offset,
|
||||
bool IsBC1)
|
||||
{
|
||||
Color[] CLUT = new Color[4];
|
||||
|
||||
int c0 = Get16(Input, Offset + 0);
|
||||
int c1 = Get16(Input, Offset + 2);
|
||||
|
||||
CLUT[0] = DecodeRGB565(c0);
|
||||
CLUT[1] = DecodeRGB565(c1);
|
||||
CLUT[2] = CalculateCLUT2(CLUT[0], CLUT[1], c0, c1, IsBC1);
|
||||
CLUT[3] = CalculateCLUT3(CLUT[0], CLUT[1], c0, c1, IsBC1);
|
||||
|
||||
int Indices = Get32(Input, Offset + 4);
|
||||
|
||||
int IdxShift = 0;
|
||||
|
||||
byte[] Output = new byte[4 * 4 * 4];
|
||||
|
||||
int OOffset = 0;
|
||||
|
||||
for (int TY = 0; TY < 4; TY++)
|
||||
{
|
||||
for (int TX = 0; TX < 4; TX++)
|
||||
{
|
||||
int Idx = (Indices >> IdxShift) & 3;
|
||||
|
||||
IdxShift += 2;
|
||||
|
||||
Color Pixel = CLUT[Idx];
|
||||
|
||||
Output[OOffset + 0] = Pixel.R;
|
||||
Output[OOffset + 1] = Pixel.G;
|
||||
Output[OOffset + 2] = Pixel.B;
|
||||
Output[OOffset + 3] = Pixel.A;
|
||||
|
||||
OOffset += 4;
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
private static Color CalculateCLUT2(Color C0, Color C1, int c0, int c1, bool IsBC1)
|
||||
{
|
||||
if (c0 > c1 || !IsBC1)
|
||||
{
|
||||
return Color.FromArgb(
|
||||
(2 * C0.R + C1.R) / 3,
|
||||
(2 * C0.G + C1.G) / 3,
|
||||
(2 * C0.B + C1.B) / 3);
|
||||
}
|
||||
else
|
||||
{
|
||||
return Color.FromArgb(
|
||||
(C0.R + C1.R) / 2,
|
||||
(C0.G + C1.G) / 2,
|
||||
(C0.B + C1.B) / 2);
|
||||
}
|
||||
}
|
||||
|
||||
private static Color CalculateCLUT3(Color C0, Color C1, int c0, int c1, bool IsBC1)
|
||||
{
|
||||
if (c0 > c1 || !IsBC1)
|
||||
{
|
||||
return
|
||||
Color.FromArgb(
|
||||
(2 * C1.R + C0.R) / 3,
|
||||
(2 * C1.G + C0.G) / 3,
|
||||
(2 * C1.B + C0.B) / 3);
|
||||
}
|
||||
|
||||
return Color.Transparent;
|
||||
}
|
||||
|
||||
private static Color DecodeRGB565(int Value)
|
||||
{
|
||||
int B = ((Value >> 0) & 0x1f) << 3;
|
||||
int G = ((Value >> 5) & 0x3f) << 2;
|
||||
int R = ((Value >> 11) & 0x1f) << 3;
|
||||
|
||||
return Color.FromArgb(
|
||||
R | (R >> 5),
|
||||
G | (G >> 6),
|
||||
B | (B >> 5));
|
||||
}
|
||||
|
||||
private static int Get16(byte[] Data, int Address)
|
||||
{
|
||||
return
|
||||
Data[Address + 0] << 0 |
|
||||
Data[Address + 1] << 8;
|
||||
}
|
||||
|
||||
private static int Get32(byte[] Data, int Address)
|
||||
{
|
||||
return
|
||||
Data[Address + 0] << 0 |
|
||||
Data[Address + 1] << 8 |
|
||||
Data[Address + 2] << 16 |
|
||||
Data[Address + 3] << 24;
|
||||
}
|
||||
}
|
||||
}
|
57
Ryujinx.Graphics/Gpu/BlockLinearSwizzle.cs
Normal file
57
Ryujinx.Graphics/Gpu/BlockLinearSwizzle.cs
Normal file
|
@ -0,0 +1,57 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
class BlockLinearSwizzle : ISwizzle
|
||||
{
|
||||
private int BhShift;
|
||||
private int BppShift;
|
||||
private int BhMask;
|
||||
|
||||
private int XShift;
|
||||
private int GobStride;
|
||||
|
||||
public BlockLinearSwizzle(int Width, int Bpp, int BlockHeight = 16)
|
||||
{
|
||||
BhMask = (BlockHeight * 8) - 1;
|
||||
|
||||
BhShift = CountLsbZeros(BlockHeight * 8);
|
||||
BppShift = CountLsbZeros(Bpp);
|
||||
|
||||
int WidthInGobs = Width * Bpp / 64;
|
||||
|
||||
GobStride = 512 * BlockHeight * WidthInGobs;
|
||||
|
||||
XShift = CountLsbZeros(512 * BlockHeight);
|
||||
}
|
||||
|
||||
private int CountLsbZeros(int Value)
|
||||
{
|
||||
int Count = 0;
|
||||
|
||||
while (((Value >> Count) & 1) == 0)
|
||||
{
|
||||
Count++;
|
||||
}
|
||||
|
||||
return Count;
|
||||
}
|
||||
|
||||
public int GetSwizzleOffset(int X, int Y)
|
||||
{
|
||||
X <<= BppShift;
|
||||
|
||||
int Position = (Y >> BhShift) * GobStride;
|
||||
|
||||
Position += (X >> 6) << XShift;
|
||||
|
||||
Position += ((Y & BhMask) >> 3) << 9;
|
||||
|
||||
Position += ((X & 0x3f) >> 5) << 8;
|
||||
Position += ((Y & 0x07) >> 1) << 6;
|
||||
Position += ((X & 0x1f) >> 4) << 5;
|
||||
Position += ((Y & 0x01) >> 0) << 4;
|
||||
Position += ((X & 0x0f) >> 0) << 0;
|
||||
|
||||
return Position;
|
||||
}
|
||||
}
|
||||
}
|
11
Ryujinx.Graphics/Gpu/INvGpuEngine.cs
Normal file
11
Ryujinx.Graphics/Gpu/INvGpuEngine.cs
Normal file
|
@ -0,0 +1,11 @@
|
|||
using ChocolArm64.Memory;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
interface INvGpuEngine
|
||||
{
|
||||
int[] Registers { get; }
|
||||
|
||||
void CallMethod(AMemory Memory, NsGpuPBEntry PBEntry);
|
||||
}
|
||||
}
|
7
Ryujinx.Graphics/Gpu/ISwizzle.cs
Normal file
7
Ryujinx.Graphics/Gpu/ISwizzle.cs
Normal file
|
@ -0,0 +1,7 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
interface ISwizzle
|
||||
{
|
||||
int GetSwizzleOffset(int X, int Y);
|
||||
}
|
||||
}
|
20
Ryujinx.Graphics/Gpu/LinearSwizzle.cs
Normal file
20
Ryujinx.Graphics/Gpu/LinearSwizzle.cs
Normal file
|
@ -0,0 +1,20 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
class LinearSwizzle : ISwizzle
|
||||
{
|
||||
private int Bpp;
|
||||
private int Stride;
|
||||
|
||||
public LinearSwizzle(int Width, int Bpp)
|
||||
{
|
||||
this.Bpp = Bpp;
|
||||
|
||||
Stride = Width * Bpp;
|
||||
}
|
||||
|
||||
public int GetSwizzleOffset(int X, int Y)
|
||||
{
|
||||
return X * Bpp + Y * Stride;
|
||||
}
|
||||
}
|
||||
}
|
420
Ryujinx.Graphics/Gpu/MacroInterpreter.cs
Normal file
420
Ryujinx.Graphics/Gpu/MacroInterpreter.cs
Normal file
|
@ -0,0 +1,420 @@
|
|||
using ChocolArm64.Memory;
|
||||
using System;
|
||||
using System.Collections.Generic;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
class MacroInterpreter
|
||||
{
|
||||
private enum AssignmentOperation
|
||||
{
|
||||
IgnoreAndFetch = 0,
|
||||
Move = 1,
|
||||
MoveAndSetMaddr = 2,
|
||||
FetchAndSend = 3,
|
||||
MoveAndSend = 4,
|
||||
FetchAndSetMaddr = 5,
|
||||
MoveAndSetMaddrThenFetchAndSend = 6,
|
||||
MoveAndSetMaddrThenSendHigh = 7
|
||||
}
|
||||
|
||||
private enum AluOperation
|
||||
{
|
||||
AluReg = 0,
|
||||
AddImmediate = 1,
|
||||
BitfieldReplace = 2,
|
||||
BitfieldExtractLslImm = 3,
|
||||
BitfieldExtractLslReg = 4,
|
||||
ReadImmediate = 5
|
||||
}
|
||||
|
||||
private enum AluRegOperation
|
||||
{
|
||||
Add = 0,
|
||||
AddWithCarry = 1,
|
||||
Subtract = 2,
|
||||
SubtractWithBorrow = 3,
|
||||
BitwiseExclusiveOr = 8,
|
||||
BitwiseOr = 9,
|
||||
BitwiseAnd = 10,
|
||||
BitwiseAndNot = 11,
|
||||
BitwiseNotAnd = 12
|
||||
}
|
||||
|
||||
private NvGpuFifo PFifo;
|
||||
private INvGpuEngine Engine;
|
||||
|
||||
public Queue<int> Fifo { get; private set; }
|
||||
|
||||
private int[] Gprs;
|
||||
|
||||
private int MethAddr;
|
||||
private int MethIncr;
|
||||
|
||||
private bool Carry;
|
||||
|
||||
private int OpCode;
|
||||
|
||||
private int PipeOp;
|
||||
|
||||
private long Pc;
|
||||
|
||||
public MacroInterpreter(NvGpuFifo PFifo, INvGpuEngine Engine)
|
||||
{
|
||||
this.PFifo = PFifo;
|
||||
this.Engine = Engine;
|
||||
|
||||
Fifo = new Queue<int>();
|
||||
|
||||
Gprs = new int[8];
|
||||
}
|
||||
|
||||
public void Execute(AMemory Memory, long Position, int Param)
|
||||
{
|
||||
Reset();
|
||||
|
||||
Gprs[1] = Param;
|
||||
|
||||
Pc = Position;
|
||||
|
||||
FetchOpCode(Memory);
|
||||
|
||||
while (Step(Memory));
|
||||
|
||||
//Due to the delay slot, we still need to execute
|
||||
//one more instruction before we actually exit.
|
||||
Step(Memory);
|
||||
}
|
||||
|
||||
private void Reset()
|
||||
{
|
||||
for (int Index = 0; Index < Gprs.Length; Index++)
|
||||
{
|
||||
Gprs[Index] = 0;
|
||||
}
|
||||
|
||||
MethAddr = 0;
|
||||
MethIncr = 0;
|
||||
|
||||
Carry = false;
|
||||
}
|
||||
|
||||
private bool Step(AMemory Memory)
|
||||
{
|
||||
long BaseAddr = Pc - 4;
|
||||
|
||||
FetchOpCode(Memory);
|
||||
|
||||
if ((OpCode & 7) < 7)
|
||||
{
|
||||
//Operation produces a value.
|
||||
AssignmentOperation AsgOp = (AssignmentOperation)((OpCode >> 4) & 7);
|
||||
|
||||
int Result = GetAluResult();
|
||||
|
||||
switch (AsgOp)
|
||||
{
|
||||
//Fetch parameter and ignore result.
|
||||
case AssignmentOperation.IgnoreAndFetch:
|
||||
{
|
||||
SetDstGpr(FetchParam());
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//Move result.
|
||||
case AssignmentOperation.Move:
|
||||
{
|
||||
SetDstGpr(Result);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//Move result and use as Method Address.
|
||||
case AssignmentOperation.MoveAndSetMaddr:
|
||||
{
|
||||
SetDstGpr(Result);
|
||||
|
||||
SetMethAddr(Result);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//Fetch parameter and send result.
|
||||
case AssignmentOperation.FetchAndSend:
|
||||
{
|
||||
SetDstGpr(FetchParam());
|
||||
|
||||
Send(Memory, Result);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//Move and send result.
|
||||
case AssignmentOperation.MoveAndSend:
|
||||
{
|
||||
SetDstGpr(Result);
|
||||
|
||||
Send(Memory, Result);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//Fetch parameter and use result as Method Address.
|
||||
case AssignmentOperation.FetchAndSetMaddr:
|
||||
{
|
||||
SetDstGpr(FetchParam());
|
||||
|
||||
SetMethAddr(Result);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//Move result and use as Method Address, then fetch and send paramter.
|
||||
case AssignmentOperation.MoveAndSetMaddrThenFetchAndSend:
|
||||
{
|
||||
SetDstGpr(Result);
|
||||
|
||||
SetMethAddr(Result);
|
||||
|
||||
Send(Memory, FetchParam());
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//Move result and use as Method Address, then send bits 17:12 of result.
|
||||
case AssignmentOperation.MoveAndSetMaddrThenSendHigh:
|
||||
{
|
||||
SetDstGpr(Result);
|
||||
|
||||
SetMethAddr(Result);
|
||||
|
||||
Send(Memory, (Result >> 12) & 0x3f);
|
||||
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
//Branch.
|
||||
bool OnNotZero = ((OpCode >> 4) & 1) != 0;
|
||||
|
||||
bool Taken = OnNotZero
|
||||
? GetGprA() != 0
|
||||
: GetGprA() == 0;
|
||||
|
||||
if (Taken)
|
||||
{
|
||||
Pc = BaseAddr + (GetImm() << 2);
|
||||
|
||||
bool NoDelays = (OpCode & 0x20) != 0;
|
||||
|
||||
if (NoDelays)
|
||||
{
|
||||
FetchOpCode(Memory);
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
}
|
||||
|
||||
bool Exit = (OpCode & 0x80) != 0;
|
||||
|
||||
return !Exit;
|
||||
}
|
||||
|
||||
private void FetchOpCode(AMemory Memory)
|
||||
{
|
||||
OpCode = PipeOp;
|
||||
|
||||
PipeOp = Memory.ReadInt32(Pc);
|
||||
|
||||
Pc += 4;
|
||||
}
|
||||
|
||||
private int GetAluResult()
|
||||
{
|
||||
AluOperation Op = (AluOperation)(OpCode & 7);
|
||||
|
||||
switch (Op)
|
||||
{
|
||||
case AluOperation.AluReg:
|
||||
{
|
||||
AluRegOperation AluOp = (AluRegOperation)((OpCode >> 17) & 0x1f);
|
||||
|
||||
return GetAluResult(AluOp, GetGprA(), GetGprB());
|
||||
}
|
||||
|
||||
case AluOperation.AddImmediate:
|
||||
{
|
||||
return GetGprA() + GetImm();
|
||||
}
|
||||
|
||||
case AluOperation.BitfieldReplace:
|
||||
case AluOperation.BitfieldExtractLslImm:
|
||||
case AluOperation.BitfieldExtractLslReg:
|
||||
{
|
||||
int BfSrcBit = (OpCode >> 17) & 0x1f;
|
||||
int BfSize = (OpCode >> 22) & 0x1f;
|
||||
int BfDstBit = (OpCode >> 27) & 0x1f;
|
||||
|
||||
int BfMask = (1 << BfSize) - 1;
|
||||
|
||||
int Dst = GetGprA();
|
||||
int Src = GetGprB();
|
||||
|
||||
switch (Op)
|
||||
{
|
||||
case AluOperation.BitfieldReplace:
|
||||
{
|
||||
Src = (int)((uint)Src >> BfSrcBit) & BfMask;
|
||||
|
||||
Dst &= ~(BfMask << BfDstBit);
|
||||
|
||||
Dst |= Src << BfDstBit;
|
||||
|
||||
return Dst;
|
||||
}
|
||||
|
||||
case AluOperation.BitfieldExtractLslImm:
|
||||
{
|
||||
Src = (int)((uint)Src >> Dst) & BfMask;
|
||||
|
||||
return Src << BfDstBit;
|
||||
}
|
||||
|
||||
case AluOperation.BitfieldExtractLslReg:
|
||||
{
|
||||
Src = (int)((uint)Src >> BfSrcBit) & BfMask;
|
||||
|
||||
return Src << Dst;
|
||||
}
|
||||
}
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case AluOperation.ReadImmediate:
|
||||
{
|
||||
return Read(GetGprA() + GetImm());
|
||||
}
|
||||
}
|
||||
|
||||
throw new ArgumentException(nameof(OpCode));
|
||||
}
|
||||
|
||||
private int GetAluResult(AluRegOperation AluOp, int A, int B)
|
||||
{
|
||||
switch (AluOp)
|
||||
{
|
||||
case AluRegOperation.Add:
|
||||
{
|
||||
ulong Result = (ulong)A + (ulong)B;
|
||||
|
||||
Carry = Result > 0xffffffff;
|
||||
|
||||
return (int)Result;
|
||||
}
|
||||
|
||||
case AluRegOperation.AddWithCarry:
|
||||
{
|
||||
ulong Result = (ulong)A + (ulong)B + (Carry ? 1UL : 0UL);
|
||||
|
||||
Carry = Result > 0xffffffff;
|
||||
|
||||
return (int)Result;
|
||||
}
|
||||
|
||||
case AluRegOperation.Subtract:
|
||||
{
|
||||
ulong Result = (ulong)A - (ulong)B;
|
||||
|
||||
Carry = Result < 0x100000000;
|
||||
|
||||
return (int)Result;
|
||||
}
|
||||
|
||||
case AluRegOperation.SubtractWithBorrow:
|
||||
{
|
||||
ulong Result = (ulong)A - (ulong)B - (Carry ? 0UL : 1UL);
|
||||
|
||||
Carry = Result < 0x100000000;
|
||||
|
||||
return (int)Result;
|
||||
}
|
||||
|
||||
case AluRegOperation.BitwiseExclusiveOr: return A ^ B;
|
||||
case AluRegOperation.BitwiseOr: return A | B;
|
||||
case AluRegOperation.BitwiseAnd: return A & B;
|
||||
case AluRegOperation.BitwiseAndNot: return A & ~B;
|
||||
case AluRegOperation.BitwiseNotAnd: return ~(A & B);
|
||||
}
|
||||
|
||||
throw new ArgumentOutOfRangeException(nameof(AluOp));
|
||||
}
|
||||
|
||||
private int GetImm()
|
||||
{
|
||||
//Note: The immediate is signed, the sign-extension is intended here.
|
||||
return OpCode >> 14;
|
||||
}
|
||||
|
||||
private void SetMethAddr(int Value)
|
||||
{
|
||||
MethAddr = (Value >> 0) & 0xfff;
|
||||
MethIncr = (Value >> 12) & 0x3f;
|
||||
}
|
||||
|
||||
private void SetDstGpr(int Value)
|
||||
{
|
||||
Gprs[(OpCode >> 8) & 7] = Value;
|
||||
}
|
||||
|
||||
private int GetGprA()
|
||||
{
|
||||
return GetGprValue((OpCode >> 11) & 7);
|
||||
}
|
||||
|
||||
private int GetGprB()
|
||||
{
|
||||
return GetGprValue((OpCode >> 14) & 7);
|
||||
}
|
||||
|
||||
private int GetGprValue(int Index)
|
||||
{
|
||||
return Index != 0 ? Gprs[Index] : 0;
|
||||
}
|
||||
|
||||
private int FetchParam()
|
||||
{
|
||||
int Value;
|
||||
|
||||
//If we don't have any parameters in the FIFO,
|
||||
//keep running the PFIFO engine until it writes the parameters.
|
||||
while (!Fifo.TryDequeue(out Value))
|
||||
{
|
||||
if (!PFifo.Step())
|
||||
{
|
||||
return 0;
|
||||
}
|
||||
}
|
||||
|
||||
return Value;
|
||||
}
|
||||
|
||||
private int Read(int Reg)
|
||||
{
|
||||
return Engine.Registers[Reg];
|
||||
}
|
||||
|
||||
private void Send(AMemory Memory, int Value)
|
||||
{
|
||||
NsGpuPBEntry PBEntry = new NsGpuPBEntry(MethAddr, 0, Value);
|
||||
|
||||
Engine.CallMethod(Memory, PBEntry);
|
||||
|
||||
MethAddr += MethIncr;
|
||||
}
|
||||
}
|
||||
}
|
|
@ -1,5 +1,5 @@
|
|||
using ChocolArm64.Memory;
|
||||
using Ryujinx.Graphics.Gal;
|
||||
using System.Threading;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
|
@ -9,7 +9,13 @@ namespace Ryujinx.Graphics.Gpu
|
|||
|
||||
internal NsGpuMemoryMgr MemoryMgr { get; private set; }
|
||||
|
||||
internal NsGpuPGraph PGraph { get; private set; }
|
||||
public NvGpuFifo Fifo;
|
||||
|
||||
internal NvGpuEngine3d Engine3d;
|
||||
|
||||
private Thread FifoProcessing;
|
||||
|
||||
private bool KeepRunning;
|
||||
|
||||
public NsGpu(IGalRenderer Renderer)
|
||||
{
|
||||
|
@ -17,7 +23,15 @@ namespace Ryujinx.Graphics.Gpu
|
|||
|
||||
MemoryMgr = new NsGpuMemoryMgr();
|
||||
|
||||
PGraph = new NsGpuPGraph(this);
|
||||
Fifo = new NvGpuFifo(this);
|
||||
|
||||
Engine3d = new NvGpuEngine3d(this);
|
||||
|
||||
KeepRunning = true;
|
||||
|
||||
FifoProcessing = new Thread(ProcessFifo);
|
||||
|
||||
FifoProcessing.Start();
|
||||
}
|
||||
|
||||
public long GetCpuAddr(long Position)
|
||||
|
@ -35,11 +49,6 @@ namespace Ryujinx.Graphics.Gpu
|
|||
return MemoryMgr.Map(CpuAddr, GpuAddr, Size);
|
||||
}
|
||||
|
||||
public void ProcessPushBuffer(NsGpuPBEntry[] PushBuffer, AMemory Memory)
|
||||
{
|
||||
PGraph.ProcessPushBuffer(PushBuffer, Memory);
|
||||
}
|
||||
|
||||
public long ReserveMemory(long Size, long Align)
|
||||
{
|
||||
return MemoryMgr.Reserve(Size, Align);
|
||||
|
@ -49,5 +58,15 @@ namespace Ryujinx.Graphics.Gpu
|
|||
{
|
||||
return MemoryMgr.Reserve(GpuAddr, Size, Align);
|
||||
}
|
||||
|
||||
private void ProcessFifo()
|
||||
{
|
||||
while (KeepRunning)
|
||||
{
|
||||
Fifo.DispatchCalls();
|
||||
|
||||
Thread.Yield();
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
|
@ -1,13 +1,11 @@
|
|||
using System;
|
||||
using System.Collections.Generic;
|
||||
using System.Collections.ObjectModel;
|
||||
using System.IO;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
public struct NsGpuPBEntry
|
||||
{
|
||||
public NsGpuRegister Register { get; private set; }
|
||||
public int Method { get; private set; }
|
||||
|
||||
public int SubChannel { get; private set; }
|
||||
|
||||
|
@ -15,65 +13,11 @@ namespace Ryujinx.Graphics.Gpu
|
|||
|
||||
public ReadOnlyCollection<int> Arguments => Array.AsReadOnly(m_Arguments);
|
||||
|
||||
public NsGpuPBEntry(NsGpuRegister Register, int SubChannel, params int[] Arguments)
|
||||
public NsGpuPBEntry(int Method, int SubChannel, params int[] Arguments)
|
||||
{
|
||||
this.Register = Register;
|
||||
this.Method = Method;
|
||||
this.SubChannel = SubChannel;
|
||||
this.m_Arguments = Arguments;
|
||||
}
|
||||
|
||||
public static NsGpuPBEntry[] DecodePushBuffer(byte[] Data)
|
||||
{
|
||||
using (MemoryStream MS = new MemoryStream(Data))
|
||||
{
|
||||
BinaryReader Reader = new BinaryReader(MS);
|
||||
|
||||
List<NsGpuPBEntry> GpFifos = new List<NsGpuPBEntry>();
|
||||
|
||||
bool CanRead() => MS.Position + 4 <= MS.Length;
|
||||
|
||||
while (CanRead())
|
||||
{
|
||||
int Packed = Reader.ReadInt32();
|
||||
|
||||
int Reg = (Packed << 2) & 0x7ffc;
|
||||
int SubC = (Packed >> 13) & 7;
|
||||
int Args = (Packed >> 16) & 0x1fff;
|
||||
int Mode = (Packed >> 29) & 7;
|
||||
|
||||
if (Mode == 4)
|
||||
{
|
||||
//Inline Mode.
|
||||
GpFifos.Add(new NsGpuPBEntry((NsGpuRegister)Reg, SubC, Args));
|
||||
}
|
||||
else
|
||||
{
|
||||
//Word mode.
|
||||
if (Mode == 1)
|
||||
{
|
||||
//Sequential Mode.
|
||||
for (int Index = 0; Index < Args && CanRead(); Index++, Reg += 4)
|
||||
{
|
||||
GpFifos.Add(new NsGpuPBEntry((NsGpuRegister)Reg, SubC, Reader.ReadInt32()));
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
//Non-Sequential Mode.
|
||||
int[] Arguments = new int[Args];
|
||||
|
||||
for (int Index = 0; Index < Args && CanRead(); Index++)
|
||||
{
|
||||
Arguments[Index] = Reader.ReadInt32();
|
||||
}
|
||||
|
||||
GpFifos.Add(new NsGpuPBEntry((NsGpuRegister)Reg, SubC, Arguments));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return GpFifos.ToArray();
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
|
@ -1,305 +0,0 @@
|
|||
using ChocolArm64.Memory;
|
||||
using Ryujinx.Graphics.Gal;
|
||||
using System.Collections.Generic;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
class NsGpuPGraph
|
||||
{
|
||||
private NsGpu Gpu;
|
||||
|
||||
private uint[] Registers;
|
||||
|
||||
public NsGpuEngine[] SubChannels;
|
||||
|
||||
private Dictionary<long, int> CurrentVertexBuffers;
|
||||
|
||||
public NsGpuPGraph(NsGpu Gpu)
|
||||
{
|
||||
this.Gpu = Gpu;
|
||||
|
||||
Registers = new uint[0x1000];
|
||||
|
||||
SubChannels = new NsGpuEngine[8];
|
||||
|
||||
CurrentVertexBuffers = new Dictionary<long, int>();
|
||||
}
|
||||
|
||||
public void ProcessPushBuffer(NsGpuPBEntry[] PushBuffer, AMemory Memory)
|
||||
{
|
||||
bool HasQuery = false;
|
||||
|
||||
foreach (NsGpuPBEntry Entry in PushBuffer)
|
||||
{
|
||||
if (Entry.Arguments.Count == 1)
|
||||
{
|
||||
SetRegister(Entry.Register, (uint)Entry.Arguments[0]);
|
||||
}
|
||||
|
||||
switch (Entry.Register)
|
||||
{
|
||||
case NsGpuRegister.BindChannel:
|
||||
if (Entry.Arguments.Count > 0)
|
||||
{
|
||||
SubChannels[Entry.SubChannel] = (NsGpuEngine)Entry.Arguments[0];
|
||||
}
|
||||
break;
|
||||
|
||||
case NsGpuRegister._3dVertexArray0Fetch:
|
||||
SendVertexBuffers(Memory);
|
||||
break;
|
||||
|
||||
case NsGpuRegister._3dCbData0:
|
||||
if (GetRegister(NsGpuRegister._3dCbPos) == 0x20)
|
||||
{
|
||||
SendTexture(Memory);
|
||||
}
|
||||
break;
|
||||
|
||||
case NsGpuRegister._3dQueryAddressHigh:
|
||||
case NsGpuRegister._3dQueryAddressLow:
|
||||
case NsGpuRegister._3dQuerySequence:
|
||||
case NsGpuRegister._3dQueryGet:
|
||||
HasQuery = true;
|
||||
break;
|
||||
|
||||
case NsGpuRegister._3dSetShader:
|
||||
uint ShaderPrg = (uint)Entry.Arguments[0];
|
||||
uint ShaderId = (uint)Entry.Arguments[1];
|
||||
uint CodeAddr = (uint)Entry.Arguments[2];
|
||||
uint ShaderType = (uint)Entry.Arguments[3];
|
||||
uint CodeEnd = (uint)Entry.Arguments[4];
|
||||
|
||||
SendShader(
|
||||
Memory,
|
||||
ShaderPrg,
|
||||
ShaderId,
|
||||
CodeAddr,
|
||||
ShaderType,
|
||||
CodeEnd);
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (HasQuery)
|
||||
{
|
||||
long Position =
|
||||
(long)GetRegister(NsGpuRegister._3dQueryAddressHigh) << 32 |
|
||||
(long)GetRegister(NsGpuRegister._3dQueryAddressLow) << 0;
|
||||
|
||||
uint Seq = GetRegister(NsGpuRegister._3dQuerySequence);
|
||||
uint Get = GetRegister(NsGpuRegister._3dQueryGet);
|
||||
|
||||
uint Mode = Get & 3;
|
||||
|
||||
if (Mode == 0)
|
||||
{
|
||||
//Write
|
||||
Position = Gpu.MemoryMgr.GetCpuAddr(Position);
|
||||
|
||||
if (Position != -1)
|
||||
{
|
||||
Gpu.Renderer.QueueAction(delegate()
|
||||
{
|
||||
Memory.WriteUInt32(Position, Seq);
|
||||
});
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void SendVertexBuffers(AMemory Memory)
|
||||
{
|
||||
long Position =
|
||||
(long)GetRegister(NsGpuRegister._3dVertexArray0StartHigh) << 32 |
|
||||
(long)GetRegister(NsGpuRegister._3dVertexArray0StartLow) << 0;
|
||||
|
||||
long Limit =
|
||||
(long)GetRegister(NsGpuRegister._3dVertexArray0LimitHigh) << 32 |
|
||||
(long)GetRegister(NsGpuRegister._3dVertexArray0LimitLow) << 0;
|
||||
|
||||
int VbIndex = CurrentVertexBuffers.Count;
|
||||
|
||||
if (!CurrentVertexBuffers.TryAdd(Position, VbIndex))
|
||||
{
|
||||
VbIndex = CurrentVertexBuffers[Position];
|
||||
}
|
||||
|
||||
if (Limit != 0)
|
||||
{
|
||||
long Size = (Limit - Position) + 1;
|
||||
|
||||
Position = Gpu.MemoryMgr.GetCpuAddr(Position);
|
||||
|
||||
if (Position != -1)
|
||||
{
|
||||
byte[] Buffer = AMemoryHelper.ReadBytes(Memory, Position, Size);
|
||||
|
||||
int Stride = (int)GetRegister(NsGpuRegister._3dVertexArray0Fetch) & 0xfff;
|
||||
|
||||
List<GalVertexAttrib> Attribs = new List<GalVertexAttrib>();
|
||||
|
||||
for (int Attr = 0; Attr < 16; Attr++)
|
||||
{
|
||||
int Packed = (int)GetRegister(NsGpuRegister._3dVertexAttrib0Format + Attr * 4);
|
||||
|
||||
GalVertexAttrib Attrib = new GalVertexAttrib(Attr,
|
||||
(Packed >> 0) & 0x1f,
|
||||
((Packed >> 6) & 0x1) != 0,
|
||||
(Packed >> 7) & 0x3fff,
|
||||
(GalVertexAttribSize)((Packed >> 21) & 0x3f),
|
||||
(GalVertexAttribType)((Packed >> 27) & 0x7),
|
||||
((Packed >> 31) & 0x1) != 0);
|
||||
|
||||
if (Attrib.Offset < Stride)
|
||||
{
|
||||
Attribs.Add(Attrib);
|
||||
}
|
||||
}
|
||||
|
||||
Gpu.Renderer.QueueAction(delegate()
|
||||
{
|
||||
Gpu.Renderer.SendVertexBuffer(VbIndex, Buffer, Stride, Attribs.ToArray());
|
||||
});
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void SendTexture(AMemory Memory)
|
||||
{
|
||||
long TicPos = (long)GetRegister(NsGpuRegister._3dTicAddressHigh) << 32 |
|
||||
(long)GetRegister(NsGpuRegister._3dTicAddressLow) << 0;
|
||||
|
||||
uint CbData = GetRegister(NsGpuRegister._3dCbData0);
|
||||
|
||||
uint TicIndex = (CbData >> 0) & 0xfffff;
|
||||
uint TscIndex = (CbData >> 20) & 0xfff; //I guess?
|
||||
|
||||
TicPos = Gpu.MemoryMgr.GetCpuAddr(TicPos + TicIndex * 0x20);
|
||||
|
||||
if (TicPos != -1)
|
||||
{
|
||||
int Word0 = Memory.ReadInt32(TicPos + 0x0);
|
||||
int Word1 = Memory.ReadInt32(TicPos + 0x4);
|
||||
int Word2 = Memory.ReadInt32(TicPos + 0x8);
|
||||
int Word3 = Memory.ReadInt32(TicPos + 0xc);
|
||||
int Word4 = Memory.ReadInt32(TicPos + 0x10);
|
||||
int Word5 = Memory.ReadInt32(TicPos + 0x14);
|
||||
int Word6 = Memory.ReadInt32(TicPos + 0x18);
|
||||
int Word7 = Memory.ReadInt32(TicPos + 0x1c);
|
||||
|
||||
long TexAddress = Word1;
|
||||
|
||||
TexAddress |= (long)(Word2 & 0xff) << 32;
|
||||
|
||||
TexAddress = Gpu.MemoryMgr.GetCpuAddr(TexAddress);
|
||||
|
||||
if (TexAddress != -1)
|
||||
{
|
||||
NsGpuTextureFormat Format = (NsGpuTextureFormat)(Word0 & 0x7f);
|
||||
|
||||
int Width = (Word4 & 0xffff) + 1;
|
||||
int Height = (Word5 & 0xffff) + 1;
|
||||
|
||||
byte[] Buffer = GetDecodedTexture(Memory, Format, TexAddress, Width, Height);
|
||||
|
||||
if (Buffer != null)
|
||||
{
|
||||
Gpu.Renderer.QueueAction(delegate()
|
||||
{
|
||||
Gpu.Renderer.SendR8G8B8A8Texture(0, Buffer, Width, Height);
|
||||
});
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void SendShader(
|
||||
AMemory Memory,
|
||||
uint ShaderPrg,
|
||||
uint ShaderId,
|
||||
uint CodeAddr,
|
||||
uint ShaderType,
|
||||
uint CodeEnd)
|
||||
{
|
||||
long CodePos = Gpu.MemoryMgr.GetCpuAddr(CodeAddr);
|
||||
|
||||
byte[] Data = AMemoryHelper.ReadBytes(Memory, CodePos, 0x300);
|
||||
}
|
||||
|
||||
private static byte[] GetDecodedTexture(
|
||||
AMemory Memory,
|
||||
NsGpuTextureFormat Format,
|
||||
long Position,
|
||||
int Width,
|
||||
int Height)
|
||||
{
|
||||
byte[] Data = null;
|
||||
|
||||
switch (Format)
|
||||
{
|
||||
case NsGpuTextureFormat.BC1:
|
||||
{
|
||||
int Size = (Width * Height) >> 1;
|
||||
|
||||
Data = AMemoryHelper.ReadBytes(Memory, Position, Size);
|
||||
|
||||
Data = BCn.DecodeBC1(new NsGpuTexture()
|
||||
{
|
||||
Width = Width,
|
||||
Height = Height,
|
||||
Data = Data
|
||||
}, 0);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case NsGpuTextureFormat.BC2:
|
||||
{
|
||||
int Size = Width * Height;
|
||||
|
||||
Data = AMemoryHelper.ReadBytes(Memory, Position, Size);
|
||||
|
||||
Data = BCn.DecodeBC2(new NsGpuTexture()
|
||||
{
|
||||
Width = Width,
|
||||
Height = Height,
|
||||
Data = Data
|
||||
}, 0);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case NsGpuTextureFormat.BC3:
|
||||
{
|
||||
int Size = Width * Height;
|
||||
|
||||
Data = AMemoryHelper.ReadBytes(Memory, Position, Size);
|
||||
|
||||
Data = BCn.DecodeBC3(new NsGpuTexture()
|
||||
{
|
||||
Width = Width,
|
||||
Height = Height,
|
||||
Data = Data
|
||||
}, 0);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
//default: throw new NotImplementedException(Format.ToString());
|
||||
}
|
||||
|
||||
return Data;
|
||||
}
|
||||
|
||||
public uint GetRegister(NsGpuRegister Register)
|
||||
{
|
||||
return Registers[((int)Register >> 2) & 0xfff];
|
||||
}
|
||||
|
||||
public void SetRegister(NsGpuRegister Register, uint Value)
|
||||
{
|
||||
Registers[((int)Register >> 2) & 0xfff] = Value;
|
||||
}
|
||||
}
|
||||
}
|
|
@ -1,94 +0,0 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
public enum NsGpuRegister
|
||||
{
|
||||
BindChannel = 0,
|
||||
|
||||
_2dClipEnable = 0x0290,
|
||||
_2dOperation = 0x02ac,
|
||||
|
||||
_3dGlobalBase = 0x02c8,
|
||||
_3dRt0AddressHigh = 0x0800,
|
||||
_3dRt0AddressLow = 0x0804,
|
||||
_3dRt0Horiz = 0x0808,
|
||||
_3dRt0Vert = 0x080c,
|
||||
_3dRt0Format = 0x0810,
|
||||
_3dRt0BlockDimensions = 0x0814,
|
||||
_3dRt0ArrayMode = 0x0818,
|
||||
_3dRt0LayerStride = 0x081c,
|
||||
_3dRt0BaseLayer = 0x0820,
|
||||
_3dViewportScaleX = 0x0a00,
|
||||
_3dViewportScaleY = 0x0a04,
|
||||
_3dViewportScaleZ = 0x0a08,
|
||||
_3dViewportTranslateX = 0x0a0c,
|
||||
_3dViewportTranslateY = 0x0a10,
|
||||
_3dViewportTranslateZ = 0x0a14,
|
||||
_3dViewportHoriz = 0x0c00,
|
||||
_3dViewportVert = 0x0c04,
|
||||
_3dDepthRangeNear = 0x0c08,
|
||||
_3dDepthRangeFar = 0x0c0c,
|
||||
_3dClearColorR = 0x0d80,
|
||||
_3dClearColorG = 0x0d84,
|
||||
_3dClearColorB = 0x0d88,
|
||||
_3dClearColorA = 0x0d8c,
|
||||
_3dScreenScissorHoriz = 0x0ff4,
|
||||
_3dScreenScissorVert = 0x0ff8,
|
||||
_3dVertexAttrib0Format = 0x1160,
|
||||
_3dVertexAttrib1Format = 0x1164,
|
||||
_3dVertexAttrib2Format = 0x1168,
|
||||
_3dVertexAttrib3Format = 0x116c,
|
||||
_3dVertexAttrib4Format = 0x1170,
|
||||
_3dVertexAttrib5Format = 0x1174,
|
||||
_3dVertexAttrib6Format = 0x1178,
|
||||
_3dVertexAttrib7Format = 0x117c,
|
||||
_3dVertexAttrib8Format = 0x1180,
|
||||
_3dVertexAttrib9Format = 0x1184,
|
||||
_3dVertexAttrib10Format = 0x1188,
|
||||
_3dVertexAttrib11Format = 0x118c,
|
||||
_3dVertexAttrib12Format = 0x1190,
|
||||
_3dVertexAttrib13Format = 0x1194,
|
||||
_3dVertexAttrib14Format = 0x1198,
|
||||
_3dVertexAttrib15Format = 0x119c,
|
||||
_3dScreenYControl = 0x13ac,
|
||||
_3dTscAddressHigh = 0x155c,
|
||||
_3dTscAddressLow = 0x1560,
|
||||
_3dTscLimit = 0x1564,
|
||||
_3dTicAddressHigh = 0x1574,
|
||||
_3dTicAddressLow = 0x1578,
|
||||
_3dTicLimit = 0x157c,
|
||||
_3dMultiSampleMode = 0x15d0,
|
||||
_3dVertexEndGl = 0x1614,
|
||||
_3dVertexBeginGl = 0x1618,
|
||||
_3dQueryAddressHigh = 0x1b00,
|
||||
_3dQueryAddressLow = 0x1b04,
|
||||
_3dQuerySequence = 0x1b08,
|
||||
_3dQueryGet = 0x1b0c,
|
||||
_3dVertexArray0Fetch = 0x1c00,
|
||||
_3dVertexArray0StartHigh = 0x1c04,
|
||||
_3dVertexArray0StartLow = 0x1c08,
|
||||
_3dVertexArray1Fetch = 0x1c10, //todo: the rest
|
||||
_3dVertexArray0LimitHigh = 0x1f00,
|
||||
_3dVertexArray0LimitLow = 0x1f04,
|
||||
_3dCbSize = 0x2380,
|
||||
_3dCbAddressHigh = 0x2384,
|
||||
_3dCbAddressLow = 0x2388,
|
||||
_3dCbPos = 0x238c,
|
||||
_3dCbData0 = 0x2390,
|
||||
_3dCbData1 = 0x2394,
|
||||
_3dCbData2 = 0x2398,
|
||||
_3dCbData3 = 0x239c,
|
||||
_3dCbData4 = 0x23a0,
|
||||
_3dCbData5 = 0x23a4,
|
||||
_3dCbData6 = 0x23a8,
|
||||
_3dCbData7 = 0x23ac,
|
||||
_3dCbData8 = 0x23b0,
|
||||
_3dCbData9 = 0x23b4,
|
||||
_3dCbData10 = 0x23b8,
|
||||
_3dCbData11 = 0x23bc,
|
||||
_3dCbData12 = 0x23c0,
|
||||
_3dCbData13 = 0x23c4,
|
||||
_3dCbData14 = 0x23c8,
|
||||
_3dCbData15 = 0x23cc,
|
||||
_3dSetShader = 0x3890
|
||||
}
|
||||
}
|
|
@ -1,10 +0,0 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
struct NsGpuTexture
|
||||
{
|
||||
public int Width;
|
||||
public int Height;
|
||||
|
||||
public byte[] Data;
|
||||
}
|
||||
}
|
|
@ -1,9 +0,0 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
enum NsGpuTextureFormat
|
||||
{
|
||||
BC1 = 0x24,
|
||||
BC2 = 0x25,
|
||||
BC3 = 0x26
|
||||
}
|
||||
}
|
|
@ -1,13 +1,11 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
enum NsGpuEngine
|
||||
enum NvGpuEngine
|
||||
{
|
||||
None = 0,
|
||||
_2d = 0x902d,
|
||||
_3d = 0xb197,
|
||||
Compute = 0xb1c0,
|
||||
Kepler = 0xa140,
|
||||
Dma = 0xb0b5,
|
||||
GpFifo = 0xb06f
|
||||
Dma = 0xb0b5
|
||||
}
|
||||
}
|
469
Ryujinx.Graphics/Gpu/NvGpuEngine3d.cs
Normal file
469
Ryujinx.Graphics/Gpu/NvGpuEngine3d.cs
Normal file
|
@ -0,0 +1,469 @@
|
|||
using ChocolArm64.Memory;
|
||||
using Ryujinx.Graphics.Gal;
|
||||
using System;
|
||||
using System.Collections.Generic;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
class NvGpuEngine3d : INvGpuEngine
|
||||
{
|
||||
public int[] Registers { get; private set; }
|
||||
|
||||
private NsGpu Gpu;
|
||||
|
||||
private Dictionary<int, NvGpuMethod> Methods;
|
||||
|
||||
private struct ConstBuffer
|
||||
{
|
||||
public bool Enabled;
|
||||
public long Position;
|
||||
public int Size;
|
||||
}
|
||||
|
||||
private ConstBuffer[] Cbs;
|
||||
|
||||
private bool HasDataToRender;
|
||||
|
||||
public NvGpuEngine3d(NsGpu Gpu)
|
||||
{
|
||||
this.Gpu = Gpu;
|
||||
|
||||
Registers = new int[0xe00];
|
||||
|
||||
Methods = new Dictionary<int, NvGpuMethod>();
|
||||
|
||||
void AddMethod(int Meth, int Count, int Stride, NvGpuMethod Method)
|
||||
{
|
||||
while (Count-- > 0)
|
||||
{
|
||||
Methods.Add(Meth, Method);
|
||||
|
||||
Meth += Stride;
|
||||
}
|
||||
}
|
||||
|
||||
AddMethod(0x585, 1, 1, VertexEndGl);
|
||||
AddMethod(0x674, 1, 1, ClearBuffers);
|
||||
AddMethod(0x6c3, 1, 1, QueryControl);
|
||||
AddMethod(0x8e4, 16, 1, CbData);
|
||||
AddMethod(0x904, 1, 1, CbBind);
|
||||
|
||||
Cbs = new ConstBuffer[18];
|
||||
}
|
||||
|
||||
public void CallMethod(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
if (Methods.TryGetValue(PBEntry.Method, out NvGpuMethod Method))
|
||||
{
|
||||
Method(Memory, PBEntry);
|
||||
}
|
||||
else
|
||||
{
|
||||
WriteRegister(PBEntry);
|
||||
}
|
||||
}
|
||||
|
||||
private void VertexEndGl(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
SetFrameBuffer(0);
|
||||
|
||||
long[] Tags = UploadShaders(Memory);
|
||||
|
||||
Gpu.Renderer.BindProgram();
|
||||
|
||||
SetAlphaBlending();
|
||||
|
||||
UploadTextures(Memory, Tags);
|
||||
UploadUniforms(Memory);
|
||||
UploadVertexArrays(Memory);
|
||||
|
||||
HasDataToRender = true;
|
||||
}
|
||||
|
||||
private void ClearBuffers(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
if (HasDataToRender)
|
||||
{
|
||||
HasDataToRender = false;
|
||||
|
||||
Gpu.Renderer.DrawFrameBuffer(0);
|
||||
}
|
||||
|
||||
int Arg0 = PBEntry.Arguments[0];
|
||||
|
||||
int FbIndex = (Arg0 >> 6) & 0xf;
|
||||
|
||||
int Layer = (Arg0 >> 10) & 0x3ff;
|
||||
|
||||
GalClearBufferFlags Flags = (GalClearBufferFlags)(Arg0 & 0x3f);
|
||||
|
||||
SetFrameBuffer(0);
|
||||
|
||||
Gpu.Renderer.ClearBuffers(Layer, Flags);
|
||||
}
|
||||
|
||||
private void SetFrameBuffer(int FbIndex)
|
||||
{
|
||||
int Width = ReadRegister(NvGpuEngine3dReg.FrameBufferNWidth + FbIndex * 0x10);
|
||||
int Height = ReadRegister(NvGpuEngine3dReg.FrameBufferNHeight + FbIndex * 0x10);
|
||||
|
||||
Gpu.Renderer.SetFb(FbIndex, Width, Height);
|
||||
Gpu.Renderer.BindFrameBuffer(FbIndex);
|
||||
}
|
||||
|
||||
private long[] UploadShaders(AMemory Memory)
|
||||
{
|
||||
long[] Tags = new long[5];
|
||||
|
||||
long BasePosition = MakeInt64From2xInt32(NvGpuEngine3dReg.ShaderAddress);
|
||||
|
||||
for (int Index = 0; Index < 6; Index++)
|
||||
{
|
||||
int Control = ReadRegister(NvGpuEngine3dReg.ShaderNControl + Index * 0x10);
|
||||
int Offset = ReadRegister(NvGpuEngine3dReg.ShaderNOffset + Index * 0x10);
|
||||
|
||||
//Note: Vertex Program (B) is always enabled.
|
||||
bool Enable = (Control & 1) != 0 || Index == 1;
|
||||
|
||||
if (!Enable)
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
long Tag = BasePosition + (uint)Offset;
|
||||
|
||||
long Position = Gpu.GetCpuAddr(Tag);
|
||||
|
||||
//TODO: Find a better way to calculate the size.
|
||||
int Size = 0x20000;
|
||||
|
||||
byte[] Code = AMemoryHelper.ReadBytes(Memory, Position, (uint)Size);
|
||||
|
||||
GalShaderType ShaderType = GetTypeFromProgram(Index);
|
||||
|
||||
Tags[(int)ShaderType] = Tag;
|
||||
|
||||
Gpu.Renderer.CreateShader(Tag, ShaderType, Code);
|
||||
Gpu.Renderer.BindShader(Tag);
|
||||
}
|
||||
|
||||
return Tags;
|
||||
}
|
||||
|
||||
private static GalShaderType GetTypeFromProgram(int Program)
|
||||
{
|
||||
switch (Program)
|
||||
{
|
||||
case 0:
|
||||
case 1: return GalShaderType.Vertex;
|
||||
case 2: return GalShaderType.TessControl;
|
||||
case 3: return GalShaderType.TessEvaluation;
|
||||
case 4: return GalShaderType.Geometry;
|
||||
case 5: return GalShaderType.Fragment;
|
||||
}
|
||||
|
||||
throw new ArgumentOutOfRangeException(nameof(Program));
|
||||
}
|
||||
|
||||
private void SetAlphaBlending()
|
||||
{
|
||||
bool BlendEnableMaster = (ReadRegister(NvGpuEngine3dReg.BlendEnableMaster) & 1) != 0;
|
||||
|
||||
Gpu.Renderer.SetBlendEnable(BlendEnableMaster);
|
||||
|
||||
bool BlendSeparateAlpha = (ReadRegister(NvGpuEngine3dReg.BlendSeparateAlpha) & 1) != 0;
|
||||
|
||||
GalBlendEquation EquationRgb = (GalBlendEquation)ReadRegister(NvGpuEngine3dReg.BlendEquationRgb);
|
||||
|
||||
GalBlendFactor FuncSrcRgb = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.BlendFuncSrcRgb);
|
||||
GalBlendFactor FuncDstRgb = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.BlendFuncDstRgb);
|
||||
|
||||
if (BlendSeparateAlpha)
|
||||
{
|
||||
GalBlendEquation EquationAlpha = (GalBlendEquation)ReadRegister(NvGpuEngine3dReg.BlendEquationAlpha);
|
||||
|
||||
GalBlendFactor FuncSrcAlpha = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.BlendFuncSrcAlpha);
|
||||
GalBlendFactor FuncDstAlpha = (GalBlendFactor)ReadRegister(NvGpuEngine3dReg.BlendFuncDstAlpha);
|
||||
|
||||
Gpu.Renderer.SetBlendSeparate(
|
||||
EquationRgb,
|
||||
EquationAlpha,
|
||||
FuncSrcRgb,
|
||||
FuncDstRgb,
|
||||
FuncSrcAlpha,
|
||||
FuncDstAlpha);
|
||||
}
|
||||
else
|
||||
{
|
||||
Gpu.Renderer.SetBlend(EquationRgb, FuncSrcRgb, FuncDstRgb);
|
||||
}
|
||||
}
|
||||
|
||||
private void UploadTextures(AMemory Memory, long[] Tags)
|
||||
{
|
||||
long BaseShPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.ShaderAddress);
|
||||
|
||||
int TextureCbIndex = ReadRegister(NvGpuEngine3dReg.TextureCbIndex);
|
||||
|
||||
long BasePosition = Cbs[TextureCbIndex].Position;
|
||||
|
||||
long Size = (uint)Cbs[TextureCbIndex].Size;
|
||||
|
||||
int TexIndex = 0;
|
||||
|
||||
for (int Index = 0; Index < Tags.Length; Index++)
|
||||
{
|
||||
foreach (ShaderDeclInfo DeclInfo in Gpu.Renderer.GetTextureUsage(Tags[Index]))
|
||||
{
|
||||
long Position = BasePosition + Index * Size;
|
||||
|
||||
UploadTexture(Memory, Position, TexIndex, DeclInfo.Index);
|
||||
|
||||
Gpu.Renderer.SetUniform1(DeclInfo.Name, TexIndex);
|
||||
|
||||
TexIndex++;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void UploadTexture(AMemory Memory, long BasePosition, int TexIndex, int HndIndex)
|
||||
{
|
||||
long Position = BasePosition + HndIndex * 4;
|
||||
|
||||
int TextureHandle = Memory.ReadInt32(Position);
|
||||
|
||||
int TicIndex = (TextureHandle >> 0) & 0xfffff;
|
||||
int TscIndex = (TextureHandle >> 20) & 0xfff;
|
||||
|
||||
TryGetCpuAddr(NvGpuEngine3dReg.TexHeaderPoolOffset, out long TicPosition);
|
||||
TryGetCpuAddr(NvGpuEngine3dReg.TexSamplerPoolOffset, out long TscPosition);
|
||||
|
||||
TicPosition += TicIndex * 0x20;
|
||||
TscPosition += TscIndex * 0x20;
|
||||
|
||||
Gpu.Renderer.SetTexture(TexIndex, TextureFactory.MakeTexture(Gpu, Memory, TicPosition));
|
||||
Gpu.Renderer.SetSampler(TexIndex, TextureFactory.MakeSampler(Gpu, Memory, TscPosition));
|
||||
}
|
||||
|
||||
private void UploadUniforms(AMemory Memory)
|
||||
{
|
||||
long BasePosition = MakeInt64From2xInt32(NvGpuEngine3dReg.ShaderAddress);
|
||||
|
||||
for (int Index = 0; Index < 5; Index++)
|
||||
{
|
||||
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)
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
for (int Cbuf = 0; Cbuf < Cbs.Length; Cbuf++)
|
||||
{
|
||||
ConstBuffer Cb = Cbs[Cbuf];
|
||||
|
||||
if (Cb.Enabled)
|
||||
{
|
||||
long CbPosition = Cb.Position + Index * Cb.Size;
|
||||
|
||||
byte[] Data = AMemoryHelper.ReadBytes(Memory, CbPosition, (uint)Cb.Size);
|
||||
|
||||
Gpu.Renderer.SetConstBuffer(BasePosition + (uint)Offset, Cbuf, Data);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void UploadVertexArrays(AMemory Memory)
|
||||
{
|
||||
long IndexPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.IndexArrayAddress);
|
||||
|
||||
int IndexSize = ReadRegister(NvGpuEngine3dReg.IndexArrayFormat);
|
||||
int IndexFirst = ReadRegister(NvGpuEngine3dReg.IndexBatchFirst);
|
||||
int IndexCount = ReadRegister(NvGpuEngine3dReg.IndexBatchCount);
|
||||
|
||||
GalIndexFormat IndexFormat = (GalIndexFormat)IndexSize;
|
||||
|
||||
IndexSize = 1 << IndexSize;
|
||||
|
||||
if (IndexSize > 4)
|
||||
{
|
||||
throw new InvalidOperationException();
|
||||
}
|
||||
|
||||
if (IndexSize != 0)
|
||||
{
|
||||
IndexPosition = Gpu.GetCpuAddr(IndexPosition);
|
||||
|
||||
int BufferSize = IndexCount * IndexSize;
|
||||
|
||||
byte[] Data = AMemoryHelper.ReadBytes(Memory, IndexPosition, BufferSize);
|
||||
|
||||
Gpu.Renderer.SetIndexArray(Data, IndexFormat);
|
||||
}
|
||||
|
||||
List<GalVertexAttrib>[] Attribs = new List<GalVertexAttrib>[32];
|
||||
|
||||
for (int Attr = 0; Attr < 16; Attr++)
|
||||
{
|
||||
int Packed = ReadRegister(NvGpuEngine3dReg.VertexAttribNFormat + Attr);
|
||||
|
||||
int ArrayIndex = Packed & 0x1f;
|
||||
|
||||
if (Attribs[ArrayIndex] == null)
|
||||
{
|
||||
Attribs[ArrayIndex] = new List<GalVertexAttrib>();
|
||||
}
|
||||
|
||||
Attribs[ArrayIndex].Add(new GalVertexAttrib(
|
||||
((Packed >> 6) & 0x1) != 0,
|
||||
(Packed >> 7) & 0x3fff,
|
||||
(GalVertexAttribSize)((Packed >> 21) & 0x3f),
|
||||
(GalVertexAttribType)((Packed >> 27) & 0x7),
|
||||
((Packed >> 31) & 0x1) != 0));
|
||||
}
|
||||
|
||||
for (int Index = 0; Index < 32; Index++)
|
||||
{
|
||||
int Control = ReadRegister(NvGpuEngine3dReg.VertexArrayNControl + Index * 4);
|
||||
|
||||
bool Enable = (Control & 0x1000) != 0;
|
||||
|
||||
if (!Enable)
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
long VertexPosition = MakeInt64From2xInt32(NvGpuEngine3dReg.VertexArrayNAddress + Index * 4);
|
||||
long VertexEndPos = MakeInt64From2xInt32(NvGpuEngine3dReg.VertexArrayNEndAddr + Index * 4);
|
||||
|
||||
long Size = (VertexEndPos - VertexPosition) + 1;
|
||||
|
||||
int Stride = Control & 0xfff;
|
||||
|
||||
VertexPosition = Gpu.GetCpuAddr(VertexPosition);
|
||||
|
||||
byte[] Data = AMemoryHelper.ReadBytes(Memory, VertexPosition, Size);
|
||||
|
||||
GalVertexAttrib[] AttribArray = Attribs[Index]?.ToArray() ?? new GalVertexAttrib[0];
|
||||
|
||||
Gpu.Renderer.SetVertexArray(Index, Stride, Data, AttribArray);
|
||||
|
||||
int PrimCtrl = ReadRegister(NvGpuEngine3dReg.VertexBeginGl);
|
||||
|
||||
GalPrimitiveType PrimType = (GalPrimitiveType)(PrimCtrl & 0xffff);
|
||||
|
||||
if (IndexCount != 0)
|
||||
{
|
||||
Gpu.Renderer.DrawElements(Index, IndexFirst, PrimType);
|
||||
}
|
||||
else
|
||||
{
|
||||
Gpu.Renderer.DrawArrays(Index, PrimType);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void QueryControl(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
if (TryGetCpuAddr(NvGpuEngine3dReg.QueryAddress, out long Position))
|
||||
{
|
||||
int Seq = Registers[(int)NvGpuEngine3dReg.QuerySequence];
|
||||
int Ctrl = Registers[(int)NvGpuEngine3dReg.QueryControl];
|
||||
|
||||
int Mode = Ctrl & 3;
|
||||
|
||||
if (Mode == 0)
|
||||
{
|
||||
//Write.
|
||||
Memory.WriteInt32(Position, Seq);
|
||||
}
|
||||
}
|
||||
|
||||
WriteRegister(PBEntry);
|
||||
}
|
||||
|
||||
private void CbData(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
if (TryGetCpuAddr(NvGpuEngine3dReg.ConstBufferNAddress, out long Position))
|
||||
{
|
||||
int Offset = ReadRegister(NvGpuEngine3dReg.ConstBufferNOffset);
|
||||
|
||||
foreach (int Arg in PBEntry.Arguments)
|
||||
{
|
||||
Memory.WriteInt32(Position + Offset, Arg);
|
||||
|
||||
Offset += 4;
|
||||
}
|
||||
|
||||
WriteRegister(NvGpuEngine3dReg.ConstBufferNOffset, Offset);
|
||||
}
|
||||
}
|
||||
|
||||
private void CbBind(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
int Index = PBEntry.Arguments[0];
|
||||
|
||||
bool Enabled = (Index & 1) != 0;
|
||||
|
||||
Index = (Index >> 4) & 0x1f;
|
||||
|
||||
if (TryGetCpuAddr(NvGpuEngine3dReg.ConstBufferNAddress, out long Position))
|
||||
{
|
||||
Cbs[Index].Position = Position;
|
||||
Cbs[Index].Enabled = Enabled;
|
||||
|
||||
Cbs[Index].Size = ReadRegister(NvGpuEngine3dReg.ConstBufferNSize);
|
||||
}
|
||||
}
|
||||
|
||||
private int ReadCb(AMemory Memory, int Cbuf, int Offset)
|
||||
{
|
||||
long Position = Cbs[Cbuf].Position;
|
||||
|
||||
int Value = Memory.ReadInt32(Position + Offset);
|
||||
|
||||
return Value;
|
||||
}
|
||||
|
||||
private bool TryGetCpuAddr(NvGpuEngine3dReg Reg, out long Position)
|
||||
{
|
||||
Position = MakeInt64From2xInt32(Reg);
|
||||
|
||||
Position = Gpu.GetCpuAddr(Position);
|
||||
|
||||
return Position != -1;
|
||||
}
|
||||
|
||||
private long MakeInt64From2xInt32(NvGpuEngine3dReg Reg)
|
||||
{
|
||||
return
|
||||
(long)Registers[(int)Reg + 0] << 32 |
|
||||
(uint)Registers[(int)Reg + 1];
|
||||
}
|
||||
|
||||
private void WriteRegister(NsGpuPBEntry PBEntry)
|
||||
{
|
||||
int ArgsCount = PBEntry.Arguments.Count;
|
||||
|
||||
if (ArgsCount > 0)
|
||||
{
|
||||
Registers[PBEntry.Method] = PBEntry.Arguments[ArgsCount - 1];
|
||||
}
|
||||
}
|
||||
|
||||
private int ReadRegister(NvGpuEngine3dReg Reg)
|
||||
{
|
||||
return Registers[(int)Reg];
|
||||
}
|
||||
|
||||
private void WriteRegister(NvGpuEngine3dReg Reg, int Value)
|
||||
{
|
||||
Registers[(int)Reg] = Value;
|
||||
}
|
||||
}
|
||||
}
|
44
Ryujinx.Graphics/Gpu/NvGpuEngine3dReg.cs
Normal file
44
Ryujinx.Graphics/Gpu/NvGpuEngine3dReg.cs
Normal file
|
@ -0,0 +1,44 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
enum NvGpuEngine3dReg
|
||||
{
|
||||
FrameBufferNAddress = 0x200,
|
||||
FrameBufferNWidth = 0x202,
|
||||
FrameBufferNHeight = 0x203,
|
||||
FrameBufferNFormat = 0x204,
|
||||
VertexAttribNFormat = 0x458,
|
||||
BlendSeparateAlpha = 0x4cf,
|
||||
BlendEquationRgb = 0x4d0,
|
||||
BlendFuncSrcRgb = 0x4d1,
|
||||
BlendFuncDstRgb = 0x4d2,
|
||||
BlendEquationAlpha = 0x4d3,
|
||||
BlendFuncSrcAlpha = 0x4d4,
|
||||
BlendFuncDstAlpha = 0x4d6,
|
||||
BlendEnableMaster = 0x4d7,
|
||||
VertexArrayElemBase = 0x50d,
|
||||
TexHeaderPoolOffset = 0x55d,
|
||||
TexSamplerPoolOffset = 0x557,
|
||||
ShaderAddress = 0x582,
|
||||
VertexBeginGl = 0x586,
|
||||
IndexArrayAddress = 0x5f2,
|
||||
IndexArrayEndAddr = 0x5f4,
|
||||
IndexArrayFormat = 0x5f6,
|
||||
IndexBatchFirst = 0x5f7,
|
||||
IndexBatchCount = 0x5f8,
|
||||
QueryAddress = 0x6c0,
|
||||
QuerySequence = 0x6c2,
|
||||
QueryControl = 0x6c3,
|
||||
VertexArrayNControl = 0x700,
|
||||
VertexArrayNAddress = 0x701,
|
||||
VertexArrayNDivisor = 0x703,
|
||||
VertexArrayNEndAddr = 0x7c0,
|
||||
ShaderNControl = 0x800,
|
||||
ShaderNOffset = 0x801,
|
||||
ShaderNMaxGprs = 0x803,
|
||||
ShaderNType = 0x804,
|
||||
ConstBufferNSize = 0x8e0,
|
||||
ConstBufferNAddress = 0x8e1,
|
||||
ConstBufferNOffset = 0x8e3,
|
||||
TextureCbIndex = 0x982
|
||||
}
|
||||
}
|
171
Ryujinx.Graphics/Gpu/NvGpuFifo.cs
Normal file
171
Ryujinx.Graphics/Gpu/NvGpuFifo.cs
Normal file
|
@ -0,0 +1,171 @@
|
|||
using ChocolArm64.Memory;
|
||||
using System.Collections.Concurrent;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
public class NvGpuFifo
|
||||
{
|
||||
private const int MacrosCount = 0x80;
|
||||
private const int MacroIndexMask = MacrosCount - 1;
|
||||
|
||||
private NsGpu Gpu;
|
||||
|
||||
private ConcurrentQueue<(AMemory, NsGpuPBEntry)> BufferQueue;
|
||||
|
||||
private NvGpuEngine[] SubChannels;
|
||||
|
||||
private struct CachedMacro
|
||||
{
|
||||
public long Position { get; private set; }
|
||||
|
||||
private MacroInterpreter Interpreter;
|
||||
|
||||
public CachedMacro(NvGpuFifo PFifo, INvGpuEngine Engine, long Position)
|
||||
{
|
||||
this.Position = Position;
|
||||
|
||||
Interpreter = new MacroInterpreter(PFifo, Engine);
|
||||
}
|
||||
|
||||
public void PushParam(int Param)
|
||||
{
|
||||
Interpreter?.Fifo.Enqueue(Param);
|
||||
}
|
||||
|
||||
public void Execute(AMemory Memory, int Param)
|
||||
{
|
||||
Interpreter?.Execute(Memory, Position, Param);
|
||||
}
|
||||
}
|
||||
|
||||
private long CurrMacroPosition;
|
||||
private int CurrMacroBindIndex;
|
||||
|
||||
private CachedMacro[] Macros;
|
||||
|
||||
public NvGpuFifo(NsGpu Gpu)
|
||||
{
|
||||
this.Gpu = Gpu;
|
||||
|
||||
BufferQueue = new ConcurrentQueue<(AMemory, NsGpuPBEntry)>();
|
||||
|
||||
SubChannels = new NvGpuEngine[8];
|
||||
|
||||
Macros = new CachedMacro[MacrosCount];
|
||||
}
|
||||
|
||||
public void PushBuffer(AMemory Memory, NsGpuPBEntry[] Buffer)
|
||||
{
|
||||
foreach (NsGpuPBEntry PBEntry in Buffer)
|
||||
{
|
||||
BufferQueue.Enqueue((Memory, PBEntry));
|
||||
}
|
||||
}
|
||||
|
||||
public void DispatchCalls()
|
||||
{
|
||||
while (Step());
|
||||
}
|
||||
|
||||
public bool Step()
|
||||
{
|
||||
if (BufferQueue.TryDequeue(out (AMemory Memory, NsGpuPBEntry PBEntry) Tuple))
|
||||
{
|
||||
CallMethod(Tuple.Memory, Tuple.PBEntry);
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
private void CallMethod(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
if (PBEntry.Method < 0x80)
|
||||
{
|
||||
switch ((NvGpuFifoMeth)PBEntry.Method)
|
||||
{
|
||||
case NvGpuFifoMeth.BindChannel:
|
||||
{
|
||||
NvGpuEngine Engine = (NvGpuEngine)PBEntry.Arguments[0];
|
||||
|
||||
SubChannels[PBEntry.SubChannel] = Engine;
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case NvGpuFifoMeth.SetMacroUploadAddress:
|
||||
{
|
||||
CurrMacroPosition = (long)((ulong)PBEntry.Arguments[0] << 2);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case NvGpuFifoMeth.SendMacroCodeData:
|
||||
{
|
||||
long Position = Gpu.GetCpuAddr(CurrMacroPosition);
|
||||
|
||||
foreach (int Arg in PBEntry.Arguments)
|
||||
{
|
||||
Memory.WriteInt32(Position, Arg);
|
||||
|
||||
CurrMacroPosition += 4;
|
||||
|
||||
Position += 4;
|
||||
}
|
||||
break;
|
||||
}
|
||||
|
||||
case NvGpuFifoMeth.SetMacroBindingIndex:
|
||||
{
|
||||
CurrMacroBindIndex = PBEntry.Arguments[0];
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case NvGpuFifoMeth.BindMacro:
|
||||
{
|
||||
long Position = (long)((ulong)PBEntry.Arguments[0] << 2);
|
||||
|
||||
Position = Gpu.GetCpuAddr(Position);
|
||||
|
||||
Macros[CurrMacroBindIndex] = new CachedMacro(this, Gpu.Engine3d, Position);
|
||||
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
switch (SubChannels[PBEntry.SubChannel])
|
||||
{
|
||||
case NvGpuEngine._3d: Call3dMethod(Memory, PBEntry); break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void Call3dMethod(AMemory Memory, NsGpuPBEntry PBEntry)
|
||||
{
|
||||
if (PBEntry.Method < 0xe00)
|
||||
{
|
||||
Gpu.Engine3d.CallMethod(Memory, PBEntry);
|
||||
}
|
||||
else
|
||||
{
|
||||
int MacroIndex = (PBEntry.Method >> 1) & MacroIndexMask;
|
||||
|
||||
if ((PBEntry.Method & 1) != 0)
|
||||
{
|
||||
foreach (int Arg in PBEntry.Arguments)
|
||||
{
|
||||
Macros[MacroIndex].PushParam(Arg);
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
Macros[MacroIndex].Execute(Memory, PBEntry.Arguments[0]);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
11
Ryujinx.Graphics/Gpu/NvGpuFifoMeth.cs
Normal file
11
Ryujinx.Graphics/Gpu/NvGpuFifoMeth.cs
Normal file
|
@ -0,0 +1,11 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
enum NvGpuFifoMeth
|
||||
{
|
||||
BindChannel = 0,
|
||||
SetMacroUploadAddress = 0x45,
|
||||
SendMacroCodeData = 0x46,
|
||||
SetMacroBindingIndex = 0x47,
|
||||
BindMacro = 0x48
|
||||
}
|
||||
}
|
6
Ryujinx.Graphics/Gpu/NvGpuMethod.cs
Normal file
6
Ryujinx.Graphics/Gpu/NvGpuMethod.cs
Normal file
|
@ -0,0 +1,6 @@
|
|||
using ChocolArm64.Memory;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
delegate void NvGpuMethod(AMemory Memory, NsGpuPBEntry PBEntry);
|
||||
}
|
101
Ryujinx.Graphics/Gpu/NvGpuPushBuffer.cs
Normal file
101
Ryujinx.Graphics/Gpu/NvGpuPushBuffer.cs
Normal file
|
@ -0,0 +1,101 @@
|
|||
using System.Collections.Generic;
|
||||
using System.IO;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
public static class NvGpuPushBuffer
|
||||
{
|
||||
private enum SubmissionMode
|
||||
{
|
||||
Incrementing = 1,
|
||||
NonIncrementing = 3,
|
||||
Immediate = 4,
|
||||
IncrementOnce = 5
|
||||
}
|
||||
|
||||
public static NsGpuPBEntry[] Decode(byte[] Data)
|
||||
{
|
||||
using (MemoryStream MS = new MemoryStream(Data))
|
||||
{
|
||||
BinaryReader Reader = new BinaryReader(MS);
|
||||
|
||||
List<NsGpuPBEntry> PushBuffer = new List<NsGpuPBEntry>();
|
||||
|
||||
bool CanRead() => MS.Position + 4 <= MS.Length;
|
||||
|
||||
while (CanRead())
|
||||
{
|
||||
int Packed = Reader.ReadInt32();
|
||||
|
||||
int Meth = (Packed >> 0) & 0x1fff;
|
||||
int SubC = (Packed >> 13) & 7;
|
||||
int Args = (Packed >> 16) & 0x1fff;
|
||||
int Mode = (Packed >> 29) & 7;
|
||||
|
||||
switch ((SubmissionMode)Mode)
|
||||
{
|
||||
case SubmissionMode.Incrementing:
|
||||
{
|
||||
for (int Index = 0; Index < Args && CanRead(); Index++, Meth++)
|
||||
{
|
||||
PushBuffer.Add(new NsGpuPBEntry(Meth, SubC, Reader.ReadInt32()));
|
||||
}
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case SubmissionMode.NonIncrementing:
|
||||
{
|
||||
int[] Arguments = new int[Args];
|
||||
|
||||
for (int Index = 0; Index < Arguments.Length; Index++)
|
||||
{
|
||||
if (!CanRead())
|
||||
{
|
||||
break;
|
||||
}
|
||||
|
||||
Arguments[Index] = Reader.ReadInt32();
|
||||
}
|
||||
|
||||
PushBuffer.Add(new NsGpuPBEntry(Meth, SubC, Arguments));
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case SubmissionMode.Immediate:
|
||||
{
|
||||
PushBuffer.Add(new NsGpuPBEntry(Meth, SubC, Args));
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
case SubmissionMode.IncrementOnce:
|
||||
{
|
||||
if (CanRead())
|
||||
{
|
||||
PushBuffer.Add(new NsGpuPBEntry(Meth, SubC, Reader.ReadInt32()));
|
||||
}
|
||||
|
||||
if (CanRead() && Args > 1)
|
||||
{
|
||||
int[] Arguments = new int[Args - 1];
|
||||
|
||||
for (int Index = 0; Index < Arguments.Length && CanRead(); Index++)
|
||||
{
|
||||
Arguments[Index] = Reader.ReadInt32();
|
||||
}
|
||||
|
||||
PushBuffer.Add(new NsGpuPBEntry(Meth + 1, SubC, Arguments));
|
||||
}
|
||||
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return PushBuffer.ToArray();
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
|
@ -1,144 +0,0 @@
|
|||
using System;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
class SwizzleAddr
|
||||
{
|
||||
private int Width;
|
||||
|
||||
private int XB;
|
||||
private int YB;
|
||||
|
||||
public SwizzleAddr(int Width, int Height, int Pad)
|
||||
{
|
||||
int W = Pow2RoundUp(Width);
|
||||
int H = Pow2RoundUp(Height);
|
||||
|
||||
XB = CountZeros(W);
|
||||
YB = CountZeros(H);
|
||||
|
||||
int HH = H >> 1;
|
||||
|
||||
if (!IsPow2(Height) && Height <= HH + HH / 3 && YB > 3)
|
||||
{
|
||||
YB--;
|
||||
}
|
||||
|
||||
this.Width = RoundSize(Width, Pad);
|
||||
}
|
||||
|
||||
private static int Pow2RoundUp(int Value)
|
||||
{
|
||||
Value--;
|
||||
|
||||
Value |= (Value >> 1);
|
||||
Value |= (Value >> 2);
|
||||
Value |= (Value >> 4);
|
||||
Value |= (Value >> 8);
|
||||
Value |= (Value >> 16);
|
||||
|
||||
return ++Value;
|
||||
}
|
||||
|
||||
private static bool IsPow2(int Value)
|
||||
{
|
||||
return Value != 0 && (Value & (Value - 1)) == 0;
|
||||
}
|
||||
|
||||
private static int CountZeros(int Value)
|
||||
{
|
||||
int Count = 0;
|
||||
|
||||
for (int i = 0; i < 32; i++)
|
||||
{
|
||||
if ((Value & (1 << i)) != 0)
|
||||
{
|
||||
break;
|
||||
}
|
||||
|
||||
Count++;
|
||||
}
|
||||
|
||||
return Count;
|
||||
}
|
||||
|
||||
private static int RoundSize(int Size, int Pad)
|
||||
{
|
||||
int Mask = Pad - 1;
|
||||
|
||||
if ((Size & Mask) != 0)
|
||||
{
|
||||
Size &= ~Mask;
|
||||
Size += Pad;
|
||||
}
|
||||
|
||||
return Size;
|
||||
}
|
||||
|
||||
public int GetSwizzledAddress8(int X, int Y)
|
||||
{
|
||||
return GetSwizzledAddress(X, Y, 4);
|
||||
}
|
||||
|
||||
public int GetSwizzledAddress16(int X, int Y)
|
||||
{
|
||||
return GetSwizzledAddress(X, Y, 3);
|
||||
}
|
||||
|
||||
public int GetSwizzledAddress32(int X, int Y)
|
||||
{
|
||||
return GetSwizzledAddress(X, Y, 2);
|
||||
}
|
||||
|
||||
public int GetSwizzledAddress64(int X, int Y)
|
||||
{
|
||||
return GetSwizzledAddress(X, Y, 1);
|
||||
}
|
||||
|
||||
public int GetSwizzledAddress128(int X, int Y)
|
||||
{
|
||||
return GetSwizzledAddress(X, Y, 0);
|
||||
}
|
||||
|
||||
private int GetSwizzledAddress(int X, int Y, int XBase)
|
||||
{
|
||||
/*
|
||||
* Examples of patterns:
|
||||
* x x y x y y x y 0 0 0 0 64 x 64 dxt5
|
||||
* x x x x x y y y y x y y x y 0 0 0 0 512 x 512 dxt5
|
||||
* y x x x x x x y y y y x y y x y 0 0 0 0 1024 x 1024 dxt5
|
||||
* y y x x x x x x y y y y x y y x y x 0 0 0 2048 x 2048 dxt1
|
||||
* y y y x x x x x x y y y y x y y x y x x 0 0 1024 x 1024 rgba8888
|
||||
*
|
||||
* Read from right to left, LSB first.
|
||||
*/
|
||||
int XCnt = XBase;
|
||||
int YCnt = 1;
|
||||
int XUsed = 0;
|
||||
int YUsed = 0;
|
||||
int Address = 0;
|
||||
|
||||
while (XUsed < XBase + 2 && XUsed + XCnt < XB)
|
||||
{
|
||||
int XMask = (1 << XCnt) - 1;
|
||||
int YMask = (1 << YCnt) - 1;
|
||||
|
||||
Address |= (X & XMask) << XUsed + YUsed;
|
||||
Address |= (Y & YMask) << XUsed + YUsed + XCnt;
|
||||
|
||||
X >>= XCnt;
|
||||
Y >>= YCnt;
|
||||
|
||||
XUsed += XCnt;
|
||||
YUsed += YCnt;
|
||||
|
||||
XCnt = Math.Min(XB - XUsed, 1);
|
||||
YCnt = Math.Min(YB - YUsed, YCnt << 1);
|
||||
}
|
||||
|
||||
Address |= (X + Y * (Width >> XUsed)) << (XUsed + YUsed);
|
||||
|
||||
return Address;
|
||||
}
|
||||
}
|
||||
}
|
34
Ryujinx.Graphics/Gpu/Texture.cs
Normal file
34
Ryujinx.Graphics/Gpu/Texture.cs
Normal file
|
@ -0,0 +1,34 @@
|
|||
using Ryujinx.Graphics.Gal;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
struct Texture
|
||||
{
|
||||
public long Position { get; private set; }
|
||||
|
||||
public int Width { get; private set; }
|
||||
public int Height { get; private set; }
|
||||
|
||||
public int BlockHeight { get; private set; }
|
||||
|
||||
public TextureSwizzle Swizzle { get; private set; }
|
||||
|
||||
public GalTextureFormat Format { get; private set; }
|
||||
|
||||
public Texture(
|
||||
long Position,
|
||||
int Width,
|
||||
int Height,
|
||||
int BlockHeight,
|
||||
TextureSwizzle Swizzle,
|
||||
GalTextureFormat Format)
|
||||
{
|
||||
this.Position = Position;
|
||||
this.Width = Width;
|
||||
this.Height = Height;
|
||||
this.BlockHeight = BlockHeight;
|
||||
this.Swizzle = Swizzle;
|
||||
this.Format = Format;
|
||||
}
|
||||
}
|
||||
}
|
83
Ryujinx.Graphics/Gpu/TextureFactory.cs
Normal file
83
Ryujinx.Graphics/Gpu/TextureFactory.cs
Normal file
|
@ -0,0 +1,83 @@
|
|||
using ChocolArm64.Memory;
|
||||
using Ryujinx.Graphics.Gal;
|
||||
using System;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
static class TextureFactory
|
||||
{
|
||||
public static GalTexture MakeTexture(NsGpu Gpu, AMemory Memory, long TicPosition)
|
||||
{
|
||||
int[] Tic = ReadWords(Memory, TicPosition, 8);
|
||||
|
||||
GalTextureFormat Format = (GalTextureFormat)(Tic[0] & 0x7f);
|
||||
|
||||
long TextureAddress = (uint)Tic[1];
|
||||
|
||||
TextureAddress |= (long)((ushort)Tic[2]) << 32;
|
||||
|
||||
TextureAddress = Gpu.GetCpuAddr(TextureAddress);
|
||||
|
||||
TextureSwizzle Swizzle = (TextureSwizzle)((Tic[2] >> 21) & 7);
|
||||
|
||||
int BlockHeightLog2 = (Tic[3] >> 3) & 7;
|
||||
|
||||
int BlockHeight = 1 << BlockHeightLog2;
|
||||
|
||||
int Width = (Tic[4] & 0xffff) + 1;
|
||||
int Height = (Tic[5] & 0xffff) + 1;
|
||||
|
||||
Texture Texture = new Texture(
|
||||
TextureAddress,
|
||||
Width,
|
||||
Height,
|
||||
BlockHeight,
|
||||
Swizzle,
|
||||
Format);
|
||||
|
||||
byte[] Data = TextureReader.Read(Memory, Texture);
|
||||
|
||||
return new GalTexture(Data, Width, Height, Format);
|
||||
}
|
||||
|
||||
public static GalTextureSampler MakeSampler(NsGpu Gpu, AMemory Memory, long TscPosition)
|
||||
{
|
||||
int[] Tsc = ReadWords(Memory, TscPosition, 8);
|
||||
|
||||
GalTextureWrap AddressU = (GalTextureWrap)((Tsc[0] >> 0) & 7);
|
||||
GalTextureWrap AddressV = (GalTextureWrap)((Tsc[0] >> 3) & 7);
|
||||
GalTextureWrap AddressP = (GalTextureWrap)((Tsc[0] >> 6) & 7);
|
||||
|
||||
GalTextureFilter MagFilter = (GalTextureFilter) ((Tsc[1] >> 0) & 3);
|
||||
GalTextureFilter MinFilter = (GalTextureFilter) ((Tsc[1] >> 4) & 3);
|
||||
GalTextureMipFilter MipFilter = (GalTextureMipFilter)((Tsc[1] >> 6) & 3);
|
||||
|
||||
GalColorF BorderColor = new GalColorF(
|
||||
BitConverter.Int32BitsToSingle(Tsc[4]),
|
||||
BitConverter.Int32BitsToSingle(Tsc[5]),
|
||||
BitConverter.Int32BitsToSingle(Tsc[6]),
|
||||
BitConverter.Int32BitsToSingle(Tsc[7]));
|
||||
|
||||
return new GalTextureSampler(
|
||||
AddressU,
|
||||
AddressV,
|
||||
AddressP,
|
||||
MinFilter,
|
||||
MagFilter,
|
||||
MipFilter,
|
||||
BorderColor);
|
||||
}
|
||||
|
||||
private static int[] ReadWords(AMemory Memory, long Position, int Count)
|
||||
{
|
||||
int[] Words = new int[Count];
|
||||
|
||||
for (int Index = 0; Index < Count; Index++, Position += 4)
|
||||
{
|
||||
Words[Index] = Memory.ReadInt32(Position);
|
||||
}
|
||||
|
||||
return Words;
|
||||
}
|
||||
}
|
||||
}
|
127
Ryujinx.Graphics/Gpu/TextureReader.cs
Normal file
127
Ryujinx.Graphics/Gpu/TextureReader.cs
Normal file
|
@ -0,0 +1,127 @@
|
|||
using ChocolArm64.Memory;
|
||||
using Ryujinx.Graphics.Gal;
|
||||
using System;
|
||||
|
||||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
static class TextureReader
|
||||
{
|
||||
public static byte[] Read(AMemory Memory, Texture Texture)
|
||||
{
|
||||
switch (Texture.Format)
|
||||
{
|
||||
case GalTextureFormat.A8B8G8R8: return Read4Bpp (Memory, Texture);
|
||||
case GalTextureFormat.BC1: return Read8Bpt4x4 (Memory, Texture);
|
||||
case GalTextureFormat.BC2: return Read16Bpt4x4(Memory, Texture);
|
||||
case GalTextureFormat.BC3: return Read16Bpt4x4(Memory, Texture);
|
||||
}
|
||||
|
||||
throw new NotImplementedException(Texture.Format.ToString());
|
||||
}
|
||||
|
||||
private unsafe static byte[] Read4Bpp(AMemory Memory, Texture Texture)
|
||||
{
|
||||
int Width = Texture.Width;
|
||||
int Height = Texture.Height;
|
||||
|
||||
byte[] Output = new byte[Width * Height * 4];
|
||||
|
||||
ISwizzle Swizzle = GetSwizzle(Texture.Swizzle, Width, 4, Texture.BlockHeight);
|
||||
|
||||
fixed (byte* BuffPtr = Output)
|
||||
{
|
||||
long OutOffs = 0;
|
||||
|
||||
for (int Y = 0; Y < Height; Y++)
|
||||
for (int X = 0; X < Width; X++)
|
||||
{
|
||||
long Offset = (uint)Swizzle.GetSwizzleOffset(X, Y);
|
||||
|
||||
int Pixel = Memory.ReadInt32Unchecked(Texture.Position + Offset);
|
||||
|
||||
*(int*)(BuffPtr + OutOffs) = Pixel;
|
||||
|
||||
OutOffs += 4;
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
private unsafe static byte[] Read8Bpt4x4(AMemory Memory, Texture Texture)
|
||||
{
|
||||
int Width = (Texture.Width + 3) / 4;
|
||||
int Height = (Texture.Height + 3) / 4;
|
||||
|
||||
byte[] Output = new byte[Width * Height * 8];
|
||||
|
||||
ISwizzle Swizzle = GetSwizzle(Texture.Swizzle, Width, 8, Texture.BlockHeight);
|
||||
|
||||
fixed (byte* BuffPtr = Output)
|
||||
{
|
||||
long OutOffs = 0;
|
||||
|
||||
for (int Y = 0; Y < Height; Y++)
|
||||
for (int X = 0; X < Width; X++)
|
||||
{
|
||||
long Offset = (uint)Swizzle.GetSwizzleOffset(X, Y);
|
||||
|
||||
long Tile = Memory.ReadInt64Unchecked(Texture.Position + Offset);
|
||||
|
||||
*(long*)(BuffPtr + OutOffs) = Tile;
|
||||
|
||||
OutOffs += 8;
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
private unsafe static byte[] Read16Bpt4x4(AMemory Memory, Texture Texture)
|
||||
{
|
||||
int Width = (Texture.Width + 3) / 4;
|
||||
int Height = (Texture.Height + 3) / 4;
|
||||
|
||||
byte[] Output = new byte[Width * Height * 16];
|
||||
|
||||
ISwizzle Swizzle = GetSwizzle(Texture.Swizzle, Width, 16, Texture.BlockHeight);
|
||||
|
||||
fixed (byte* BuffPtr = Output)
|
||||
{
|
||||
long OutOffs = 0;
|
||||
|
||||
for (int Y = 0; Y < Height; Y++)
|
||||
for (int X = 0; X < Width; X++)
|
||||
{
|
||||
long Offset = (uint)Swizzle.GetSwizzleOffset(X, Y);
|
||||
|
||||
long Tile0 = Memory.ReadInt64Unchecked(Texture.Position + Offset + 0);
|
||||
long Tile1 = Memory.ReadInt64Unchecked(Texture.Position + Offset + 8);
|
||||
|
||||
*(long*)(BuffPtr + OutOffs + 0) = Tile0;
|
||||
*(long*)(BuffPtr + OutOffs + 8) = Tile1;
|
||||
|
||||
OutOffs += 16;
|
||||
}
|
||||
}
|
||||
|
||||
return Output;
|
||||
}
|
||||
|
||||
private static ISwizzle GetSwizzle(TextureSwizzle Swizzle, int Width, int Bpp, int BlockHeight)
|
||||
{
|
||||
switch (Swizzle)
|
||||
{
|
||||
case TextureSwizzle.Pitch:
|
||||
case TextureSwizzle.PitchColorKey:
|
||||
return new LinearSwizzle(Width, Bpp);
|
||||
|
||||
case TextureSwizzle.BlockLinear:
|
||||
case TextureSwizzle.BlockLinearColorKey:
|
||||
return new BlockLinearSwizzle(Width, Bpp, BlockHeight);
|
||||
}
|
||||
|
||||
throw new NotImplementedException(Swizzle.ToString());
|
||||
}
|
||||
}
|
||||
}
|
11
Ryujinx.Graphics/Gpu/TextureSwizzle.cs
Normal file
11
Ryujinx.Graphics/Gpu/TextureSwizzle.cs
Normal file
|
@ -0,0 +1,11 @@
|
|||
namespace Ryujinx.Graphics.Gpu
|
||||
{
|
||||
enum TextureSwizzle
|
||||
{
|
||||
_1dBuffer = 0,
|
||||
PitchColorKey = 1,
|
||||
Pitch = 2,
|
||||
BlockLinear = 3,
|
||||
BlockLinearColorKey = 4
|
||||
}
|
||||
}
|
Loading…
Add table
Add a link
Reference in a new issue