Ryujinx/Ryujinx.Graphics/Shader/CodeGen/Glsl/Instructions/InstGenMemory.cs

244 lines
7.7 KiB
C#
Raw Normal View History

New shader translator implementation (#654) * Start implementing a new shader translator * Fix shift instructions and a typo * Small refactoring on StructuredProgram, move RemovePhis method to a separate class * Initial geometry shader support * Implement TLD4 * Fix -- There's no negation on FMUL32I * Add constant folding and algebraic simplification optimizations, nits * Some leftovers from constant folding * Avoid cast for constant assignments * Add a branch elimination pass, and misc small fixes * Remove redundant branches, add expression propagation and other improvements on the code * Small leftovers -- add missing break and continue, remove unused properties, other improvements * Add null check to handle empty block cases on block visitor * Add HADD2 and HMUL2 half float shader instructions * Optimize pack/unpack sequences, some fixes related to half float instructions * Add TXQ, TLD, TLDS and TLD4S shader texture instructions, and some support for bindless textures, some refactoring on codegen * Fix copy paste mistake that caused RZ to be ignored on the AST instruction * Add workaround for conditional exit, and fix half float instruction with constant buffer * Add missing 0.0 source for TLDS.LZ variants * Simplify the switch for TLDS.LZ * Texture instructions related fixes * Implement the HFMA instruction, and some misc. fixes * Enable constant folding on UnpackHalf2x16 instructions * Refactor HFMA to use OpCode* for opcode decoding rather than on the helper methods * Remove the old shader translator * Remove ShaderDeclInfo and other unused things * Add dual vertex shader support * Add ShaderConfig, used to pass shader type and maximum cbuffer size * Move and rename some instruction enums * Move texture instructions into a separate file * Move operand GetExpression and locals management to OperandManager * Optimize opcode decoding using a simple list and binary search * Add missing condition for do-while on goto elimination * Misc. fixes on texture instructions * Simplify TLDS switch * Address PR feedback, and a nit
2019-04-17 23:57:08 +00:00
using Ryujinx.Graphics.Shader.IntermediateRepresentation;
using Ryujinx.Graphics.Shader.StructuredIr;
using static Ryujinx.Graphics.Shader.CodeGen.Glsl.Instructions.InstGenHelper;
using static Ryujinx.Graphics.Shader.StructuredIr.InstructionInfo;
namespace Ryujinx.Graphics.Shader.CodeGen.Glsl.Instructions
{
static class InstGenMemory
{
public static string LoadConstant(CodeGenContext context, AstOperation operation)
{
IAstNode src1 = operation.GetSource(1);
string offsetExpr = GetSoureExpr(context, src1, GetSrcVarType(operation.Inst, 1));
offsetExpr = Enclose(offsetExpr, src1, Instruction.ShiftRightS32, isLhs: true);
return OperandManager.GetConstantBufferName(operation.GetSource(0), offsetExpr, context.Config.Type);
}
public static string TextureSample(CodeGenContext context, AstOperation operation)
{
AstTextureOperation texOp = (AstTextureOperation)operation;
bool isBindless = (texOp.Flags & TextureFlags.Bindless) != 0;
bool isGather = (texOp.Flags & TextureFlags.Gather) != 0;
bool intCoords = (texOp.Flags & TextureFlags.IntCoords) != 0;
bool hasLodBias = (texOp.Flags & TextureFlags.LodBias) != 0;
bool hasLodLevel = (texOp.Flags & TextureFlags.LodLevel) != 0;
bool hasOffset = (texOp.Flags & TextureFlags.Offset) != 0;
bool hasOffsets = (texOp.Flags & TextureFlags.Offsets) != 0;
bool isArray = (texOp.Type & TextureType.Array) != 0;
bool isMultisample = (texOp.Type & TextureType.Multisample) != 0;
bool isShadow = (texOp.Type & TextureType.Shadow) != 0;
string texCall = intCoords ? "texelFetch" : "texture";
if (isGather)
{
texCall += "Gather";
}
else if (hasLodLevel && !intCoords)
{
texCall += "Lod";
}
if (hasOffset)
{
texCall += "Offset";
}
else if (hasOffsets)
{
texCall += "Offsets";
}
string samplerName = OperandManager.GetSamplerName(context.Config.Type, texOp);
texCall += "(" + samplerName;
int coordsCount = texOp.Type.GetCoordsCount();
int pCount = coordsCount;
int arrayIndexElem = -1;
if (isArray)
{
arrayIndexElem = pCount++;
}
// The sampler 1D shadow overload expects a
// dummy value on the middle of the vector, who knows why...
New shader translator implementation (#654) * Start implementing a new shader translator * Fix shift instructions and a typo * Small refactoring on StructuredProgram, move RemovePhis method to a separate class * Initial geometry shader support * Implement TLD4 * Fix -- There's no negation on FMUL32I * Add constant folding and algebraic simplification optimizations, nits * Some leftovers from constant folding * Avoid cast for constant assignments * Add a branch elimination pass, and misc small fixes * Remove redundant branches, add expression propagation and other improvements on the code * Small leftovers -- add missing break and continue, remove unused properties, other improvements * Add null check to handle empty block cases on block visitor * Add HADD2 and HMUL2 half float shader instructions * Optimize pack/unpack sequences, some fixes related to half float instructions * Add TXQ, TLD, TLDS and TLD4S shader texture instructions, and some support for bindless textures, some refactoring on codegen * Fix copy paste mistake that caused RZ to be ignored on the AST instruction * Add workaround for conditional exit, and fix half float instruction with constant buffer * Add missing 0.0 source for TLDS.LZ variants * Simplify the switch for TLDS.LZ * Texture instructions related fixes * Implement the HFMA instruction, and some misc. fixes * Enable constant folding on UnpackHalf2x16 instructions * Refactor HFMA to use OpCode* for opcode decoding rather than on the helper methods * Remove the old shader translator * Remove ShaderDeclInfo and other unused things * Add dual vertex shader support * Add ShaderConfig, used to pass shader type and maximum cbuffer size * Move and rename some instruction enums * Move texture instructions into a separate file * Move operand GetExpression and locals management to OperandManager * Optimize opcode decoding using a simple list and binary search * Add missing condition for do-while on goto elimination * Misc. fixes on texture instructions * Simplify TLDS switch * Address PR feedback, and a nit
2019-04-17 23:57:08 +00:00
bool hasDummy1DShadowElem = texOp.Type == (TextureType.Texture1D | TextureType.Shadow);
if (hasDummy1DShadowElem)
{
pCount++;
}
if (isShadow && !isGather)
{
pCount++;
}
// On textureGather*, the comparison value is
// always specified as an extra argument.
New shader translator implementation (#654) * Start implementing a new shader translator * Fix shift instructions and a typo * Small refactoring on StructuredProgram, move RemovePhis method to a separate class * Initial geometry shader support * Implement TLD4 * Fix -- There's no negation on FMUL32I * Add constant folding and algebraic simplification optimizations, nits * Some leftovers from constant folding * Avoid cast for constant assignments * Add a branch elimination pass, and misc small fixes * Remove redundant branches, add expression propagation and other improvements on the code * Small leftovers -- add missing break and continue, remove unused properties, other improvements * Add null check to handle empty block cases on block visitor * Add HADD2 and HMUL2 half float shader instructions * Optimize pack/unpack sequences, some fixes related to half float instructions * Add TXQ, TLD, TLDS and TLD4S shader texture instructions, and some support for bindless textures, some refactoring on codegen * Fix copy paste mistake that caused RZ to be ignored on the AST instruction * Add workaround for conditional exit, and fix half float instruction with constant buffer * Add missing 0.0 source for TLDS.LZ variants * Simplify the switch for TLDS.LZ * Texture instructions related fixes * Implement the HFMA instruction, and some misc. fixes * Enable constant folding on UnpackHalf2x16 instructions * Refactor HFMA to use OpCode* for opcode decoding rather than on the helper methods * Remove the old shader translator * Remove ShaderDeclInfo and other unused things * Add dual vertex shader support * Add ShaderConfig, used to pass shader type and maximum cbuffer size * Move and rename some instruction enums * Move texture instructions into a separate file * Move operand GetExpression and locals management to OperandManager * Optimize opcode decoding using a simple list and binary search * Add missing condition for do-while on goto elimination * Misc. fixes on texture instructions * Simplify TLDS switch * Address PR feedback, and a nit
2019-04-17 23:57:08 +00:00
bool hasExtraCompareArg = isShadow && isGather;
if (pCount == 5)
{
pCount = 4;
hasExtraCompareArg = true;
}
int srcIndex = isBindless ? 1 : 0;
string Src(VariableType type)
{
return GetSoureExpr(context, texOp.GetSource(srcIndex++), type);
}
void Append(string str)
{
texCall += ", " + str;
}
VariableType coordType = intCoords ? VariableType.S32 : VariableType.F32;
string AssemblePVector(int count)
{
if (count > 1)
{
string[] elems = new string[count];
for (int index = 0; index < count; index++)
{
if (arrayIndexElem == index)
{
elems[index] = Src(VariableType.S32);
if (!intCoords)
{
elems[index] = "float(" + elems[index] + ")";
}
}
else if (index == 1 && hasDummy1DShadowElem)
{
elems[index] = NumberFormatter.FormatFloat(0);
}
else
{
elems[index] = Src(coordType);
}
}
string prefix = intCoords ? "i" : string.Empty;
return prefix + "vec" + count + "(" + string.Join(", ", elems) + ")";
}
else
{
return Src(coordType);
}
}
Append(AssemblePVector(pCount));
if (hasExtraCompareArg)
{
Append(Src(VariableType.F32));
}
if (isMultisample)
{
Append(Src(VariableType.S32));
}
else if (hasLodLevel)
{
Append(Src(coordType));
}
string AssembleOffsetVector(int count)
{
if (count > 1)
{
string[] elems = new string[count];
for (int index = 0; index < count; index++)
{
elems[index] = Src(VariableType.S32);
}
return "ivec" + count + "(" + string.Join(", ", elems) + ")";
}
else
{
return Src(VariableType.S32);
}
}
if (hasOffset)
{
Append(AssembleOffsetVector(coordsCount));
}
else if (hasOffsets)
{
texCall += $", ivec{coordsCount}[4](";
texCall += AssembleOffsetVector(coordsCount) + ", ";
texCall += AssembleOffsetVector(coordsCount) + ", ";
texCall += AssembleOffsetVector(coordsCount) + ", ";
texCall += AssembleOffsetVector(coordsCount) + ")";
}
if (hasLodBias)
{
Append(Src(VariableType.F32));
}
// textureGather* optional extra component index,
// not needed for shadow samplers.
New shader translator implementation (#654) * Start implementing a new shader translator * Fix shift instructions and a typo * Small refactoring on StructuredProgram, move RemovePhis method to a separate class * Initial geometry shader support * Implement TLD4 * Fix -- There's no negation on FMUL32I * Add constant folding and algebraic simplification optimizations, nits * Some leftovers from constant folding * Avoid cast for constant assignments * Add a branch elimination pass, and misc small fixes * Remove redundant branches, add expression propagation and other improvements on the code * Small leftovers -- add missing break and continue, remove unused properties, other improvements * Add null check to handle empty block cases on block visitor * Add HADD2 and HMUL2 half float shader instructions * Optimize pack/unpack sequences, some fixes related to half float instructions * Add TXQ, TLD, TLDS and TLD4S shader texture instructions, and some support for bindless textures, some refactoring on codegen * Fix copy paste mistake that caused RZ to be ignored on the AST instruction * Add workaround for conditional exit, and fix half float instruction with constant buffer * Add missing 0.0 source for TLDS.LZ variants * Simplify the switch for TLDS.LZ * Texture instructions related fixes * Implement the HFMA instruction, and some misc. fixes * Enable constant folding on UnpackHalf2x16 instructions * Refactor HFMA to use OpCode* for opcode decoding rather than on the helper methods * Remove the old shader translator * Remove ShaderDeclInfo and other unused things * Add dual vertex shader support * Add ShaderConfig, used to pass shader type and maximum cbuffer size * Move and rename some instruction enums * Move texture instructions into a separate file * Move operand GetExpression and locals management to OperandManager * Optimize opcode decoding using a simple list and binary search * Add missing condition for do-while on goto elimination * Misc. fixes on texture instructions * Simplify TLDS switch * Address PR feedback, and a nit
2019-04-17 23:57:08 +00:00
if (isGather && !isShadow)
{
Append(Src(VariableType.S32));
}
texCall += ")" + (isGather || !isShadow ? GetMask(texOp.ComponentMask) : "");
return texCall;
}
public static string TextureSize(CodeGenContext context, AstOperation operation)
{
AstTextureOperation texOp = (AstTextureOperation)operation;
bool isBindless = (texOp.Flags & TextureFlags.Bindless) != 0;
string samplerName = OperandManager.GetSamplerName(context.Config.Type, texOp);
IAstNode src0 = operation.GetSource(isBindless ? 1 : 0);
string src0Expr = GetSoureExpr(context, src0, GetSrcVarType(operation.Inst, 0));
return $"textureSize({samplerName}, {src0Expr}){GetMask(texOp.ComponentMask)}";
}
private static string GetMask(int compMask)
{
string mask = ".";
for (int index = 0; index < 4; index++)
{
if ((compMask & (1 << index)) != 0)
{
mask += "rgba".Substring(index, 1);
}
}
return mask;
}
}
}