ryujinx/Ryujinx
1
0
Fork 0
mirror of https://github.com/Ryujinx/Ryujinx.git synced 2024-11-09 08:18:34 +00:00
Code Issues Packages Projects Releases Wiki Activity
Ryujinx/Ryujinx.Graphics.Shader/StructuredIr/InstructionInfo.cs

150 lines
9.7 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 System;
namespace Ryujinx.Graphics.Shader.StructuredIr
{
static class InstructionInfo
{
private struct InstInfo
{
public VariableType DestType { get; }
public VariableType[] SrcTypes { get; }
public InstInfo(VariableType destType, params VariableType[] srcTypes)
{
DestType = destType;
SrcTypes = srcTypes;
}
}
private static InstInfo[] _infoTbl;
static InstructionInfo()
{
_infoTbl = new InstInfo[(int)Instruction.Count];
Initial work
2019-10-13 06:02:07 +00:00
// Inst Destination type Source 1 type Source 2 type Source 3 type Source 4 type
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
Add(Instruction.Absolute, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.Add, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.BitfieldExtractS32, VariableType.S32, VariableType.S32, VariableType.S32, VariableType.S32);
Add(Instruction.BitfieldExtractU32, VariableType.U32, VariableType.U32, VariableType.S32, VariableType.S32);
Add(Instruction.BitfieldInsert, VariableType.Int, VariableType.Int, VariableType.Int, VariableType.S32, VariableType.S32);
Add(Instruction.BitfieldReverse, VariableType.Int, VariableType.Int);
Add(Instruction.BitwiseAnd, VariableType.Int, VariableType.Int, VariableType.Int);
Add(Instruction.BitwiseExclusiveOr, VariableType.Int, VariableType.Int, VariableType.Int);
Add(Instruction.BitwiseNot, VariableType.Int, VariableType.Int);
Add(Instruction.BitwiseOr, VariableType.Int, VariableType.Int, VariableType.Int);
Add(Instruction.BranchIfTrue, VariableType.None, VariableType.Bool);
Add(Instruction.BranchIfFalse, VariableType.None, VariableType.Bool);
Add(Instruction.Ceiling, VariableType.F32, VariableType.F32, VariableType.F32);
Add(Instruction.Clamp, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.ClampU32, VariableType.U32, VariableType.U32, VariableType.U32, VariableType.U32);
Add(Instruction.CompareEqual, VariableType.Bool, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.CompareGreater, VariableType.Bool, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.CompareGreaterOrEqual, VariableType.Bool, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.CompareGreaterOrEqualU32, VariableType.Bool, VariableType.U32, VariableType.U32);
Add(Instruction.CompareGreaterU32, VariableType.Bool, VariableType.U32, VariableType.U32);
Add(Instruction.CompareLess, VariableType.Bool, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.CompareLessOrEqual, VariableType.Bool, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.CompareLessOrEqualU32, VariableType.Bool, VariableType.U32, VariableType.U32);
Add(Instruction.CompareLessU32, VariableType.Bool, VariableType.U32, VariableType.U32);
Add(Instruction.CompareNotEqual, VariableType.Bool, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.ConditionalSelect, VariableType.Scalar, VariableType.Bool, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.ConvertFPToS32, VariableType.S32, VariableType.F32);
Add(Instruction.ConvertS32ToFP, VariableType.F32, VariableType.S32);
Add(Instruction.ConvertU32ToFP, VariableType.F32, VariableType.U32);
Add(Instruction.Cosine, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.Divide, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.ExponentB2, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.Floor, VariableType.F32, VariableType.F32);
Add(Instruction.FusedMultiplyAdd, VariableType.F32, VariableType.F32, VariableType.F32, VariableType.F32);
Add(Instruction.IsNan, VariableType.Bool, VariableType.F32);
Initial work
2019-10-13 06:02:07 +00:00
Add(Instruction.LoadAttribute, VariableType.F32, VariableType.S32, VariableType.S32);
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
Add(Instruction.LoadConstant, VariableType.F32, VariableType.S32, VariableType.S32);
Initial work
2019-10-13 06:02:07 +00:00
Add(Instruction.LoadGlobal, VariableType.F32, VariableType.S32, VariableType.S32);
Add(Instruction.LoadLocal, VariableType.F32, VariableType.S32);
Add(Instruction.LoadStorage, VariableType.F32, VariableType.S32, VariableType.S32);
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
Add(Instruction.LogarithmB2, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.LogicalAnd, VariableType.Bool, VariableType.Bool, VariableType.Bool);
Add(Instruction.LogicalExclusiveOr, VariableType.Bool, VariableType.Bool, VariableType.Bool);
Add(Instruction.LogicalNot, VariableType.Bool, VariableType.Bool);
Add(Instruction.LogicalOr, VariableType.Bool, VariableType.Bool, VariableType.Bool);
Add(Instruction.ShiftLeft, VariableType.Int, VariableType.Int, VariableType.Int);
Add(Instruction.ShiftRightS32, VariableType.S32, VariableType.S32, VariableType.Int);
Add(Instruction.ShiftRightU32, VariableType.U32, VariableType.U32, VariableType.Int);
Add(Instruction.Maximum, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.MaximumU32, VariableType.U32, VariableType.U32, VariableType.U32);
Add(Instruction.Minimum, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.MinimumU32, VariableType.U32, VariableType.U32, VariableType.U32);
Add(Instruction.Multiply, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.Negate, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.PackHalf2x16, VariableType.U32, VariableType.F32, VariableType.F32);
Add(Instruction.ReciprocalSquareRoot, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.Sine, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.SquareRoot, VariableType.Scalar, VariableType.Scalar);
Initial work
2019-10-13 06:02:07 +00:00
Add(Instruction.StoreGlobal, VariableType.None, VariableType.S32, VariableType.S32, VariableType.F32);
Add(Instruction.StoreLocal, VariableType.None, VariableType.S32, VariableType.F32);
Add(Instruction.StoreStorage, VariableType.None, VariableType.S32, VariableType.S32, VariableType.F32);
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
Add(Instruction.Subtract, VariableType.Scalar, VariableType.Scalar, VariableType.Scalar);
Add(Instruction.TextureSample, VariableType.F32);
Add(Instruction.TextureSize, VariableType.S32, VariableType.S32, VariableType.S32);
Add(Instruction.Truncate, VariableType.F32, VariableType.F32);
Add(Instruction.UnpackHalf2x16, VariableType.F32, VariableType.U32);
}
private static void Add(Instruction inst, VariableType destType, params VariableType[] srcTypes)
{
_infoTbl[(int)inst] = new InstInfo(destType, srcTypes);
}
public static VariableType GetDestVarType(Instruction inst)
{
return GetFinalVarType(_infoTbl[(int)(inst & Instruction.Mask)].DestType, inst);
}
public static VariableType GetSrcVarType(Instruction inst, int index)
{
if (inst == Instruction.TextureSample)
{
return VariableType.F32;
}
return GetFinalVarType(_infoTbl[(int)(inst & Instruction.Mask)].SrcTypes[index], inst);
}
private static VariableType GetFinalVarType(VariableType type, Instruction inst)
{
if (type == VariableType.Scalar)
{
return (inst & Instruction.FP) != 0
? VariableType.F32
: VariableType.S32;
}
else if (type == VariableType.Int)
{
return VariableType.S32;
}
else if (type == VariableType.None)
{
throw new ArgumentException($"Invalid operand for instruction \"{inst}\".");
}
return type;
}
public static bool IsUnary(Instruction inst)
{
if (inst == Instruction.Copy)
{
return true;
}
else if (inst == Instruction.TextureSample)
{
return false;
}
return _infoTbl[(int)(inst & Instruction.Mask)].SrcTypes.Length == 1;
}
}
}
Reference in a new issue Copy permalink