InfinityX/RyuKen
1
0
Fork 0
forked from Mirror/Ryujinx
Code Pull requests Activity
RyuKen/Ryujinx.Graphics.Shader/StructuredIr/InstructionInfo.cs

216 lines
16 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;
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
using Ryujinx.Graphics.Shader.Translation;
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 System;
namespace Ryujinx.Graphics.Shader.StructuredIr
{
static class InstructionInfo
{
Make structs readonly when applicable (#4002) * Make all structs readonly when applicable. It should reduce amount of needless defensive copies * Make structs with trivial boilerplate equality code record structs * Remove unnecessary readonly modifiers from TextureCreateInfo * Make BitMap structs readonly too
2022-12-05 13:47:39 +00:00
private readonly struct InstInfo
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
{
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
public AggregateType DestType { get; }
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
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
public AggregateType[] SrcTypes { get; }
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
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
public InstInfo(AggregateType destType, params AggregateType[] srcTypes)
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
{
DestType = destType;
SrcTypes = srcTypes;
}
}
private static InstInfo[] _infoTbl;
static InstructionInfo()
{
_infoTbl = new InstInfo[(int)Instruction.Count];
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
// Inst Destination type Source 1 type Source 2 type Source 3 type Source 4 type
Add(Instruction.AtomicAdd, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.AtomicAnd, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.AtomicCompareAndSwap, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32, AggregateType.U32);
Add(Instruction.AtomicMaxS32, AggregateType.S32, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.AtomicMaxU32, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.AtomicMinS32, AggregateType.S32, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.AtomicMinU32, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.AtomicOr, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.AtomicSwap, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.AtomicXor, AggregateType.U32, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.Absolute, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.Add, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.Ballot, AggregateType.U32, AggregateType.Bool);
Add(Instruction.BitCount, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitfieldExtractS32, AggregateType.S32, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitfieldExtractU32, AggregateType.U32, AggregateType.U32, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitfieldInsert, AggregateType.S32, AggregateType.S32, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitfieldReverse, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitwiseAnd, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitwiseExclusiveOr, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitwiseNot, AggregateType.S32, AggregateType.S32);
Add(Instruction.BitwiseOr, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.BranchIfTrue, AggregateType.Void, AggregateType.Bool);
Add(Instruction.BranchIfFalse, AggregateType.Void, AggregateType.Bool);
Add(Instruction.Call, AggregateType.Scalar);
Add(Instruction.Ceiling, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.Clamp, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.ClampU32, AggregateType.U32, AggregateType.U32, AggregateType.U32, AggregateType.U32);
Add(Instruction.CompareEqual, AggregateType.Bool, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.CompareGreater, AggregateType.Bool, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.CompareGreaterOrEqual, AggregateType.Bool, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.CompareGreaterOrEqualU32, AggregateType.Bool, AggregateType.U32, AggregateType.U32);
Add(Instruction.CompareGreaterU32, AggregateType.Bool, AggregateType.U32, AggregateType.U32);
Add(Instruction.CompareLess, AggregateType.Bool, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.CompareLessOrEqual, AggregateType.Bool, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.CompareLessOrEqualU32, AggregateType.Bool, AggregateType.U32, AggregateType.U32);
Add(Instruction.CompareLessU32, AggregateType.Bool, AggregateType.U32, AggregateType.U32);
Add(Instruction.CompareNotEqual, AggregateType.Bool, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.ConditionalSelect, AggregateType.Scalar, AggregateType.Bool, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.ConvertFP32ToFP64, AggregateType.FP64, AggregateType.FP32);
Add(Instruction.ConvertFP64ToFP32, AggregateType.FP32, AggregateType.FP64);
Add(Instruction.ConvertFP32ToS32, AggregateType.S32, AggregateType.FP32);
Add(Instruction.ConvertFP32ToU32, AggregateType.U32, AggregateType.FP32);
Add(Instruction.ConvertFP64ToS32, AggregateType.S32, AggregateType.FP64);
Add(Instruction.ConvertFP64ToU32, AggregateType.U32, AggregateType.FP64);
Add(Instruction.ConvertS32ToFP32, AggregateType.FP32, AggregateType.S32);
Add(Instruction.ConvertS32ToFP64, AggregateType.FP64, AggregateType.S32);
Add(Instruction.ConvertU32ToFP32, AggregateType.FP32, AggregateType.U32);
Add(Instruction.ConvertU32ToFP64, AggregateType.FP64, AggregateType.U32);
Add(Instruction.Cosine, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.Ddx, AggregateType.FP32, AggregateType.FP32);
Add(Instruction.Ddy, AggregateType.FP32, AggregateType.FP32);
Add(Instruction.Divide, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.ExponentB2, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.FindLSB, AggregateType.S32, AggregateType.S32);
Add(Instruction.FindMSBS32, AggregateType.S32, AggregateType.S32);
Add(Instruction.FindMSBU32, AggregateType.S32, AggregateType.U32);
Add(Instruction.Floor, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.FusedMultiplyAdd, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.ImageLoad, AggregateType.FP32);
Add(Instruction.ImageStore, AggregateType.Void);
Add(Instruction.ImageAtomic, AggregateType.S32);
Add(Instruction.IsNan, AggregateType.Bool, AggregateType.Scalar);
Add(Instruction.LoadAttribute, AggregateType.FP32, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.LoadConstant, AggregateType.FP32, AggregateType.S32, AggregateType.S32);
Add(Instruction.LoadGlobal, AggregateType.U32, AggregateType.S32, AggregateType.S32);
Add(Instruction.LoadLocal, AggregateType.U32, AggregateType.S32);
Add(Instruction.LoadShared, AggregateType.U32, AggregateType.S32);
Add(Instruction.LoadStorage, AggregateType.U32, AggregateType.S32, AggregateType.S32);
Add(Instruction.Lod, AggregateType.FP32);
Add(Instruction.LogarithmB2, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.LogicalAnd, AggregateType.Bool, AggregateType.Bool, AggregateType.Bool);
Add(Instruction.LogicalExclusiveOr, AggregateType.Bool, AggregateType.Bool, AggregateType.Bool);
Add(Instruction.LogicalNot, AggregateType.Bool, AggregateType.Bool);
Add(Instruction.LogicalOr, AggregateType.Bool, AggregateType.Bool, AggregateType.Bool);
Add(Instruction.Maximum, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.MaximumU32, AggregateType.U32, AggregateType.U32, AggregateType.U32);
Add(Instruction.Minimum, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.MinimumU32, AggregateType.U32, AggregateType.U32, AggregateType.U32);
Add(Instruction.Multiply, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.MultiplyHighS32, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.MultiplyHighU32, AggregateType.U32, AggregateType.U32, AggregateType.U32);
Add(Instruction.Negate, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.PackDouble2x32, AggregateType.FP64, AggregateType.U32, AggregateType.U32);
Add(Instruction.PackHalf2x16, AggregateType.U32, AggregateType.FP32, AggregateType.FP32);
Add(Instruction.ReciprocalSquareRoot, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.Round, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.ShiftLeft, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.ShiftRightS32, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.ShiftRightU32, AggregateType.U32, AggregateType.U32, AggregateType.S32);
Add(Instruction.Shuffle, AggregateType.FP32, AggregateType.FP32, AggregateType.U32, AggregateType.U32, AggregateType.Bool);
Add(Instruction.ShuffleDown, AggregateType.FP32, AggregateType.FP32, AggregateType.U32, AggregateType.U32, AggregateType.Bool);
Add(Instruction.ShuffleUp, AggregateType.FP32, AggregateType.FP32, AggregateType.U32, AggregateType.U32, AggregateType.Bool);
Add(Instruction.ShuffleXor, AggregateType.FP32, AggregateType.FP32, AggregateType.U32, AggregateType.U32, AggregateType.Bool);
Add(Instruction.Sine, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.SquareRoot, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.StoreAttribute, AggregateType.Void, AggregateType.S32, AggregateType.S32, AggregateType.FP32);
Add(Instruction.StoreGlobal, AggregateType.Void, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.StoreLocal, AggregateType.Void, AggregateType.S32, AggregateType.U32);
Add(Instruction.StoreShared, AggregateType.Void, AggregateType.S32, AggregateType.U32);
Add(Instruction.StoreShared16, AggregateType.Void, AggregateType.S32, AggregateType.U32);
Add(Instruction.StoreShared8, AggregateType.Void, AggregateType.S32, AggregateType.U32);
Add(Instruction.StoreStorage, AggregateType.Void, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.StoreStorage16, AggregateType.Void, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.StoreStorage8, AggregateType.Void, AggregateType.S32, AggregateType.S32, AggregateType.U32);
Add(Instruction.Subtract, AggregateType.Scalar, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.SwizzleAdd, AggregateType.FP32, AggregateType.FP32, AggregateType.FP32, AggregateType.S32);
Add(Instruction.TextureSample, AggregateType.FP32);
Add(Instruction.TextureSize, AggregateType.S32, AggregateType.S32, AggregateType.S32);
Add(Instruction.Truncate, AggregateType.Scalar, AggregateType.Scalar);
Add(Instruction.UnpackDouble2x32, AggregateType.U32, AggregateType.FP64);
Add(Instruction.UnpackHalf2x16, AggregateType.FP32, AggregateType.U32);
Add(Instruction.VectorExtract, AggregateType.Scalar, AggregateType.Vector4, AggregateType.S32);
Add(Instruction.VoteAll, AggregateType.Bool, AggregateType.Bool);
Add(Instruction.VoteAllEqual, AggregateType.Bool, AggregateType.Bool);
Add(Instruction.VoteAny, AggregateType.Bool, AggregateType.Bool);
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
}
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
private static void Add(Instruction inst, AggregateType destType, params AggregateType[] srcTypes)
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
{
_infoTbl[(int)inst] = new InstInfo(destType, srcTypes);
}
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
public static AggregateType GetDestVarType(Instruction inst)
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
{
return GetFinalVarType(_infoTbl[(int)(inst & Instruction.Mask)].DestType, inst);
}
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
public static AggregateType GetSrcVarType(Instruction inst, int index)
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 per-source type memory change tracking, simplified state change tracking, other fixes
2019-10-26 17:50:52 +00:00
// TODO: Return correct type depending on source index,
// that can improve the decompiler output.
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
if (inst == Instruction.ImageLoad ||
inst == Instruction.ImageStore ||
Implement Shader Instructions SUATOM and SURED (#2090) * Initial Implementation * Further improvements (no support for float/64-bit types) * Merge atomic and reduce instructions, add missing format switch * Fix rebase issues. * Not used. * Whoops. Fixed. * Partial implementation of inc/dec, cleanup and TODOs * Remove testing path * Address Feedback
2021-08-31 05:51:57 +00:00
inst == Instruction.ImageAtomic ||
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
inst == Instruction.Lod ||
Support non-constant texture offsets on non-NVIDIA gpus
2019-12-11 06:54:18 +00:00
inst == Instruction.TextureSample)
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
{
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
return AggregateType.FP32;
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
}
Implement CAL and RET shader instructions (#1618) * Add support for CAL and RET shader instructions * Remove unused stuff * Fix a bug that could cause the wrong values to be passed to a function * Avoid repopulating function id dictionary every time * PR feedback * Fix vertex shader A/B merge
2020-10-25 20:00:44 +00:00
else if (inst == Instruction.Call)
{
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
return AggregateType.S32;
Implement CAL and RET shader instructions (#1618) * Add support for CAL and RET shader instructions * Remove unused stuff * Fix a bug that could cause the wrong values to be passed to a function * Avoid repopulating function id dictionary every time * PR feedback * Fix vertex shader A/B merge
2020-10-25 20:00:44 +00:00
}
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
return GetFinalVarType(_infoTbl[(int)(inst & Instruction.Mask)].SrcTypes[index], inst);
}
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
private static AggregateType GetFinalVarType(AggregateType type, Instruction inst)
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
{
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
if (type == AggregateType.Scalar)
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
{
Initial support for double precision shader instructions. (#963) * Implement DADD, DFMA and DMUL shader instructions * Rename FP to FP32 * Correct double immediate * Classic mistake
2020-03-03 14:02:08 +00:00
if ((inst & Instruction.FP32) != 0)
{
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
return AggregateType.FP32;
Initial support for double precision shader instructions. (#963) * Implement DADD, DFMA and DMUL shader instructions * Rename FP to FP32 * Correct double immediate * Classic mistake
2020-03-03 14:02:08 +00:00
}
else if ((inst & Instruction.FP64) != 0)
{
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
return AggregateType.FP64;
Initial support for double precision shader instructions. (#963) * Implement DADD, DFMA and DMUL shader instructions * Rename FP to FP32 * Correct double immediate * Classic mistake
2020-03-03 14:02:08 +00:00
}
else
{
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
return AggregateType.S32;
Initial support for double precision shader instructions. (#963) * Implement DADD, DFMA and DMUL shader instructions * Rename FP to FP32 * Correct double immediate * Classic mistake
2020-03-03 14:02:08 +00:00
}
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
}
Use vector outputs for texture operations (#3939) * Change AggregateType to include vector type counts * Replace VariableType uses with AggregateType and delete VariableType * Support new local vector types on SPIR-V and GLSL * Start using vector outputs for texture operations * Use vectors on more texture operations * Use vector output for ImageLoad operations * Replace all uses of single destination texture constructors with multi destination ones * Update textureGatherOffsets replacement to split vector operations * Shader cache version bump Co-authored-by: Ac_K <Acoustik666@gmail.com>
2022-12-29 15:09:34 +00:00
else if (type == AggregateType.Void)
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
{
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;
}
}
}
Copy permalink