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
2019-04-18 01:57:08 +02:00
|
|
|
using Ryujinx.Graphics.Shader.IntermediateRepresentation;
|
|
|
|
using Ryujinx.Graphics.Shader.Translation;
|
|
|
|
|
|
|
|
using static Ryujinx.Graphics.Shader.IntermediateRepresentation.OperandHelper;
|
|
|
|
|
|
|
|
namespace Ryujinx.Graphics.Shader.Instructions
|
|
|
|
{
|
|
|
|
static class Lop3Expression
|
|
|
|
{
|
|
|
|
public static Operand GetFromTruthTable(
|
|
|
|
EmitterContext context,
|
|
|
|
Operand srcA,
|
|
|
|
Operand srcB,
|
|
|
|
Operand srcC,
|
|
|
|
int imm)
|
|
|
|
{
|
|
|
|
Operand expr = null;
|
|
|
|
|
2019-07-02 04:39:22 +02:00
|
|
|
// Handle some simple cases, or cases where
|
|
|
|
// the KMap would yield poor results (like XORs).
|
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
2019-04-18 01:57:08 +02:00
|
|
|
if (imm == 0x96 || imm == 0x69)
|
|
|
|
{
|
2019-07-02 04:39:22 +02:00
|
|
|
// XOR (0x96) and XNOR (0x69).
|
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
2019-04-18 01:57:08 +02:00
|
|
|
if (imm == 0x69)
|
|
|
|
{
|
|
|
|
srcA = context.BitwiseNot(srcA);
|
|
|
|
}
|
|
|
|
|
|
|
|
expr = context.BitwiseExclusiveOr(srcA, srcB);
|
|
|
|
expr = context.BitwiseExclusiveOr(expr, srcC);
|
|
|
|
|
|
|
|
return expr;
|
|
|
|
}
|
|
|
|
else if (imm == 0)
|
|
|
|
{
|
2019-07-02 04:39:22 +02:00
|
|
|
// Always false.
|
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
2019-04-18 01:57:08 +02:00
|
|
|
return Const(IrConsts.False);
|
|
|
|
}
|
|
|
|
else if (imm == 0xff)
|
|
|
|
{
|
2019-07-02 04:39:22 +02:00
|
|
|
// Always true.
|
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
2019-04-18 01:57:08 +02:00
|
|
|
return Const(IrConsts.True);
|
|
|
|
}
|
|
|
|
|
|
|
|
int map;
|
|
|
|
|
2019-07-02 04:39:22 +02:00
|
|
|
// Encode into gray code.
|
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
2019-04-18 01:57:08 +02:00
|
|
|
map = ((imm >> 0) & 1) << 0;
|
|
|
|
map |= ((imm >> 1) & 1) << 4;
|
|
|
|
map |= ((imm >> 2) & 1) << 1;
|
|
|
|
map |= ((imm >> 3) & 1) << 5;
|
|
|
|
map |= ((imm >> 4) & 1) << 3;
|
|
|
|
map |= ((imm >> 5) & 1) << 7;
|
|
|
|
map |= ((imm >> 6) & 1) << 2;
|
|
|
|
map |= ((imm >> 7) & 1) << 6;
|
|
|
|
|
2019-07-02 04:39:22 +02:00
|
|
|
// Solve KMap, get sum of products.
|
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
2019-04-18 01:57:08 +02:00
|
|
|
int visited = 0;
|
|
|
|
|
|
|
|
for (int index = 0; index < 8 && visited != 0xff; index++)
|
|
|
|
{
|
|
|
|
if ((map & (1 << index)) == 0)
|
|
|
|
{
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
int mask = 0;
|
|
|
|
|
|
|
|
for (int mSize = 4; mSize != 0; mSize >>= 1)
|
|
|
|
{
|
|
|
|
mask = RotateLeft4((1 << mSize) - 1, index & 3) << (index & 4);
|
|
|
|
|
|
|
|
if ((map & mask) == mask)
|
|
|
|
{
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-07-02 04:39:22 +02:00
|
|
|
// The mask should wrap, if we are on the high row, shift to low etc.
|
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
2019-04-18 01:57:08 +02:00
|
|
|
int mask2 = (index & 4) != 0 ? mask >> 4 : mask << 4;
|
|
|
|
|
|
|
|
if ((map & mask2) == mask2)
|
|
|
|
{
|
|
|
|
mask |= mask2;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((mask & visited) == mask)
|
|
|
|
{
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool notA = (mask & 0x33) != 0;
|
|
|
|
bool notB = (mask & 0x99) != 0;
|
|
|
|
bool notC = (mask & 0x0f) != 0;
|
|
|
|
|
|
|
|
bool aChanges = (mask & 0xcc) != 0 && notA;
|
|
|
|
bool bChanges = (mask & 0x66) != 0 && notB;
|
|
|
|
bool cChanges = (mask & 0xf0) != 0 && notC;
|
|
|
|
|
|
|
|
Operand localExpr = null;
|
|
|
|
|
|
|
|
void And(Operand source)
|
|
|
|
{
|
|
|
|
if (localExpr != null)
|
|
|
|
{
|
|
|
|
localExpr = context.BitwiseAnd(localExpr, source);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
localExpr = source;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!aChanges)
|
|
|
|
{
|
|
|
|
And(context.BitwiseNot(srcA, notA));
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!bChanges)
|
|
|
|
{
|
|
|
|
And(context.BitwiseNot(srcB, notB));
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!cChanges)
|
|
|
|
{
|
|
|
|
And(context.BitwiseNot(srcC, notC));
|
|
|
|
}
|
|
|
|
|
|
|
|
if (expr != null)
|
|
|
|
{
|
|
|
|
expr = context.BitwiseOr(expr, localExpr);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
expr = localExpr;
|
|
|
|
}
|
|
|
|
|
|
|
|
visited |= mask;
|
|
|
|
}
|
|
|
|
|
|
|
|
return expr;
|
|
|
|
}
|
|
|
|
|
|
|
|
private static int RotateLeft4(int value, int shift)
|
|
|
|
{
|
|
|
|
return ((value << shift) | (value >> (4 - shift))) & 0xf;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|