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 System.Collections.Generic;
|
|
|
|
|
|
|
|
namespace Ryujinx.Graphics.Shader.Translation
|
|
|
|
{
|
2020-10-25 21:00:44 +01:00
|
|
|
class ControlFlowGraph
|
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
|
|
|
{
|
2020-10-25 21:00:44 +01:00
|
|
|
public BasicBlock[] Blocks { get; }
|
|
|
|
public BasicBlock[] PostOrderBlocks { get; }
|
|
|
|
public int[] PostOrderMap { get; }
|
|
|
|
|
|
|
|
public ControlFlowGraph(BasicBlock[] blocks)
|
|
|
|
{
|
|
|
|
Blocks = blocks;
|
|
|
|
|
2023-06-28 08:59:13 +02:00
|
|
|
HashSet<BasicBlock> visited = new();
|
2020-10-25 21:00:44 +01:00
|
|
|
|
2023-06-28 08:59:13 +02:00
|
|
|
Stack<BasicBlock> blockStack = new();
|
2020-10-25 21:00:44 +01:00
|
|
|
|
2023-06-28 08:59:13 +02:00
|
|
|
List<BasicBlock> postOrderBlocks = new(blocks.Length);
|
2020-10-25 21:00:44 +01:00
|
|
|
|
|
|
|
PostOrderMap = new int[blocks.Length];
|
|
|
|
|
|
|
|
visited.Add(blocks[0]);
|
|
|
|
|
|
|
|
blockStack.Push(blocks[0]);
|
|
|
|
|
|
|
|
while (blockStack.TryPop(out BasicBlock block))
|
|
|
|
{
|
|
|
|
if (block.Next != null && visited.Add(block.Next))
|
|
|
|
{
|
|
|
|
blockStack.Push(block);
|
|
|
|
blockStack.Push(block.Next);
|
|
|
|
}
|
|
|
|
else if (block.Branch != null && visited.Add(block.Branch))
|
|
|
|
{
|
|
|
|
blockStack.Push(block);
|
|
|
|
blockStack.Push(block.Branch);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
PostOrderMap[block.Index] = postOrderBlocks.Count;
|
|
|
|
|
|
|
|
postOrderBlocks.Add(block);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
PostOrderBlocks = postOrderBlocks.ToArray();
|
|
|
|
}
|
|
|
|
|
|
|
|
public static ControlFlowGraph Create(Operation[] operations)
|
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
|
|
|
{
|
2023-06-28 08:59:13 +02:00
|
|
|
Dictionary<Operand, BasicBlock> labels = new();
|
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
|
|
|
|
2023-06-28 08:59:13 +02:00
|
|
|
List<BasicBlock> blocks = new();
|
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
|
|
|
|
|
|
|
BasicBlock currentBlock = null;
|
|
|
|
|
|
|
|
void NextBlock(BasicBlock nextBlock)
|
|
|
|
{
|
|
|
|
if (currentBlock != null && !EndsWithUnconditionalInst(currentBlock.GetLastOp()))
|
|
|
|
{
|
|
|
|
currentBlock.Next = nextBlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
currentBlock = nextBlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
void NewNextBlock()
|
|
|
|
{
|
2023-06-28 08:59:13 +02:00
|
|
|
BasicBlock block = new(blocks.Count);
|
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
|
|
|
|
|
|
|
blocks.Add(block);
|
|
|
|
|
|
|
|
NextBlock(block);
|
|
|
|
}
|
|
|
|
|
|
|
|
bool needsNewBlock = true;
|
|
|
|
|
|
|
|
for (int index = 0; index < operations.Length; index++)
|
|
|
|
{
|
|
|
|
Operation operation = operations[index];
|
|
|
|
|
|
|
|
if (operation.Inst == Instruction.MarkLabel)
|
|
|
|
{
|
|
|
|
Operand label = operation.Dest;
|
|
|
|
|
|
|
|
if (labels.TryGetValue(label, out BasicBlock nextBlock))
|
|
|
|
{
|
|
|
|
nextBlock.Index = blocks.Count;
|
|
|
|
|
|
|
|
blocks.Add(nextBlock);
|
|
|
|
|
|
|
|
NextBlock(nextBlock);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
NewNextBlock();
|
|
|
|
|
|
|
|
labels.Add(label, currentBlock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
if (needsNewBlock)
|
|
|
|
{
|
|
|
|
NewNextBlock();
|
|
|
|
}
|
|
|
|
|
|
|
|
currentBlock.Operations.AddLast(operation);
|
|
|
|
}
|
|
|
|
|
2023-06-28 08:59:13 +02:00
|
|
|
needsNewBlock = operation.Inst == Instruction.Branch ||
|
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
|
|
|
operation.Inst == Instruction.BranchIfTrue ||
|
|
|
|
operation.Inst == Instruction.BranchIfFalse;
|
|
|
|
|
|
|
|
if (needsNewBlock)
|
|
|
|
{
|
|
|
|
Operand label = operation.Dest;
|
|
|
|
|
|
|
|
if (!labels.TryGetValue(label, out BasicBlock branchBlock))
|
|
|
|
{
|
|
|
|
branchBlock = new BasicBlock();
|
|
|
|
|
|
|
|
labels.Add(label, branchBlock);
|
|
|
|
}
|
|
|
|
|
|
|
|
currentBlock.Branch = branchBlock;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-10-12 22:35:31 +02:00
|
|
|
// Remove unreachable blocks.
|
|
|
|
bool hasUnreachable;
|
|
|
|
|
|
|
|
do
|
|
|
|
{
|
|
|
|
hasUnreachable = false;
|
|
|
|
|
|
|
|
for (int blkIndex = 1; blkIndex < blocks.Count; blkIndex++)
|
|
|
|
{
|
|
|
|
BasicBlock block = blocks[blkIndex];
|
|
|
|
|
|
|
|
if (block.Predecessors.Count == 0)
|
|
|
|
{
|
|
|
|
block.Next = null;
|
|
|
|
block.Branch = null;
|
|
|
|
blocks.RemoveAt(blkIndex--);
|
|
|
|
hasUnreachable = true;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
block.Index = blkIndex;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} while (hasUnreachable);
|
|
|
|
|
2020-10-25 21:00:44 +01:00
|
|
|
return new ControlFlowGraph(blocks.ToArray());
|
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
|
|
|
}
|
|
|
|
|
|
|
|
private static bool EndsWithUnconditionalInst(INode node)
|
|
|
|
{
|
|
|
|
if (node is Operation operation)
|
|
|
|
{
|
|
|
|
switch (operation.Inst)
|
|
|
|
{
|
|
|
|
case Instruction.Branch:
|
|
|
|
case Instruction.Discard:
|
|
|
|
case Instruction.Return:
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
2023-06-28 08:59:13 +02:00
|
|
|
}
|