1
0
Fork 0
mirror of https://github.com/Ryujinx/Ryujinx.git synced 2024-10-01 12:30:00 +02:00
Ryujinx/Ryujinx.Graphics.Shader/CodeGen/Glsl/Instructions/InstGenMemory.cs

673 lines
23 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
2019-04-18 01:57:08 +02:00
using Ryujinx.Graphics.Shader.IntermediateRepresentation;
using Ryujinx.Graphics.Shader.StructuredIr;
2019-10-13 08:02:07 +02:00
using System;
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 static Ryujinx.Graphics.Shader.CodeGen.Glsl.Instructions.InstGenHelper;
using static Ryujinx.Graphics.Shader.StructuredIr.InstructionInfo;
namespace Ryujinx.Graphics.Shader.CodeGen.Glsl.Instructions
{
static class InstGenMemory
{
public static string ImageLoadOrStore(CodeGenContext context, AstOperation operation)
{
AstTextureOperation texOp = (AstTextureOperation)operation;
bool isBindless = (texOp.Flags & TextureFlags.Bindless) != 0;
// TODO: Bindless texture support. For now we just return 0/do nothing.
if (isBindless)
{
return texOp.Inst == Instruction.ImageLoad ? NumberFormatter.FormatFloat(0) : "// imageStore(bindless)";
}
bool isArray = (texOp.Type & SamplerType.Array) != 0;
bool isIndexed = (texOp.Type & SamplerType.Indexed) != 0;
string texCall = texOp.Inst == Instruction.ImageLoad ? "imageLoad" : "imageStore";
int srcIndex = isBindless ? 1 : 0;
string Src(VariableType type)
{
return GetSoureExpr(context, texOp.GetSource(srcIndex++), type);
}
string indexExpr = null;
if (isIndexed)
{
indexExpr = Src(VariableType.S32);
}
string imageName = OperandManager.GetImageName(context.Config.Stage, texOp, indexExpr);
texCall += "(" + imageName;
int coordsCount = texOp.Type.GetDimensions();
int pCount = coordsCount + (isArray ? 1 : 0);
void Append(string str)
{
texCall += ", " + str;
}
string ApplyScaling(string vector)
{
int index = context.FindImageDescriptorIndex(texOp);
TextureUsageFlags flags = TextureUsageFlags.NeedsScaleValue;
if ((context.Config.Stage == ShaderStage.Fragment || context.Config.Stage == ShaderStage.Compute) &&
texOp.Inst == Instruction.ImageLoad &&
!isBindless &&
!isIndexed)
{
// Image scales start after texture ones.
int scaleIndex = context.TextureDescriptors.Count + index;
if (pCount == 3 && isArray)
{
// The array index is not scaled, just x and y.
vector = "ivec3(Helper_TexelFetchScale((" + vector + ").xy, " + scaleIndex + "), (" + vector + ").z)";
}
else if (pCount == 2 && !isArray)
{
vector = "Helper_TexelFetchScale(" + vector + ", " + scaleIndex + ")";
}
else
{
flags |= TextureUsageFlags.ResScaleUnsupported;
}
}
else
{
flags |= TextureUsageFlags.ResScaleUnsupported;
}
if (!isBindless)
{
context.ImageDescriptors[index] = context.ImageDescriptors[index].SetFlag(flags);
}
return vector;
}
if (pCount > 1)
{
string[] elems = new string[pCount];
for (int index = 0; index < pCount; index++)
{
elems[index] = Src(VariableType.S32);
}
Append(ApplyScaling("ivec" + pCount + "(" + string.Join(", ", elems) + ")"));
}
else
{
Append(Src(VariableType.S32));
}
if (texOp.Inst == Instruction.ImageStore)
{
VariableType type = texOp.Format.GetComponentType();
string[] cElems = new string[4];
for (int index = 0; index < 4; index++)
{
if (srcIndex < texOp.SourcesCount)
{
cElems[index] = Src(type);
}
else
{
cElems[index] = type switch
{
VariableType.S32 => NumberFormatter.FormatInt(0),
VariableType.U32 => NumberFormatter.FormatUint(0),
_ => NumberFormatter.FormatFloat(0)
};
}
}
string prefix = type switch
{
VariableType.S32 => "i",
VariableType.U32 => "u",
_ => string.Empty
};
Append(prefix + "vec4(" + string.Join(", ", cElems) + ")");
}
texCall += ")" + (texOp.Inst == Instruction.ImageLoad ? GetMask(texOp.Index) : "");
return texCall;
}
2019-10-13 08:02:07 +02:00
public static string LoadAttribute(CodeGenContext context, AstOperation operation)
{
IAstNode src1 = operation.GetSource(0);
IAstNode src2 = operation.GetSource(1);
if (!(src1 is AstOperand attr) || attr.Type != OperandType.Attribute)
{
throw new InvalidOperationException("First source of LoadAttribute must be a attribute.");
}
string indexExpr = GetSoureExpr(context, src2, GetSrcVarType(operation.Inst, 1));
return OperandManager.GetAttributeName(attr, context.Config, isOutAttr: false, indexExpr);
2019-10-13 08:02:07 +02:00
}
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
2019-04-18 01:57:08 +02:00
public static string LoadConstant(CodeGenContext context, AstOperation operation)
{
2019-10-13 08:02:07 +02:00
IAstNode src1 = operation.GetSource(0);
IAstNode src2 = operation.GetSource(1);
string offsetExpr = GetSoureExpr(context, src2, GetSrcVarType(operation.Inst, 1));
offsetExpr = Enclose(offsetExpr, src2, Instruction.ShiftRightS32, isLhs: true);
if (src1 is AstOperand oper && oper.Type == OperandType.Constant)
{
return OperandManager.GetConstantBufferName(oper.Value, offsetExpr, context.Config.Stage, context.CbIndexable);
}
else
{
string slotExpr = GetSoureExpr(context, src1, GetSrcVarType(operation.Inst, 0));
return OperandManager.GetConstantBufferName(slotExpr, offsetExpr, context.Config.Stage);
}
2019-10-13 08:02:07 +02:00
}
public static string LoadLocal(CodeGenContext context, AstOperation operation)
{
return LoadLocalOrShared(context, operation, DefaultNames.LocalMemoryName);
}
public static string LoadShared(CodeGenContext context, AstOperation operation)
{
return LoadLocalOrShared(context, operation, DefaultNames.SharedMemoryName);
}
private static string LoadLocalOrShared(CodeGenContext context, AstOperation operation, string arrayName)
2019-10-13 08:02:07 +02:00
{
IAstNode src1 = operation.GetSource(0);
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
2019-10-13 08:02:07 +02:00
string offsetExpr = GetSoureExpr(context, src1, GetSrcVarType(operation.Inst, 0));
return $"{arrayName}[{offsetExpr}]";
2019-10-13 08:02:07 +02:00
}
public static string LoadStorage(CodeGenContext context, AstOperation operation)
{
IAstNode src1 = operation.GetSource(0);
IAstNode src2 = operation.GetSource(1);
string indexExpr = GetSoureExpr(context, src1, GetSrcVarType(operation.Inst, 0));
string offsetExpr = GetSoureExpr(context, src2, GetSrcVarType(operation.Inst, 1));
return GetStorageBufferAccessor(indexExpr, offsetExpr, context.Config.Stage);
2019-10-13 08:02:07 +02:00
}
public static string Lod(CodeGenContext context, AstOperation operation)
{
AstTextureOperation texOp = (AstTextureOperation)operation;
int coordsCount = texOp.Type.GetDimensions();
bool isBindless = (texOp.Flags & TextureFlags.Bindless) != 0;
// TODO: Bindless texture support. For now we just return 0.
if (isBindless)
{
return NumberFormatter.FormatFloat(0);
}
bool isIndexed = (texOp.Type & SamplerType.Indexed) != 0;
string indexExpr = null;
if (isIndexed)
{
indexExpr = GetSoureExpr(context, texOp.GetSource(0), VariableType.S32);
}
string samplerName = OperandManager.GetSamplerName(context.Config.Stage, texOp, indexExpr);
int coordsIndex = isBindless || isIndexed ? 1 : 0;
string coordsExpr;
if (coordsCount > 1)
{
string[] elems = new string[coordsCount];
for (int index = 0; index < coordsCount; index++)
{
elems[index] = GetSoureExpr(context, texOp.GetSource(coordsIndex + index), VariableType.F32);
}
coordsExpr = "vec" + coordsCount + "(" + string.Join(", ", elems) + ")";
}
else
{
coordsExpr = GetSoureExpr(context, texOp.GetSource(coordsIndex), VariableType.F32);
}
return $"textureQueryLod({samplerName}, {coordsExpr}){GetMask(texOp.Index)}";
}
2019-10-13 08:02:07 +02:00
public static string StoreLocal(CodeGenContext context, AstOperation operation)
{
return StoreLocalOrShared(context, operation, DefaultNames.LocalMemoryName);
}
public static string StoreShared(CodeGenContext context, AstOperation operation)
{
return StoreLocalOrShared(context, operation, DefaultNames.SharedMemoryName);
}
private static string StoreLocalOrShared(CodeGenContext context, AstOperation operation, string arrayName)
2019-10-13 08:02:07 +02:00
{
IAstNode src1 = operation.GetSource(0);
IAstNode src2 = operation.GetSource(1);
string offsetExpr = GetSoureExpr(context, src1, GetSrcVarType(operation.Inst, 0));
VariableType srcType = OperandManager.GetNodeDestType(context, src2);
2019-10-13 08:02:07 +02:00
string src = TypeConversion.ReinterpretCast(context, src2, srcType, VariableType.U32);
2019-10-13 08:02:07 +02:00
return $"{arrayName}[{offsetExpr}] = {src}";
2019-10-13 08:02:07 +02:00
}
public static string StoreStorage(CodeGenContext context, AstOperation operation)
{
IAstNode src1 = operation.GetSource(0);
IAstNode src2 = operation.GetSource(1);
IAstNode src3 = operation.GetSource(2);
2019-10-13 08:02:07 +02:00
string indexExpr = GetSoureExpr(context, src1, GetSrcVarType(operation.Inst, 0));
string offsetExpr = GetSoureExpr(context, src2, GetSrcVarType(operation.Inst, 1));
2019-10-13 08:02:07 +02:00
VariableType srcType = OperandManager.GetNodeDestType(context, src3);
2019-10-13 08:02:07 +02:00
string src = TypeConversion.ReinterpretCast(context, src3, srcType, VariableType.U32);
2019-10-13 08:02:07 +02:00
Implement lazy flush-on-read for Buffers (SSBO/Copy) (#1790) * Initial implementation of buffer flush (VERY WIP) * Host shaders need to be rebuilt for the SSBO write flag. * New approach with reserved regions and gl sync * Fix a ton of buffer issues. * Remove unused buffer unmapped behaviour * Revert "Remove unused buffer unmapped behaviour" This reverts commit f1700e52fb8760180ac5e0987a07d409d1e70ece. * Delete modified ranges on unmap Fixes potential crashes in Super Smash Bros, where a previously modified range could lie on either side of an unmap. * Cache some more delegates. * Dispose Sync on Close * Also create host sync for GPFifo syncpoint increment. * Copy buffer optimization, add docs * Fix race condition with OpenGL Sync * Enable read tracking on CommandBuffer, insert syncpoint on WaitForIdle * Performance: Only flush individual pages of SSBO at a time This avoids flushing large amounts of data when only a small amount is actually used. * Signal Modified rather than flushing after clear * Fix some docs and code style. * Introduce a new test for tracking memory protection. Sucessfully demonstrates that the bug causing write protection to be cleared by a read action has been fixed. (these tests fail on master) * Address Comments * Add host sync for SetReference This ensures that any indirect draws will correctly flush any related buffer data written before them. Fixes some flashing and misplaced world geometry in MH rise. * Make PageAlign static * Re-enable read tracking, for reads.
2021-01-17 21:08:06 +01:00
SetStorageWriteFlag(context, src1, context.Config.Stage);
string sb = GetStorageBufferAccessor(indexExpr, offsetExpr, context.Config.Stage);
2019-10-13 08:02:07 +02:00
return $"{sb} = {src}";
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
}
public static string TextureSample(CodeGenContext context, AstOperation operation)
{
AstTextureOperation texOp = (AstTextureOperation)operation;
bool isBindless = (texOp.Flags & TextureFlags.Bindless) != 0;
bool isGather = (texOp.Flags & TextureFlags.Gather) != 0;
bool hasDerivatives = (texOp.Flags & TextureFlags.Derivatives) != 0;
bool intCoords = (texOp.Flags & TextureFlags.IntCoords) != 0;
bool hasLodBias = (texOp.Flags & TextureFlags.LodBias) != 0;
bool hasLodLevel = (texOp.Flags & TextureFlags.LodLevel) != 0;
bool hasOffset = (texOp.Flags & TextureFlags.Offset) != 0;
bool hasOffsets = (texOp.Flags & TextureFlags.Offsets) != 0;
2019-10-13 08:02:07 +02:00
bool isArray = (texOp.Type & SamplerType.Array) != 0;
bool isIndexed = (texOp.Type & SamplerType.Indexed) != 0;
bool isMultisample = (texOp.Type & SamplerType.Multisample) != 0;
bool isShadow = (texOp.Type & SamplerType.Shadow) != 0;
2019-10-13 08:02:07 +02:00
// TODO: Bindless texture support. For now we just return 0.
if (isBindless)
{
return NumberFormatter.FormatFloat(0);
}
2019-10-13 08:02:07 +02:00
// This combination is valid, but not available on GLSL.
// For now, ignore the LOD level and do a normal sample.
// TODO: How to implement it properly?
if (hasLodLevel && isArray && isShadow)
{
hasLodLevel = 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
string texCall = intCoords ? "texelFetch" : "texture";
if (isGather)
{
texCall += "Gather";
}
else if (hasDerivatives)
{
texCall += "Grad";
}
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
else if (hasLodLevel && !intCoords)
{
texCall += "Lod";
}
if (hasOffset)
{
texCall += "Offset";
}
else if (hasOffsets)
{
texCall += "Offsets";
}
int srcIndex = isBindless ? 1 : 0;
string Src(VariableType type)
{
return GetSoureExpr(context, texOp.GetSource(srcIndex++), type);
}
string indexExpr = null;
if (isIndexed)
{
indexExpr = Src(VariableType.S32);
}
string samplerName = OperandManager.GetSamplerName(context.Config.Stage, texOp, indexExpr);
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
texCall += "(" + samplerName;
int coordsCount = texOp.Type.GetDimensions();
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 pCount = coordsCount;
int arrayIndexElem = -1;
if (isArray)
{
arrayIndexElem = pCount++;
}
// The sampler 1D shadow overload expects a
// dummy value on the middle of the vector, who knows why...
bool hasDummy1DShadowElem = texOp.Type == (SamplerType.Texture1D | SamplerType.Shadow);
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 (hasDummy1DShadowElem)
{
pCount++;
}
if (isShadow && !isGather)
{
pCount++;
}
// On textureGather*, the comparison value is
// always specified as an extra argument.
New shader translator implementation (#654) * Start implementing a new shader translator * Fix shift instructions and a typo * Small refactoring on StructuredProgram, move RemovePhis method to a separate class * Initial geometry shader support * Implement TLD4 * Fix -- There's no negation on FMUL32I * Add constant folding and algebraic simplification optimizations, nits * Some leftovers from constant folding * Avoid cast for constant assignments * Add a branch elimination pass, and misc small fixes * Remove redundant branches, add expression propagation and other improvements on the code * Small leftovers -- add missing break and continue, remove unused properties, other improvements * Add null check to handle empty block cases on block visitor * Add HADD2 and HMUL2 half float shader instructions * Optimize pack/unpack sequences, some fixes related to half float instructions * Add TXQ, TLD, TLDS and TLD4S shader texture instructions, and some support for bindless textures, some refactoring on codegen * Fix copy paste mistake that caused RZ to be ignored on the AST instruction * Add workaround for conditional exit, and fix half float instruction with constant buffer * Add missing 0.0 source for TLDS.LZ variants * Simplify the switch for TLDS.LZ * Texture instructions related fixes * Implement the HFMA instruction, and some misc. fixes * Enable constant folding on UnpackHalf2x16 instructions * Refactor HFMA to use OpCode* for opcode decoding rather than on the helper methods * Remove the old shader translator * Remove ShaderDeclInfo and other unused things * Add dual vertex shader support * Add ShaderConfig, used to pass shader type and maximum cbuffer size * Move and rename some instruction enums * Move texture instructions into a separate file * Move operand GetExpression and locals management to OperandManager * Optimize opcode decoding using a simple list and binary search * Add missing condition for do-while on goto elimination * Misc. fixes on texture instructions * Simplify TLDS switch * Address PR feedback, and a nit
2019-04-18 01:57:08 +02:00
bool hasExtraCompareArg = isShadow && isGather;
if (pCount == 5)
{
pCount = 4;
hasExtraCompareArg = true;
}
void Append(string str)
{
texCall += ", " + str;
}
VariableType coordType = intCoords ? VariableType.S32 : VariableType.F32;
string AssemblePVector(int count)
{
if (count > 1)
{
string[] elems = new string[count];
for (int index = 0; index < count; index++)
{
if (arrayIndexElem == index)
{
elems[index] = Src(VariableType.S32);
if (!intCoords)
{
elems[index] = "float(" + elems[index] + ")";
}
}
else if (index == 1 && hasDummy1DShadowElem)
{
elems[index] = NumberFormatter.FormatFloat(0);
}
else
{
elems[index] = Src(coordType);
}
}
string prefix = intCoords ? "i" : string.Empty;
return prefix + "vec" + count + "(" + string.Join(", ", elems) + ")";
}
else
{
return Src(coordType);
}
}
string ApplyScaling(string vector)
{
if (intCoords)
{
int index = context.FindTextureDescriptorIndex(texOp);
TextureUsageFlags flags = TextureUsageFlags.NeedsScaleValue;
if ((context.Config.Stage == ShaderStage.Fragment || context.Config.Stage == ShaderStage.Compute) &&
!isBindless &&
!isIndexed)
{
if (pCount == 3 && isArray)
{
// The array index is not scaled, just x and y.
vector = "ivec3(Helper_TexelFetchScale((" + vector + ").xy, " + index + "), (" + vector + ").z)";
}
else if (pCount == 2 && !isArray)
{
vector = "Helper_TexelFetchScale(" + vector + ", " + index + ")";
}
else
{
flags |= TextureUsageFlags.ResScaleUnsupported;
}
}
else
{
// Resolution scaling cannot be applied to this texture right now.
// Flag so that we know to blacklist scaling on related textures when binding them.
flags |= TextureUsageFlags.ResScaleUnsupported;
}
if (!isBindless)
{
context.TextureDescriptors[index] = context.TextureDescriptors[index].SetFlag(flags);
}
}
return vector;
}
Append(ApplyScaling(AssemblePVector(pCount)));
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
string AssembleDerivativesVector(int count)
{
if (count > 1)
{
string[] elems = new string[count];
for (int index = 0; index < count; index++)
{
elems[index] = Src(VariableType.F32);
}
return "vec" + count + "(" + string.Join(", ", elems) + ")";
}
else
{
return Src(VariableType.F32);
}
}
if (hasExtraCompareArg)
{
Append(Src(VariableType.F32));
}
if (hasDerivatives)
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
{
Append(AssembleDerivativesVector(coordsCount)); // dPdx
Append(AssembleDerivativesVector(coordsCount)); // dPdy
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 (isMultisample)
{
Append(Src(VariableType.S32));
}
else if (hasLodLevel)
{
Append(Src(coordType));
}
string AssembleOffsetVector(int count)
{
if (count > 1)
{
string[] elems = new string[count];
for (int index = 0; index < count; index++)
{
elems[index] = Src(VariableType.S32);
}
return "ivec" + count + "(" + string.Join(", ", elems) + ")";
}
else
{
return Src(VariableType.S32);
}
}
if (hasOffset)
{
Append(AssembleOffsetVector(coordsCount));
}
else if (hasOffsets)
{
texCall += $", ivec{coordsCount}[4](";
texCall += AssembleOffsetVector(coordsCount) + ", ";
texCall += AssembleOffsetVector(coordsCount) + ", ";
texCall += AssembleOffsetVector(coordsCount) + ", ";
texCall += AssembleOffsetVector(coordsCount) + ")";
}
if (hasLodBias)
{
Append(Src(VariableType.F32));
}
// textureGather* optional extra component index,
// not needed for shadow samplers.
New shader translator implementation (#654) * Start implementing a new shader translator * Fix shift instructions and a typo * Small refactoring on StructuredProgram, move RemovePhis method to a separate class * Initial geometry shader support * Implement TLD4 * Fix -- There's no negation on FMUL32I * Add constant folding and algebraic simplification optimizations, nits * Some leftovers from constant folding * Avoid cast for constant assignments * Add a branch elimination pass, and misc small fixes * Remove redundant branches, add expression propagation and other improvements on the code * Small leftovers -- add missing break and continue, remove unused properties, other improvements * Add null check to handle empty block cases on block visitor * Add HADD2 and HMUL2 half float shader instructions * Optimize pack/unpack sequences, some fixes related to half float instructions * Add TXQ, TLD, TLDS and TLD4S shader texture instructions, and some support for bindless textures, some refactoring on codegen * Fix copy paste mistake that caused RZ to be ignored on the AST instruction * Add workaround for conditional exit, and fix half float instruction with constant buffer * Add missing 0.0 source for TLDS.LZ variants * Simplify the switch for TLDS.LZ * Texture instructions related fixes * Implement the HFMA instruction, and some misc. fixes * Enable constant folding on UnpackHalf2x16 instructions * Refactor HFMA to use OpCode* for opcode decoding rather than on the helper methods * Remove the old shader translator * Remove ShaderDeclInfo and other unused things * Add dual vertex shader support * Add ShaderConfig, used to pass shader type and maximum cbuffer size * Move and rename some instruction enums * Move texture instructions into a separate file * Move operand GetExpression and locals management to OperandManager * Optimize opcode decoding using a simple list and binary search * Add missing condition for do-while on goto elimination * Misc. fixes on texture instructions * Simplify TLDS switch * Address PR feedback, and a nit
2019-04-18 01:57:08 +02:00
if (isGather && !isShadow)
{
Append(Src(VariableType.S32));
}
texCall += ")" + (isGather || !isShadow ? GetMask(texOp.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
2019-04-18 01:57:08 +02:00
return texCall;
}
public static string TextureSize(CodeGenContext context, AstOperation operation)
{
AstTextureOperation texOp = (AstTextureOperation)operation;
bool isBindless = (texOp.Flags & TextureFlags.Bindless) != 0;
// TODO: Bindless texture support. For now we just return 0.
if (isBindless)
{
return NumberFormatter.FormatInt(0);
}
bool isIndexed = (texOp.Type & SamplerType.Indexed) != 0;
string indexExpr = null;
if (isIndexed)
{
indexExpr = GetSoureExpr(context, texOp.GetSource(0), VariableType.S32);
}
New shader translator implementation (#654) * Start implementing a new shader translator * Fix shift instructions and a typo * Small refactoring on StructuredProgram, move RemovePhis method to a separate class * Initial geometry shader support * Implement TLD4 * Fix -- There's no negation on FMUL32I * Add constant folding and algebraic simplification optimizations, nits * Some leftovers from constant folding * Avoid cast for constant assignments * Add a branch elimination pass, and misc small fixes * Remove redundant branches, add expression propagation and other improvements on the code * Small leftovers -- add missing break and continue, remove unused properties, other improvements * Add null check to handle empty block cases on block visitor * Add HADD2 and HMUL2 half float shader instructions * Optimize pack/unpack sequences, some fixes related to half float instructions * Add TXQ, TLD, TLDS and TLD4S shader texture instructions, and some support for bindless textures, some refactoring on codegen * Fix copy paste mistake that caused RZ to be ignored on the AST instruction * Add workaround for conditional exit, and fix half float instruction with constant buffer * Add missing 0.0 source for TLDS.LZ variants * Simplify the switch for TLDS.LZ * Texture instructions related fixes * Implement the HFMA instruction, and some misc. fixes * Enable constant folding on UnpackHalf2x16 instructions * Refactor HFMA to use OpCode* for opcode decoding rather than on the helper methods * Remove the old shader translator * Remove ShaderDeclInfo and other unused things * Add dual vertex shader support * Add ShaderConfig, used to pass shader type and maximum cbuffer size * Move and rename some instruction enums * Move texture instructions into a separate file * Move operand GetExpression and locals management to OperandManager * Optimize opcode decoding using a simple list and binary search * Add missing condition for do-while on goto elimination * Misc. fixes on texture instructions * Simplify TLDS switch * Address PR feedback, and a nit
2019-04-18 01:57:08 +02:00
string samplerName = OperandManager.GetSamplerName(context.Config.Stage, texOp, indexExpr);
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 lodSrcIndex = isBindless || isIndexed ? 1 : 0;
IAstNode lod = operation.GetSource(lodSrcIndex);
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
string lodExpr = GetSoureExpr(context, lod, GetSrcVarType(operation.Inst, lodSrcIndex));
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
2020-03-22 22:31:31 +01:00
if (texOp.Index == 3)
{
return $"textureQueryLevels({samplerName})";
}
else
{
return $"textureSize({samplerName}, {lodExpr}){GetMask(texOp.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
2019-04-18 01:57:08 +02:00
}
Implement lazy flush-on-read for Buffers (SSBO/Copy) (#1790) * Initial implementation of buffer flush (VERY WIP) * Host shaders need to be rebuilt for the SSBO write flag. * New approach with reserved regions and gl sync * Fix a ton of buffer issues. * Remove unused buffer unmapped behaviour * Revert "Remove unused buffer unmapped behaviour" This reverts commit f1700e52fb8760180ac5e0987a07d409d1e70ece. * Delete modified ranges on unmap Fixes potential crashes in Super Smash Bros, where a previously modified range could lie on either side of an unmap. * Cache some more delegates. * Dispose Sync on Close * Also create host sync for GPFifo syncpoint increment. * Copy buffer optimization, add docs * Fix race condition with OpenGL Sync * Enable read tracking on CommandBuffer, insert syncpoint on WaitForIdle * Performance: Only flush individual pages of SSBO at a time This avoids flushing large amounts of data when only a small amount is actually used. * Signal Modified rather than flushing after clear * Fix some docs and code style. * Introduce a new test for tracking memory protection. Sucessfully demonstrates that the bug causing write protection to be cleared by a read action has been fixed. (these tests fail on master) * Address Comments * Add host sync for SetReference This ensures that any indirect draws will correctly flush any related buffer data written before them. Fixes some flashing and misplaced world geometry in MH rise. * Make PageAlign static * Re-enable read tracking, for reads.
2021-01-17 21:08:06 +01:00
private static void SetStorageWriteFlag(CodeGenContext context, IAstNode indexExpr, ShaderStage stage)
{
// Attempt to find a BufferDescriptor with the given index.
// If it cannot be resolved or is not constant, assume that the slot expression could potentially index any of them,
// and set the flag on all storage buffers.
int index = -1;
if (indexExpr is AstOperand operand && operand.Type == OperandType.Constant)
{
index = context.SBufferDescriptors.FindIndex(buffer => buffer.Slot == operand.Value);
}
if (index != -1)
{
context.SBufferDescriptors[index] = context.SBufferDescriptors[index].SetFlag(BufferUsageFlags.Write);
}
else
{
for (int i = 0; i < context.SBufferDescriptors.Count; i++)
{
context.SBufferDescriptors[i] = context.SBufferDescriptors[i].SetFlag(BufferUsageFlags.Write);
}
}
}
private static string GetStorageBufferAccessor(string slotExpr, string offsetExpr, ShaderStage stage)
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
{
string sbName = OperandManager.GetShaderStagePrefix(stage);
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
sbName += "_" + DefaultNames.StorageNamePrefix;
return $"{sbName}[{slotExpr}].{DefaultNames.DataName}[{offsetExpr}]";
}
private static string GetMask(int index)
{
return '.' + "rgba".Substring(index, 1);
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
}
}
}