archive/Cemu
1
0
Fork 0
mirror of https://github.com/cemu-project/Cemu.git synced 2025-07-10 00:41:19 +12:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

Latte: Small refactor for shader decompiler

Browse source 
Latte: Small refactor for shader decompiler
This commit is contained in:
Exzap 2023-03-01 22:16:57 +01:00
parent 4a564e2447
commit f8ea594447
9 changed files with 67 additions and 149 deletions
Download patch file Download diff file Expand all files Collapse all files

97
src/Cafe/HW/Latte/LegacyShaderDecompiler/LatteDecompilerEmitGLSL.cpp
View file

@ -3363,7 +3363,7 @@ void _emitCFRingWriteCode(LatteDecompilerShaderContext* shaderContext, LatteDeco
if ((cfInstruction->memWriteCompMask&(1 << i)) == 0)
continue;
if (shaderContext->useTFViaSSBO)
if (shaderContext->options->useTFViaSSBO)
{
uint32 u32Offset = streamWrite->exportArrayBase + i;
src->addFmt("sb_buffer[sbBase{} + {}]", streamWrite->bufferIndex, u32Offset);
@ -3483,7 +3483,7 @@ void _emitStreamWriteCode(LatteDecompilerShaderContext* shaderContext, LatteDeco
if ((cfInstruction->memWriteCompMask&(1 << i)) == 0)
continue;
if (shaderContext->useTFViaSSBO)
if (shaderContext->options->useTFViaSSBO)
{
uint32 u32Offset = cfInstruction->exportArrayBase + i;
src->addFmt("sb_buffer[sbBase{} + {}]", streamoutBufferIndex, u32Offset);
@ -3831,7 +3831,7 @@ void LatteDecompiler_emitGLSLHelperFunctions(LatteDecompilerShaderContext* shade
//fCStr_shaderSource->add("float mul_nonIEEE(float a, float b){float r = a*b;r = intBitsToFloat(floatBitsToInt(r)&(((floatBitsToInt(a) != 0) && (floatBitsToInt(b) != 0))?0xFFFFFFFF:0));return r;}" STR_LINEBREAK); works
if( LatteGPUState.glVendor == GLVENDOR_NVIDIA && !ActiveSettings::DumpShadersEnabled())
fCStr_shaderSource->add("float mul_nonIEEE(float a, float b){return mix(0.0, a*b, (a != 0.0) && (b != 0.0));}" _CRLF); // compiles faster on Nvidia and also results in lower RAM usage
fCStr_shaderSource->add("float mul_nonIEEE(float a, float b){return mix(0.0, a*b, (a != 0.0) && (b != 0.0));}" _CRLF); // compiles faster on Nvidia and also results in lower RAM usage (OpenGL)
else
fCStr_shaderSource->add("float mul_nonIEEE(float a, float b){ if( a == 0.0 || b == 0.0 ) return 0.0; return a*b; }" _CRLF);
}
@ -3841,51 +3841,6 @@ void LatteDecompiler_emitGLSLHelperFunctions(LatteDecompilerShaderContext* shade
}
}
void _addPixelShaderExtraDebugInfo(LatteDecompilerShaderContext* shaderContext, StringBuf* fCStr_shaderSource)
{
#ifdef CEMU_DEBUG_ASSERT
fCStr_shaderSource->add("// Color buffers:" _CRLF);
for(uint32 i=0; i<8; i++)
{
uint32 regColorBuffer = shaderContext->contextRegisters[mmCB_COLOR0_BASE+i];
uint32 regColorSize = shaderContext->contextRegisters[mmCB_COLOR0_SIZE+i];
uint32 regColorInfo = shaderContext->contextRegisters[mmCB_COLOR0_INFO+i];
uint32 regColorView = shaderContext->contextRegisters[mmCB_COLOR0_VIEW+i];
MPTR colorBufferPhysMem = regColorBuffer;
if( regColorBuffer == MPTR_NULL )
continue;
uint32 colorBufferFormat = (regColorInfo>>2)&0x3F; // format
uint32 colorBufferTileMode = 0;
colorBufferTileMode = (regColorInfo >> 8) & 0xF;
switch ( (regColorInfo >> 12) & 7 )
{
case 4:
colorBufferFormat |= 0x100;
break;
case 1:
colorBufferFormat |= 0x200;
break;
case 5:
colorBufferFormat |= 0x300;
break;
case 6:
colorBufferFormat |= 0x400;
break;
case 7:
colorBufferFormat |= 0x800;
break;
default:
break;
}
uint32 colorBufferWidth = (regColorSize>>0)&0xFFFF;
uint32 colorBufferHeight = (regColorSize>>16)&0xFFFF;
fCStr_shaderSource->addFmt("// Color{}: {}x{} at 0x{:08x} fmt {:04x} tm {}" _CRLF, i, colorBufferWidth, colorBufferHeight, colorBufferPhysMem, colorBufferFormat, colorBufferTileMode);
}
#endif
}
#include "Cafe/HW/Latte/LegacyShaderDecompiler/LatteDecompilerEmitGLSLHeader.hpp"
void LatteDecompiler_emitAttributeImport(LatteDecompilerShaderContext* shaderContext, LatteParsedFetchShaderAttribute_t& attrib)
@ -3954,18 +3909,14 @@ void LatteDecompiler_emitGLSLShader(LatteDecompilerShaderContext* shaderContext,
src->add("#extension GL_ARB_texture_gather : enable" _CRLF);
src->add("#extension GL_ARB_separate_shader_objects : enable" _CRLF);
if (shaderContext->analyzer.hasStreamoutWrite || shaderContext->usesGeometryShader )
if (shaderContext->analyzer.hasStreamoutWrite || shaderContext->options->usesGeometryShader )
src->add("#extension GL_ARB_enhanced_layouts : enable" _CRLF);
// debug info
src->addFmt("// shader %08x%08x" _CRLF, (uint32)(shaderContext->shaderBaseHash >> 32), (uint32)(shaderContext->shaderBaseHash & 0xFFFFFFFF));
src->addFmt("// shader {:016x}" _CRLF, shaderContext->shaderBaseHash);
#ifdef CEMU_DEBUG_ASSERT
src->addFmt("// usesIntegerValues: {}" _CRLF, shaderContext->analyzer.usesIntegerValues?"true":"false");
src->addFmt(_CRLF);
if( shader->shaderType == LatteConst::ShaderType::Pixel )
_addPixelShaderExtraDebugInfo(shaderContext, src);
#endif
// header part (definitions for inputs and outputs)
LatteDecompiler::emitHeader(shaderContext);
@ -3982,7 +3933,6 @@ void LatteDecompiler_emitGLSLShader(LatteDecompilerShaderContext* shaderContext,
{
if (shaderContext->analyzer.usesRelativeGPRRead || (shaderContext->analyzer.gprUseMask[i / 8] & (1 << (i & 7))) != 0)
{
//fCStr_appendFormatted(fCStr_shaderSource, "ivec4 R{}i, R{}i, R{}i, R{}i;" STR_LINEBREAK, i*4+0, i*4+1, i*4+2, i*4+3);
if (shaderContext->typeTracker.genIntReg)
src->addFmt("ivec4 R{}i = ivec4(0);" _CRLF, i);
else if (shaderContext->typeTracker.genFloatReg)
@ -4035,7 +3985,6 @@ void LatteDecompiler_emitGLSLShader(LatteDecompilerShaderContext* shaderContext,
src->add("bool predResult = true;" _CRLF);
if(shaderContext->analyzer.modifiesPixelActiveState )
{
// cemu_assert_debug(shaderContext->analyzer.activeStackMaxDepth == 0);
src->addFmt("bool activeMaskStack[{}];" _CRLF, shaderContext->analyzer.activeStackMaxDepth+1);
src->addFmt("bool activeMaskStackC[{}];" _CRLF, shaderContext->analyzer.activeStackMaxDepth+2);
for (sint32 i = 0; i < shaderContext->analyzer.activeStackMaxDepth; i++)
@ -4058,8 +4007,11 @@ void LatteDecompiler_emitGLSLShader(LatteDecompilerShaderContext* shaderContext,
}
}
// helper variables for cube maps (todo: Only emit when used)
src->addFmt("vec3 cubeMapSTM;" _CRLF);
src->addFmt("int cubeMapFaceId;" _CRLF);
if (shaderContext->analyzer.hasRedcCUBE)
{
src->add("vec3 cubeMapSTM;" _CRLF);
src->add("int cubeMapFaceId;" _CRLF);
}
for(sint32 i=0; i<LATTE_NUM_MAX_TEX_UNITS; i++)
{
if(!shaderContext->output->textureUnitMask[i])
@ -4106,25 +4058,18 @@ void LatteDecompiler_emitGLSLShader(LatteDecompilerShaderContext* shaderContext,
cemu_assert_unimplemented();
}
if (shaderContext->fetchShaderCount == 1)
LatteFetchShader* parsedFetchShader = shaderContext->fetchShader;
for(auto& bufferGroup : parsedFetchShader->bufferGroups)
{
LatteFetchShader* parsedFetchShader = shaderContext->fetchShaderList[0];
for(auto& bufferGroup : parsedFetchShader->bufferGroups)
{
for(sint32 i=0; i<bufferGroup.attribCount; i++)
LatteDecompiler_emitAttributeImport(shaderContext, bufferGroup.attrib[i]);
}
for (auto& bufferGroup : parsedFetchShader->bufferGroupsInvalid)
{
// these attributes point to non-existent buffers
// todo - figure out how the hardware actually handles this, currently we assume the input values are zero
for (sint32 i = 0; i < bufferGroup.attribCount; i++)
LatteDecompiler_emitAttributeImport(shaderContext, bufferGroup.attrib[i]);
}
for(sint32 i=0; i<bufferGroup.attribCount; i++)
LatteDecompiler_emitAttributeImport(shaderContext, bufferGroup.attrib[i]);
}
else
for (auto& bufferGroup : parsedFetchShader->bufferGroupsInvalid)
{
cemu_assert_unimplemented();
// these attributes point to non-existent buffers
// todo - figure out how the hardware actually handles this, currently we assume the input values are zero
for (sint32 i = 0; i < bufferGroup.attribCount; i++)
LatteDecompiler_emitAttributeImport(shaderContext, bufferGroup.attrib[i]);
}
}
else if (shader->shaderType == LatteConst::ShaderType::Pixel)
@ -4172,7 +4117,7 @@ void LatteDecompiler_emitGLSLShader(LatteDecompilerShaderContext* shaderContext,
continue;
}
if (shaderContext->usesGeometryShader)
if (shaderContext->options->usesGeometryShader)
{
// import from geometry shader
if (shaderContext->typeTracker.defaultDataType == LATTE_DECOMPILER_DTYPE_SIGNED_INT)
@ -4216,7 +4161,7 @@ void LatteDecompiler_emitGLSLShader(LatteDecompilerShaderContext* shaderContext,
// vertex shader should write renderstate point size at the end if required but not modified by shader
if (shaderContext->analyzer.outputPointSize && shaderContext->analyzer.writesPointSize == false)
{
if (shader->shaderType == LatteConst::ShaderType::Vertex && shaderContext->usesGeometryShader == false)
if (shader->shaderType == LatteConst::ShaderType::Vertex && shaderContext->options->usesGeometryShader == false)
src->add("gl_PointSize = uf_pointSize;" _CRLF);
}
// end of shader main