ShuriZma
/
duckstation
mirror of https://github.com/stenzek/duckstation
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

GPU/HW: Clear ROV depth on fill/copy/write 

It wasn't being specified before, whoops.
This commit is contained in:
Stenzek 2024-11-27 17:10:53 +10:00
parent eb390a9b5d
commit 97700b85de
No known key found for this signature in database
3 changed files with 35 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
src/core/gpu_hw.cpp
View File

@ -1467,7 +1467,7 @@ bool GPU_HW::CompilePipelines(Error* error)
std::unique_ptr<GPUShader> fs = g_gpu_device->CreateShader( std::unique_ptr<GPUShader> fs = g_gpu_device->CreateShader(
GPUShaderStage::Fragment, shadergen.GetLanguage(), GPUShaderStage::Fragment, shadergen.GetLanguage(),
shadergen.GenerateVRAMFillFragmentShader(ConvertToBoolUnchecked(wrapped), ConvertToBoolUnchecked(interlaced), shadergen.GenerateVRAMFillFragmentShader(ConvertToBoolUnchecked(wrapped), ConvertToBoolUnchecked(interlaced),
m_write_mask_as_depth), m_write_mask_as_depth, needs_rov_depth),
error); error);
if (!fs) if (!fs)
return false; return false;
@ -1485,9 +1485,9 @@ bool GPU_HW::CompilePipelines(Error* error)
// VRAM copy // VRAM copy
{ {
std::unique_ptr<GPUShader> fs = std::unique_ptr<GPUShader> fs = g_gpu_device->CreateShader(
g_gpu_device->CreateShader(GPUShaderStage::Fragment, shadergen.GetLanguage(), GPUShaderStage::Fragment, shadergen.GetLanguage(),
shadergen.GenerateVRAMCopyFragmentShader(m_write_mask_as_depth), error); shadergen.GenerateVRAMCopyFragmentShader(m_write_mask_as_depth, needs_rov_depth), error);
if (!fs) if (!fs)
return false; return false;
@ -1516,7 +1516,7 @@ bool GPU_HW::CompilePipelines(Error* error)
const bool use_ssbo = features.texture_buffers_emulated_with_ssbo; const bool use_ssbo = features.texture_buffers_emulated_with_ssbo;
std::unique_ptr<GPUShader> fs = g_gpu_device->CreateShader( std::unique_ptr<GPUShader> fs = g_gpu_device->CreateShader(
GPUShaderStage::Fragment, shadergen.GetLanguage(), GPUShaderStage::Fragment, shadergen.GetLanguage(),
shadergen.GenerateVRAMWriteFragmentShader(use_buffer, use_ssbo, m_write_mask_as_depth), error); shadergen.GenerateVRAMWriteFragmentShader(use_buffer, use_ssbo, m_write_mask_as_depth, needs_rov_depth), error);
if (!fs) if (!fs)
return false; return false;

33
src/core/gpu_hw_shadergen.cpp
View File

@ -1453,14 +1453,17 @@ uint SampleVRAM(uint2 coords)
return ss.str(); return ss.str();
} }
std::string GPU_HW_ShaderGen::GenerateVRAMWriteFragmentShader(bool use_buffer, bool use_ssbo, std::string GPU_HW_ShaderGen::GenerateVRAMWriteFragmentShader(bool use_buffer, bool use_ssbo, bool write_mask_as_depth,
bool write_mask_as_depth) const bool write_depth_as_rt) const
{ {
Assert(!write_mask_as_depth || (write_mask_as_depth != write_depth_as_rt));
std::stringstream ss; std::stringstream ss;
WriteHeader(ss); WriteHeader(ss);
WriteColorConversionFunctions(ss); WriteColorConversionFunctions(ss);
DefineMacro(ss, "WRITE_MASK_AS_DEPTH", write_mask_as_depth); DefineMacro(ss, "WRITE_MASK_AS_DEPTH", write_mask_as_depth);
DefineMacro(ss, "WRITE_DEPTH_AS_RT", write_depth_as_rt);
DefineMacro(ss, "USE_BUFFER", use_buffer); DefineMacro(ss, "USE_BUFFER", use_buffer);
ss << "CONSTANT float2 VRAM_SIZE = float2(" << VRAM_WIDTH << ".0, " << VRAM_HEIGHT << ".0);\n"; ss << "CONSTANT float2 VRAM_SIZE = float2(" << VRAM_WIDTH << ".0, " << VRAM_HEIGHT << ".0);\n";
@ -1496,7 +1499,7 @@ std::string GPU_HW_ShaderGen::GenerateVRAMWriteFragmentShader(bool use_buffer, b
ss << "#define GET_VALUE(buffer_offset) (LOAD_TEXTURE_BUFFER(samp0, int(buffer_offset)).r)\n\n"; ss << "#define GET_VALUE(buffer_offset) (LOAD_TEXTURE_BUFFER(samp0, int(buffer_offset)).r)\n\n";
} }
DeclareFragmentEntryPoint(ss, 0, 1, {}, true, 1, false, write_mask_as_depth); DeclareFragmentEntryPoint(ss, 0, 1, {}, true, 1 + BoolToUInt32(write_depth_as_rt), false, write_mask_as_depth);
ss << R"( ss << R"(
{ {
float2 coords = floor(v_pos.xy / u_resolution_scale); float2 coords = floor(v_pos.xy / u_resolution_scale);
@ -1523,20 +1526,25 @@ std::string GPU_HW_ShaderGen::GenerateVRAMWriteFragmentShader(bool use_buffer, b
o_col0 = RGBA5551ToRGBA8(value); o_col0 = RGBA5551ToRGBA8(value);
#if WRITE_MASK_AS_DEPTH #if WRITE_MASK_AS_DEPTH
o_depth = (o_col0.a == 1.0) ? u_depth_value : 0.0; o_depth = (o_col0.a == 1.0) ? u_depth_value : 0.0;
#elif WRITE_DEPTH_AS_RT
o_col1 = float4(1.0f, 0.0f, 0.0f, 0.0f);
#endif #endif
})"; })";
return ss.str(); return ss.str();
} }
std::string GPU_HW_ShaderGen::GenerateVRAMCopyFragmentShader(bool write_mask_as_depth) const std::string GPU_HW_ShaderGen::GenerateVRAMCopyFragmentShader(bool write_mask_as_depth, bool write_depth_as_rt) const
{ {
Assert(!write_mask_as_depth || (write_mask_as_depth != write_depth_as_rt));
// TODO: This won't currently work because we can't bind the texture to both the shader and framebuffer. // TODO: This won't currently work because we can't bind the texture to both the shader and framebuffer.
const bool msaa = false; const bool msaa = false;
std::stringstream ss; std::stringstream ss;
WriteHeader(ss); WriteHeader(ss);
DefineMacro(ss, "WRITE_MASK_AS_DEPTH", write_mask_as_depth); DefineMacro(ss, "WRITE_MASK_AS_DEPTH", write_mask_as_depth);
DefineMacro(ss, "WRITE_DEPTH_AS_RT", write_depth_as_rt);
DefineMacro(ss, "MSAA_COPY", msaa); DefineMacro(ss, "MSAA_COPY", msaa);
DeclareUniformBuffer(ss, DeclareUniformBuffer(ss,
@ -1545,7 +1553,8 @@ std::string GPU_HW_ShaderGen::GenerateVRAMCopyFragmentShader(bool write_mask_as_
true); true);
DeclareTexture(ss, "samp0", 0, msaa); DeclareTexture(ss, "samp0", 0, msaa);
DeclareFragmentEntryPoint(ss, 0, 1, {}, true, 1, false, write_mask_as_depth, false, false, msaa); DeclareFragmentEntryPoint(ss, 0, 1, {}, true, 1 + BoolToUInt32(write_depth_as_rt), false, write_mask_as_depth, false,
false, msaa);
ss << R"( ss << R"(
{ {
float2 dst_coords = floor(v_pos.xy); float2 dst_coords = floor(v_pos.xy);
@ -1575,25 +1584,31 @@ std::string GPU_HW_ShaderGen::GenerateVRAMCopyFragmentShader(bool write_mask_as_
o_col0 = float4(color.xyz, u_set_mask_bit ? 1.0 : color.a); o_col0 = float4(color.xyz, u_set_mask_bit ? 1.0 : color.a);
#if WRITE_MASK_AS_DEPTH #if WRITE_MASK_AS_DEPTH
o_depth = (u_set_mask_bit ? 1.0f : ((o_col0.a == 1.0) ? u_depth_value : 0.0)); o_depth = (u_set_mask_bit ? 1.0f : ((o_col0.a == 1.0) ? u_depth_value : 0.0));
#elif WRITE_DEPTH_AS_RT
o_col1 = float4(1.0f, 0.0f, 0.0f, 0.0f);
#endif #endif
})"; })";
return ss.str(); return ss.str();
} }
std::string GPU_HW_ShaderGen::GenerateVRAMFillFragmentShader(bool wrapped, bool interlaced, std::string GPU_HW_ShaderGen::GenerateVRAMFillFragmentShader(bool wrapped, bool interlaced, bool write_mask_as_depth,
bool write_mask_as_depth) const bool write_depth_as_rt) const
{ {
Assert(!write_mask_as_depth || (write_mask_as_depth != write_depth_as_rt));
std::stringstream ss; std::stringstream ss;
WriteHeader(ss); WriteHeader(ss);
DefineMacro(ss, "WRITE_MASK_AS_DEPTH", write_mask_as_depth); DefineMacro(ss, "WRITE_MASK_AS_DEPTH", write_mask_as_depth);
DefineMacro(ss, "WRITE_DEPTH_AS_RT", write_depth_as_rt);
DefineMacro(ss, "WRAPPED", wrapped); DefineMacro(ss, "WRAPPED", wrapped);
DefineMacro(ss, "INTERLACED", interlaced); DefineMacro(ss, "INTERLACED", interlaced);
DeclareUniformBuffer( DeclareUniformBuffer(
ss, {"uint2 u_dst_coords", "uint2 u_end_coords", "float4 u_fill_color", "uint u_interlaced_displayed_field"}, true); ss, {"uint2 u_dst_coords", "uint2 u_end_coords", "float4 u_fill_color", "uint u_interlaced_displayed_field"}, true);
DeclareFragmentEntryPoint(ss, 0, 1, {}, interlaced || wrapped, 1, false, write_mask_as_depth, false, false, false); DeclareFragmentEntryPoint(ss, 0, 1, {}, interlaced || wrapped, 1 + BoolToUInt32(write_depth_as_rt), false,
write_mask_as_depth, false, false, false);
ss << R"( ss << R"(
{ {
#if INTERLACED || WRAPPED #if INTERLACED || WRAPPED
@ -1617,6 +1632,8 @@ std::string GPU_HW_ShaderGen::GenerateVRAMFillFragmentShader(bool wrapped, bool
o_col0 = u_fill_color; o_col0 = u_fill_color;
#if WRITE_MASK_AS_DEPTH #if WRITE_MASK_AS_DEPTH
o_depth = u_fill_color.a; o_depth = u_fill_color.a;
#elif WRITE_DEPTH_AS_RT
o_col1 = float4(1.0f, 0.0f, 0.0f, 0.0f);
#endif #endif
})"; })";

8
src/core/gpu_hw_shadergen.h
View File

@ -26,9 +26,11 @@ public:
std::string GenerateWireframeGeometryShader() const; std::string GenerateWireframeGeometryShader() const;
std::string GenerateWireframeFragmentShader() const; std::string GenerateWireframeFragmentShader() const;
std::string GenerateVRAMReadFragmentShader(u32 resolution_scale, u32 multisamples) const; std::string GenerateVRAMReadFragmentShader(u32 resolution_scale, u32 multisamples) const;
std::string GenerateVRAMWriteFragmentShader(bool use_buffer, bool use_ssbo, bool write_mask_as_depth) const; std::string GenerateVRAMWriteFragmentShader(bool use_buffer, bool use_ssbo, bool write_mask_as_depth,
std::string GenerateVRAMCopyFragmentShader(bool write_mask_as_depth) const; bool write_depth_as_rt) const;
std::string GenerateVRAMFillFragmentShader(bool wrapped, bool interlaced, bool write_mask_as_depth) const; std::string GenerateVRAMCopyFragmentShader(bool write_mask_as_depth, bool write_depth_as_rt) const;
std::string GenerateVRAMFillFragmentShader(bool wrapped, bool interlaced, bool write_mask_as_depth,
bool write_depth_as_rt) const;
std::string GenerateVRAMUpdateDepthFragmentShader(bool msaa) const; std::string GenerateVRAMUpdateDepthFragmentShader(bool msaa) const;
std::string GenerateVRAMExtractFragmentShader(u32 resolution_scale, u32 multisamples, bool color_24bit, std::string GenerateVRAMExtractFragmentShader(u32 resolution_scale, u32 multisamples, bool color_24bit,
bool depth_buffer) const; bool depth_buffer) const;