diff options
Diffstat (limited to 'src')
-rw-r--r-- | src/video_core/engines/shader_bytecode.h | 2 | ||||
-rw-r--r-- | src/video_core/macro_interpreter.cpp | 4 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_device.cpp | 2 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_device.h | 5 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_shader_cache.cpp | 7 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_shader_decompiler.cpp | 105 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_shader_decompiler.h | 1 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_shader_disk_cache.cpp | 10 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_state.cpp | 4 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_texture_cache.cpp | 3 | ||||
-rw-r--r-- | src/video_core/renderer_vulkan/vk_shader_decompiler.cpp | 14 | ||||
-rw-r--r-- | src/video_core/shader/shader_ir.cpp | 16 | ||||
-rw-r--r-- | src/video_core/shader/shader_ir.h | 15 | ||||
-rw-r--r-- | src/video_core/texture_cache/surface_base.cpp | 9 | ||||
-rw-r--r-- | src/video_core/texture_cache/surface_base.h | 10 | ||||
-rw-r--r-- | src/video_core/texture_cache/surface_params.cpp | 13 | ||||
-rw-r--r-- | src/video_core/texture_cache/texture_cache.h | 37 |
17 files changed, 195 insertions, 62 deletions
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h index 4aaad2979..79d469b88 100644 --- a/src/video_core/engines/shader_bytecode.h +++ b/src/video_core/engines/shader_bytecode.h @@ -78,7 +78,7 @@ union Attribute { constexpr explicit Attribute(u64 value) : value(value) {} enum class Index : u64 { - PointSize = 6, + LayerViewportPointSize = 6, Position = 7, Attribute_0 = 8, Attribute_31 = 39, diff --git a/src/video_core/macro_interpreter.cpp b/src/video_core/macro_interpreter.cpp index c766ed692..9f59a2dc1 100644 --- a/src/video_core/macro_interpreter.cpp +++ b/src/video_core/macro_interpreter.cpp @@ -4,14 +4,18 @@ #include "common/assert.h" #include "common/logging/log.h" +#include "common/microprofile.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/macro_interpreter.h" +MICROPROFILE_DEFINE(MacroInterp, "GPU", "Execute macro interpreter", MP_RGB(128, 128, 192)); + namespace Tegra { MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {} void MacroInterpreter::Execute(u32 offset, std::vector<u32> parameters) { + MICROPROFILE_SCOPE(MacroInterp); Reset(); registers[1] = parameters[0]; this->parameters = std::move(parameters); diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp index 6238ddaaa..85424a4c9 100644 --- a/src/video_core/renderer_opengl/gl_device.cpp +++ b/src/video_core/renderer_opengl/gl_device.cpp @@ -27,6 +27,7 @@ Device::Device() { shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT); max_vertex_attributes = GetInteger<u32>(GL_MAX_VERTEX_ATTRIBS); max_varyings = GetInteger<u32>(GL_MAX_VARYING_VECTORS); + has_vertex_viewport_layer = GLAD_GL_ARB_shader_viewport_layer_array; has_variable_aoffi = TestVariableAoffi(); has_component_indexing_bug = TestComponentIndexingBug(); } @@ -35,6 +36,7 @@ Device::Device(std::nullptr_t) { uniform_buffer_alignment = 0; max_vertex_attributes = 16; max_varyings = 15; + has_vertex_viewport_layer = true; has_variable_aoffi = true; has_component_indexing_bug = false; } diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h index 939edb440..dc883722d 100644 --- a/src/video_core/renderer_opengl/gl_device.h +++ b/src/video_core/renderer_opengl/gl_device.h @@ -30,6 +30,10 @@ public: return max_varyings; } + bool HasVertexViewportLayer() const { + return has_vertex_viewport_layer; + } + bool HasVariableAoffi() const { return has_variable_aoffi; } @@ -46,6 +50,7 @@ private: std::size_t shader_storage_alignment{}; u32 max_vertex_attributes{}; u32 max_varyings{}; + bool has_vertex_viewport_layer{}; bool has_variable_aoffi{}; bool has_component_indexing_bug{}; }; diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index 5d76ee12d..32dd9eae7 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -190,8 +190,11 @@ CachedProgram SpecializeShader(const std::string& code, const GLShader::ShaderEn const auto texture_buffer_usage{variant.texture_buffer_usage}; std::string source = "#version 430 core\n" - "#extension GL_ARB_separate_shader_objects : enable\n\n"; - source += fmt::format("#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++); + "#extension GL_ARB_separate_shader_objects : enable\n"; + if (entries.shader_viewport_layer_array) { + source += "#extension GL_ARB_shader_viewport_layer_array : enable\n"; + } + source += fmt::format("\n#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++); for (const auto& cbuf : entries.const_buffers) { source += diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp index 47cc2011f..119073776 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -14,6 +14,7 @@ #include "common/alignment.h" #include "common/assert.h" #include "common/common_types.h" +#include "common/logging/log.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_rasterizer.h" @@ -246,6 +247,8 @@ public: usage.is_read, usage.is_written); } entries.clip_distances = ir.GetClipDistances(); + entries.shader_viewport_layer_array = + stage == ShaderStage::Vertex && (ir.UsesLayer() || ir.UsesViewportIndex()); entries.shader_length = ir.GetLength(); return entries; } @@ -282,22 +285,35 @@ private: } void DeclareVertexRedeclarations() { - bool clip_distances_declared = false; - code.AddLine("out gl_PerVertex {{"); ++code.scope; code.AddLine("vec4 gl_Position;"); - for (const auto o : ir.GetOutputAttributes()) { - if (o == Attribute::Index::PointSize) - code.AddLine("float gl_PointSize;"); - if (!clip_distances_declared && (o == Attribute::Index::ClipDistances0123 || - o == Attribute::Index::ClipDistances4567)) { + for (const auto attribute : ir.GetOutputAttributes()) { + if (attribute == Attribute::Index::ClipDistances0123 || + attribute == Attribute::Index::ClipDistances4567) { code.AddLine("float gl_ClipDistance[];"); - clip_distances_declared = true; + break; } } + if (stage != ShaderStage::Vertex || device.HasVertexViewportLayer()) { + if (ir.UsesLayer()) { + code.AddLine("int gl_Layer;"); + } + if (ir.UsesViewportIndex()) { + code.AddLine("int gl_ViewportIndex;"); + } + } else if ((ir.UsesLayer() || ir.UsesViewportIndex()) && stage == ShaderStage::Vertex && + !device.HasVertexViewportLayer()) { + LOG_ERROR( + Render_OpenGL, + "GL_ARB_shader_viewport_layer_array is not available and its required by a shader"); + } + + if (ir.UsesPointSize()) { + code.AddLine("float gl_PointSize;"); + } --code.scope; code.AddLine("}};"); @@ -805,6 +821,45 @@ private: return CastOperand(VisitOperand(operation, operand_index), type); } + std::optional<std::pair<std::string, bool>> GetOutputAttribute(const AbufNode* abuf) { + switch (const auto attribute = abuf->GetIndex()) { + case Attribute::Index::Position: + return std::make_pair("gl_Position"s + GetSwizzle(abuf->GetElement()), false); + case Attribute::Index::LayerViewportPointSize: + switch (abuf->GetElement()) { + case 0: + UNIMPLEMENTED(); + return {}; + case 1: + if (stage == ShaderStage::Vertex && !device.HasVertexViewportLayer()) { + return {}; + } + return std::make_pair("gl_Layer", true); + case 2: + if (stage == ShaderStage::Vertex && !device.HasVertexViewportLayer()) { + return {}; + } + return std::make_pair("gl_ViewportIndex", true); + case 3: + UNIMPLEMENTED_MSG("Requires some state changes for gl_PointSize to work in shader"); + return std::make_pair("gl_PointSize", false); + } + return {}; + case Attribute::Index::ClipDistances0123: + return std::make_pair(fmt::format("gl_ClipDistance[{}]", abuf->GetElement()), false); + case Attribute::Index::ClipDistances4567: + return std::make_pair(fmt::format("gl_ClipDistance[{}]", abuf->GetElement() + 4), + false); + default: + if (IsGenericAttribute(attribute)) { + return std::make_pair( + GetOutputAttribute(attribute) + GetSwizzle(abuf->GetElement()), false); + } + UNIMPLEMENTED_MSG("Unhandled output attribute: {}", static_cast<u32>(attribute)); + return {}; + } + } + std::string CastOperand(const std::string& value, Type type) const { switch (type) { case Type::Bool: @@ -1001,6 +1056,8 @@ private: const Node& src = operation[1]; std::string target; + bool is_integer = false; + if (const auto gpr = std::get_if<GprNode>(&*dest)) { if (gpr->GetIndex() == Register::ZeroIndex) { // Writing to Register::ZeroIndex is a no op @@ -1009,26 +1066,12 @@ private: target = GetRegister(gpr->GetIndex()); } else if (const auto abuf = std::get_if<AbufNode>(&*dest)) { UNIMPLEMENTED_IF(abuf->IsPhysicalBuffer()); - - target = [&]() -> std::string { - switch (const auto attribute = abuf->GetIndex(); abuf->GetIndex()) { - case Attribute::Index::Position: - return "gl_Position"s + GetSwizzle(abuf->GetElement()); - case Attribute::Index::PointSize: - return "gl_PointSize"; - case Attribute::Index::ClipDistances0123: - return fmt::format("gl_ClipDistance[{}]", abuf->GetElement()); - case Attribute::Index::ClipDistances4567: - return fmt::format("gl_ClipDistance[{}]", abuf->GetElement() + 4); - default: - if (IsGenericAttribute(attribute)) { - return GetOutputAttribute(attribute) + GetSwizzle(abuf->GetElement()); - } - UNIMPLEMENTED_MSG("Unhandled output attribute: {}", - static_cast<u32>(attribute)); - return "0"; - } - }(); + const auto result = GetOutputAttribute(abuf); + if (!result) { + return {}; + } + target = result->first; + is_integer = result->second; } else if (const auto lmem = std::get_if<LmemNode>(&*dest)) { target = fmt::format("{}[ftou({}) / 4]", GetLocalMemory(), Visit(lmem->GetAddress())); } else if (const auto gmem = std::get_if<GmemNode>(&*dest)) { @@ -1040,7 +1083,11 @@ private: UNREACHABLE_MSG("Assign called without a proper target"); } - code.AddLine("{} = {};", target, Visit(src)); + if (is_integer) { + code.AddLine("{} = ftoi({});", target, Visit(src)); + } else { + code.AddLine("{} = {};", target, Visit(src)); + } return {}; } diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h index 14d11c7fc..02586736d 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.h +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h @@ -78,6 +78,7 @@ struct ShaderEntries { std::vector<ImageEntry> images; std::vector<GlobalMemoryEntry> global_memory_entries; std::array<bool, Maxwell::NumClipDistances> clip_distances{}; + bool shader_viewport_layer_array{}; std::size_t shader_length{}; }; diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp index 10688397b..7893d1e26 100644 --- a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp @@ -373,6 +373,12 @@ std::optional<ShaderDiskCacheDecompiled> ShaderDiskCacheOpenGL::LoadDecompiledEn } } + bool shader_viewport_layer_array{}; + if (!LoadObjectFromPrecompiled(shader_viewport_layer_array)) { + return {}; + } + entry.entries.shader_viewport_layer_array = shader_viewport_layer_array; + u64 shader_length{}; if (!LoadObjectFromPrecompiled(shader_length)) { return {}; @@ -445,6 +451,10 @@ bool ShaderDiskCacheOpenGL::SaveDecompiledFile(u64 unique_identifier, const std: } } + if (!SaveObjectToPrecompiled(entries.shader_viewport_layer_array)) { + return false; + } + if (!SaveObjectToPrecompiled(static_cast<u64>(entries.shader_length))) { return false; } diff --git a/src/video_core/renderer_opengl/gl_state.cpp b/src/video_core/renderer_opengl/gl_state.cpp index d86e137ac..0eae98afe 100644 --- a/src/video_core/renderer_opengl/gl_state.cpp +++ b/src/video_core/renderer_opengl/gl_state.cpp @@ -6,8 +6,11 @@ #include <glad/glad.h> #include "common/assert.h" #include "common/logging/log.h" +#include "common/microprofile.h" #include "video_core/renderer_opengl/gl_state.h" +MICROPROFILE_DEFINE(OpenGL_State, "OpenGL", "State Change", MP_RGB(192, 128, 128)); + namespace OpenGL { using Maxwell = Tegra::Engines::Maxwell3D::Regs; @@ -524,6 +527,7 @@ void OpenGLState::ApplySamplers() const { } void OpenGLState::Apply() const { + MICROPROFILE_SCOPE(OpenGL_State); ApplyFramebufferState(); ApplyVertexArrayState(); ApplyShaderProgram(); diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index 08ae1a429..b1f6bc7c2 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -31,6 +31,8 @@ using VideoCore::Surface::SurfaceType; MICROPROFILE_DEFINE(OpenGL_Texture_Upload, "OpenGL", "Texture Upload", MP_RGB(128, 192, 128)); MICROPROFILE_DEFINE(OpenGL_Texture_Download, "OpenGL", "Texture Download", MP_RGB(128, 192, 128)); +MICROPROFILE_DEFINE(OpenGL_Texture_Buffer_Copy, "OpenGL", "Texture Buffer Copy", + MP_RGB(128, 192, 128)); namespace { @@ -535,6 +537,7 @@ void TextureCacheOpenGL::ImageBlit(View& src_view, View& dst_view, } void TextureCacheOpenGL::BufferCopy(Surface& src_surface, Surface& dst_surface) { + MICROPROFILE_SCOPE(OpenGL_Texture_Buffer_Copy); const auto& src_params = src_surface->GetSurfaceParams(); const auto& dst_params = dst_surface->GetSurfaceParams(); UNIMPLEMENTED_IF(src_params.num_levels > 1 || dst_params.num_levels > 1); diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp index 1bb04607b..9b2d8e987 100644 --- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp +++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp @@ -430,20 +430,17 @@ private: instance_index = DeclareBuiltIn(spv::BuiltIn::InstanceIndex, spv::StorageClass::Input, t_in_uint, "instance_index"); - bool is_point_size_declared = false; bool is_clip_distances_declared = false; for (const auto index : ir.GetOutputAttributes()) { - if (index == Attribute::Index::PointSize) { - is_point_size_declared = true; - } else if (index == Attribute::Index::ClipDistances0123 || - index == Attribute::Index::ClipDistances4567) { + if (index == Attribute::Index::ClipDistances0123 || + index == Attribute::Index::ClipDistances4567) { is_clip_distances_declared = true; } } std::vector<Id> members; members.push_back(t_float4); - if (is_point_size_declared) { + if (ir.UsesPointSize()) { members.push_back(t_float); } if (is_clip_distances_declared) { @@ -466,7 +463,7 @@ private: position_index = MemberDecorateBuiltIn(spv::BuiltIn::Position, "position", true); point_size_index = - MemberDecorateBuiltIn(spv::BuiltIn::PointSize, "point_size", is_point_size_declared); + MemberDecorateBuiltIn(spv::BuiltIn::PointSize, "point_size", ir.UsesPointSize()); clip_distances_index = MemberDecorateBuiltIn(spv::BuiltIn::ClipDistance, "clip_distances", is_clip_distances_declared); @@ -712,7 +709,8 @@ private: case Attribute::Index::Position: return AccessElement(t_out_float, per_vertex, position_index, abuf->GetElement()); - case Attribute::Index::PointSize: + case Attribute::Index::LayerViewportPointSize: + UNIMPLEMENTED_IF(abuf->GetElement() != 3); return AccessElement(t_out_float, per_vertex, point_size_index); case Attribute::Index::ClipDistances0123: return AccessElement(t_out_float, per_vertex, clip_distances_index, diff --git a/src/video_core/shader/shader_ir.cpp b/src/video_core/shader/shader_ir.cpp index 972defff3..78bd1cf1e 100644 --- a/src/video_core/shader/shader_ir.cpp +++ b/src/video_core/shader/shader_ir.cpp @@ -98,6 +98,22 @@ Node ShaderIR::GetPhysicalInputAttribute(Tegra::Shader::Register physical_addres } Node ShaderIR::GetOutputAttribute(Attribute::Index index, u64 element, Node buffer) { + if (index == Attribute::Index::LayerViewportPointSize) { + switch (element) { + case 0: + UNIMPLEMENTED(); + break; + case 1: + uses_layer = true; + break; + case 2: + uses_viewport_index = true; + break; + case 3: + uses_point_size = true; + break; + } + } if (index == Attribute::Index::ClipDistances0123 || index == Attribute::Index::ClipDistances4567) { const auto clip_index = diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h index b6faf7ad3..126c78136 100644 --- a/src/video_core/shader/shader_ir.h +++ b/src/video_core/shader/shader_ir.h @@ -115,6 +115,18 @@ public: return static_cast<std::size_t>(coverage_end * sizeof(u64)); } + bool UsesLayer() const { + return uses_layer; + } + + bool UsesViewportIndex() const { + return uses_viewport_index; + } + + bool UsesPointSize() const { + return uses_point_size; + } + bool HasPhysicalAttributes() const { return uses_physical_attributes; } @@ -346,6 +358,9 @@ private: std::set<Image> used_images; std::array<bool, Tegra::Engines::Maxwell3D::Regs::NumClipDistances> used_clip_distances{}; std::map<GlobalMemoryBase, GlobalMemoryUsage> used_global_memory; + bool uses_layer{}; + bool uses_viewport_index{}; + bool uses_point_size{}; bool uses_physical_attributes{}; // Shader uses AL2P or physical attribute read/writes Tegra::Shader::Header header; diff --git a/src/video_core/texture_cache/surface_base.cpp b/src/video_core/texture_cache/surface_base.cpp index 7a0fdb19b..6af9044ca 100644 --- a/src/video_core/texture_cache/surface_base.cpp +++ b/src/video_core/texture_cache/surface_base.cpp @@ -75,9 +75,12 @@ MatchStructureResult SurfaceBaseImpl::MatchesStructure(const SurfaceParams& rhs) // Linear Surface check if (!params.is_tiled) { - if (std::tie(params.width, params.height, params.pitch) == - std::tie(rhs.width, rhs.height, rhs.pitch)) { - return MatchStructureResult::FullMatch; + if (std::tie(params.height, params.pitch) == std::tie(rhs.height, rhs.pitch)) { + if (params.width == rhs.width) { + return MatchStructureResult::FullMatch; + } else { + return MatchStructureResult::SemiMatch; + } } return MatchStructureResult::None; } diff --git a/src/video_core/texture_cache/surface_base.h b/src/video_core/texture_cache/surface_base.h index 8ba386a8a..bcce8d863 100644 --- a/src/video_core/texture_cache/surface_base.h +++ b/src/video_core/texture_cache/surface_base.h @@ -200,8 +200,9 @@ public: modification_tick = tick; } - void MarkAsRenderTarget(const bool is_target) { + void MarkAsRenderTarget(const bool is_target, const u32 index) { this->is_target = is_target; + this->index = index; } void MarkAsPicked(const bool is_picked) { @@ -221,6 +222,10 @@ public: return is_target; } + u32 GetRenderTarget() const { + return index; + } + bool IsRegistered() const { return is_registered; } @@ -307,10 +312,13 @@ private: return view; } + static constexpr u32 NO_RT = 0xFFFFFFFF; + bool is_modified{}; bool is_target{}; bool is_registered{}; bool is_picked{}; + u32 index{NO_RT}; u64 modification_tick{}; }; diff --git a/src/video_core/texture_cache/surface_params.cpp b/src/video_core/texture_cache/surface_params.cpp index 9c56e2b4f..fd5472451 100644 --- a/src/video_core/texture_cache/surface_params.cpp +++ b/src/video_core/texture_cache/surface_params.cpp @@ -290,12 +290,19 @@ std::size_t SurfaceParams::GetLayerSize(bool as_host_size, bool uncompressed) co std::size_t SurfaceParams::GetInnerMipmapMemorySize(u32 level, bool as_host_size, bool uncompressed) const { - const bool tiled{as_host_size ? false : is_tiled}; const u32 width{GetMipmapSize(uncompressed, GetMipWidth(level), GetDefaultBlockWidth())}; const u32 height{GetMipmapSize(uncompressed, GetMipHeight(level), GetDefaultBlockHeight())}; const u32 depth{is_layered ? 1U : GetMipDepth(level)}; - return Tegra::Texture::CalculateSize(tiled, GetBytesPerPixel(), width, height, depth, - GetMipBlockHeight(level), GetMipBlockDepth(level)); + if (is_tiled) { + return Tegra::Texture::CalculateSize(!as_host_size, GetBytesPerPixel(), width, height, + depth, GetMipBlockHeight(level), + GetMipBlockDepth(level)); + } else if (as_host_size || IsBuffer()) { + return GetBytesPerPixel() * width * height * depth; + } else { + // Linear Texture Case + return pitch * height * depth; + } } bool SurfaceParams::operator==(const SurfaceParams& rhs) const { diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index c9e72531a..7f9623c62 100644 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -133,11 +133,11 @@ public: regs.zeta.memory_layout.block_depth, regs.zeta.memory_layout.type)}; auto surface_view = GetSurface(gpu_addr, depth_params, preserve_contents, true); if (depth_buffer.target) - depth_buffer.target->MarkAsRenderTarget(false); + depth_buffer.target->MarkAsRenderTarget(false, NO_RT); depth_buffer.target = surface_view.first; depth_buffer.view = surface_view.second; if (depth_buffer.target) - depth_buffer.target->MarkAsRenderTarget(true); + depth_buffer.target->MarkAsRenderTarget(true, DEPTH_RT); return surface_view.second; } @@ -167,11 +167,11 @@ public: auto surface_view = GetSurface(gpu_addr, SurfaceParams::CreateForFramebuffer(system, index), preserve_contents, true); if (render_targets[index].target) - render_targets[index].target->MarkAsRenderTarget(false); + render_targets[index].target->MarkAsRenderTarget(false, NO_RT); render_targets[index].target = surface_view.first; render_targets[index].view = surface_view.second; if (render_targets[index].target) - render_targets[index].target->MarkAsRenderTarget(true); + render_targets[index].target->MarkAsRenderTarget(true, static_cast<u32>(index)); return surface_view.second; } @@ -191,7 +191,7 @@ public: if (depth_buffer.target == nullptr) { return; } - depth_buffer.target->MarkAsRenderTarget(false); + depth_buffer.target->MarkAsRenderTarget(false, NO_RT); depth_buffer.target = nullptr; depth_buffer.view = nullptr; } @@ -200,7 +200,7 @@ public: if (render_targets[index].target == nullptr) { return; } - render_targets[index].target->MarkAsRenderTarget(false); + render_targets[index].target->MarkAsRenderTarget(false, NO_RT); render_targets[index].target = nullptr; render_targets[index].view = nullptr; } @@ -270,6 +270,16 @@ protected: // and reading it from a sepparate buffer. virtual void BufferCopy(TSurface& src_surface, TSurface& dst_surface) = 0; + void ManageRenderTargetUnregister(TSurface& surface) { + auto& maxwell3d = system.GPU().Maxwell3D(); + const u32 index = surface->GetRenderTarget(); + if (index == DEPTH_RT) { + maxwell3d.dirty_flags.zeta_buffer = true; + } else { + maxwell3d.dirty_flags.color_buffer.set(index, true); + } + } + void Register(TSurface surface) { const GPUVAddr gpu_addr = surface->GetGpuAddr(); const CacheAddr cache_ptr = ToCacheAddr(system.GPU().MemoryManager().GetPointer(gpu_addr)); @@ -294,6 +304,9 @@ protected: if (guard_render_targets && surface->IsProtected()) { return; } + if (!guard_render_targets && surface->IsRenderTarget()) { + ManageRenderTargetUnregister(surface); + } const GPUVAddr gpu_addr = surface->GetGpuAddr(); const CacheAddr cache_ptr = surface->GetCacheAddr(); const std::size_t size = surface->GetSizeInBytes(); @@ -649,15 +662,6 @@ private: } return {current_surface, *view}; } - // The next case is unsafe, so if we r in accurate GPU, just skip it - if (Settings::values.use_accurate_gpu_emulation) { - return RecycleSurface(overlaps, params, gpu_addr, preserve_contents, - MatchTopologyResult::FullMatch); - } - // This is the case the texture is a part of the parent. - if (current_surface->MatchesSubTexture(params, gpu_addr)) { - return RebuildSurface(current_surface, params, is_render); - } } else { // If there are many overlaps, odds are they are subtextures of the candidate // surface. We try to construct a new surface based on the candidate parameters, @@ -793,6 +797,9 @@ private: static constexpr u64 registry_page_size{1 << registry_page_bits}; std::unordered_map<CacheAddr, std::vector<TSurface>> registry; + static constexpr u32 DEPTH_RT = 8; + static constexpr u32 NO_RT = 0xFFFFFFFF; + // The L1 Cache is used for fast texture lookup before checking the overlaps // This avoids calculating size and other stuffs. std::unordered_map<CacheAddr, TSurface> l1_cache; |