diff options
Diffstat (limited to 'src/video_core/renderer_opengl')
5 files changed, 79 insertions, 35 deletions
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index bf6b5c3a0..546e86532 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -324,6 +324,11 @@ std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_c bool using_depth_fb) { const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs; + if (regs.rt[0].format == Tegra::RenderTargetFormat::NONE) { + LOG_ERROR(HW_GPU, "RenderTargetFormat is not configured"); + using_color_fb = false; + } + // TODO(bunnei): Implement this const bool has_stencil = false; diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp index 257aa9571..f6efce818 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp @@ -109,6 +109,9 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm, true}, // DXT45 {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm, true}, // DXN1 + {GL_COMPRESSED_RG_RGTC2, GL_RG, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm, + true}, // DXN2UNORM + {GL_COMPRESSED_SIGNED_RG_RGTC2, GL_RG, GL_INT, ComponentType::SNorm, true}, // DXN2SNORM {GL_COMPRESSED_RGBA_BPTC_UNORM_ARB, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm, true}, // BC7U {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4 @@ -218,6 +221,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr), MortonCopy<true, PixelFormat::R11FG11FB10F>, MortonCopy<true, PixelFormat::RGBA32UI>, MortonCopy<true, PixelFormat::DXT1>, MortonCopy<true, PixelFormat::DXT23>, MortonCopy<true, PixelFormat::DXT45>, MortonCopy<true, PixelFormat::DXN1>, + MortonCopy<true, PixelFormat::DXN2UNORM>, MortonCopy<true, PixelFormat::DXN2SNORM>, MortonCopy<true, PixelFormat::BC7U>, MortonCopy<true, PixelFormat::ASTC_2D_4X4>, MortonCopy<true, PixelFormat::G8R8>, MortonCopy<true, PixelFormat::BGRA8>, MortonCopy<true, PixelFormat::RGBA32F>, MortonCopy<true, PixelFormat::RG32F>, @@ -242,7 +246,10 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr), MortonCopy<false, PixelFormat::RGBA16F>, MortonCopy<false, PixelFormat::R11FG11FB10F>, MortonCopy<false, PixelFormat::RGBA32UI>, - // TODO(Subv): Swizzling DXT1/DXT23/DXT45/DXN1/BC7U/ASTC_2D_4X4 formats is not supported + // TODO(Subv): Swizzling DXT1/DXT23/DXT45/DXN1/DXN2/BC7U/ASTC_2D_4X4 formats is not + // supported + nullptr, + nullptr, nullptr, nullptr, nullptr, @@ -447,22 +454,24 @@ MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64 void CachedSurface::LoadGLBuffer() { ASSERT(params.type != SurfaceType::Fill); - u8* const texture_src_data = Memory::GetPointer(params.GetCpuAddr()); + const u8* const texture_src_data = Memory::GetPointer(params.GetCpuAddr()); ASSERT(texture_src_data); - gl_buffer.resize(params.width * params.height * GetGLBytesPerPixel(params.pixel_format)); + const u32 bytes_per_pixel = GetGLBytesPerPixel(params.pixel_format); + const u32 copy_size = params.width * params.height * bytes_per_pixel; MICROPROFILE_SCOPE(OpenGL_SurfaceLoad); - if (!params.is_tiled) { - const u32 bytes_per_pixel{params.GetFormatBpp() >> 3}; + if (params.is_tiled) { + gl_buffer.resize(copy_size); - std::memcpy(gl_buffer.data(), texture_src_data, - bytes_per_pixel * params.width * params.height); - } else { morton_to_gl_fns[static_cast<size_t>(params.pixel_format)]( params.width, params.block_height, params.height, gl_buffer.data(), params.addr); + } else { + const u8* const texture_src_data_end = texture_src_data + copy_size; + + gl_buffer.assign(texture_src_data, texture_src_data_end); } ConvertFormatAsNeeded_LoadGLBuffer(gl_buffer, params.pixel_format, params.width, params.height); @@ -757,10 +766,12 @@ void RasterizerCacheOpenGL::FlushRegion(Tegra::GPUVAddr /*addr*/, size_t /*size* } void RasterizerCacheOpenGL::InvalidateRegion(Tegra::GPUVAddr addr, size_t size) { - for (const auto& pair : surface_cache) { - const auto& surface{pair.second}; + for (auto iter = surface_cache.cbegin(); iter != surface_cache.cend();) { + const auto& surface{iter->second}; const auto& params{surface->GetSurfaceParams()}; + ++iter; + if (params.IsOverlappingRegion(addr, size)) { UnregisterSurface(surface); } diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h index 0c6652c7a..26e2ee203 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h @@ -35,31 +35,33 @@ struct SurfaceParams { DXT23 = 9, DXT45 = 10, DXN1 = 11, // This is also known as BC4 - BC7U = 12, - ASTC_2D_4X4 = 13, - G8R8 = 14, - BGRA8 = 15, - RGBA32F = 16, - RG32F = 17, - R32F = 18, - R16F = 19, - R16UNORM = 20, - RG16 = 21, - RG16F = 22, - RG16UI = 23, - RG16I = 24, - RG16S = 25, - RGB32F = 26, - SRGBA8 = 27, + DXN2UNORM = 12, + DXN2SNORM = 13, + BC7U = 14, + ASTC_2D_4X4 = 15, + G8R8 = 16, + BGRA8 = 17, + RGBA32F = 18, + RG32F = 19, + R32F = 20, + R16F = 21, + R16UNORM = 22, + RG16 = 23, + RG16F = 24, + RG16UI = 25, + RG16I = 26, + RG16S = 27, + RGB32F = 28, + SRGBA8 = 29, MaxColorFormat, // DepthStencil formats - Z24S8 = 28, - S8Z24 = 29, - Z32F = 30, - Z16 = 31, - Z32FS8 = 32, + Z24S8 = 30, + S8Z24 = 31, + Z32F = 32, + Z16 = 33, + Z32FS8 = 34, MaxDepthStencilFormat, @@ -109,6 +111,8 @@ struct SurfaceParams { 4, // DXT23 4, // DXT45 4, // DXN1 + 4, // DXN2UNORM + 4, // DXN2SNORM 4, // BC7U 4, // ASTC_2D_4X4 1, // G8R8 @@ -153,6 +157,8 @@ struct SurfaceParams { 128, // DXT23 128, // DXT45 64, // DXN1 + 128, // DXN2UNORM + 128, // DXN2SNORM 128, // BC7U 32, // ASTC_2D_4X4 16, // G8R8 @@ -221,6 +227,8 @@ struct SurfaceParams { return PixelFormat::RG32F; case Tegra::RenderTargetFormat::R11G11B10_FLOAT: return PixelFormat::R11FG11FB10F; + case Tegra::RenderTargetFormat::B5G6R5_UNORM: + return PixelFormat::B5G6R5; case Tegra::RenderTargetFormat::RGBA32_UINT: return PixelFormat::RGBA32UI; case Tegra::RenderTargetFormat::R8_UNORM: @@ -303,6 +311,16 @@ struct SurfaceParams { return PixelFormat::DXT45; case Tegra::Texture::TextureFormat::DXN1: return PixelFormat::DXN1; + case Tegra::Texture::TextureFormat::DXN2: + switch (component_type) { + case Tegra::Texture::ComponentType::UNORM: + return PixelFormat::DXN2UNORM; + case Tegra::Texture::ComponentType::SNORM: + return PixelFormat::DXN2SNORM; + } + LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", + static_cast<u32>(component_type)); + UNREACHABLE(); case Tegra::Texture::TextureFormat::BC7U: return PixelFormat::BC7U; case Tegra::Texture::TextureFormat::ASTC_2D_4X4: @@ -360,6 +378,9 @@ struct SurfaceParams { return Tegra::Texture::TextureFormat::DXT45; case PixelFormat::DXN1: return Tegra::Texture::TextureFormat::DXN1; + case PixelFormat::DXN2UNORM: + case PixelFormat::DXN2SNORM: + return Tegra::Texture::TextureFormat::DXN2; case PixelFormat::BC7U: return Tegra::Texture::TextureFormat::BC7U; case PixelFormat::ASTC_2D_4X4: @@ -441,6 +462,7 @@ struct SurfaceParams { case Tegra::RenderTargetFormat::RGB10_A2_UNORM: case Tegra::RenderTargetFormat::R8_UNORM: case Tegra::RenderTargetFormat::RG16_UNORM: + case Tegra::RenderTargetFormat::B5G6R5_UNORM: return ComponentType::UNorm; case Tegra::RenderTargetFormat::RG16_SNORM: return ComponentType::SNorm; diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp index dd240a4ce..32f06f409 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -507,6 +507,8 @@ private: /// Build the GLSL register list. void BuildRegisterList() { + regs.reserve(Register::NumRegisters); + for (size_t index = 0; index < Register::NumRegisters; ++index) { regs.emplace_back(index, suffix); } @@ -657,16 +659,17 @@ private: * @param instr Instruction to generate the if condition for. * @returns string containing the predicate condition. */ - std::string GetPredicateCondition(u64 index, bool negate) const { + std::string GetPredicateCondition(u64 index, bool negate) { using Tegra::Shader::Pred; std::string variable; // Index 7 is used as an 'Always True' condition. - if (index == static_cast<u64>(Pred::UnusedIndex)) + if (index == static_cast<u64>(Pred::UnusedIndex)) { variable = "true"; - else + } else { variable = 'p' + std::to_string(index) + '_' + suffix; - + declr_predicates.insert(variable); + } if (negate) { return "!(" + variable + ')'; } diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h index 500d4d4b1..43be69dd1 100644 --- a/src/video_core/renderer_opengl/maxwell_to_gl.h +++ b/src/video_core/renderer_opengl/maxwell_to_gl.h @@ -31,6 +31,7 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) { case Maxwell::VertexAttribute::Size::Size_8_8_8_8: return GL_UNSIGNED_BYTE; case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_16_16_16_16: return GL_UNSIGNED_SHORT; case Maxwell::VertexAttribute::Size::Size_10_10_10_2: return GL_UNSIGNED_INT_2_10_10_10_REV; @@ -85,6 +86,8 @@ inline GLenum IndexFormat(Maxwell::IndexFormat index_format) { inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) { switch (topology) { + case Maxwell::PrimitiveTopology::Points: + return GL_POINTS; case Maxwell::PrimitiveTopology::Triangles: return GL_TRIANGLES; case Maxwell::PrimitiveTopology::TriangleStrip: |