diff options
Diffstat (limited to 'src/video_core')
| -rw-r--r-- | src/video_core/gpu.cpp | 4 | ||||
| -rw-r--r-- | src/video_core/gpu.h | 4 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/gl_rasterizer_cache.cpp | 76 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/gl_rasterizer_cache.h | 110 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/maxwell_to_gl.h | 1 |
5 files changed, 140 insertions, 55 deletions
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp index 19e7f1161..c9f6b82b7 100644 --- a/src/video_core/gpu.cpp +++ b/src/video_core/gpu.cpp @@ -46,8 +46,10 @@ u32 RenderTargetBytesPerPixel(RenderTargetFormat format) { case RenderTargetFormat::RGBA32_FLOAT: case RenderTargetFormat::RGBA32_UINT: return 16; + case RenderTargetFormat::RGBA16_UINT: case RenderTargetFormat::RGBA16_FLOAT: case RenderTargetFormat::RG32_FLOAT: + case RenderTargetFormat::RG32_UINT: return 8; case RenderTargetFormat::RGBA8_UNORM: case RenderTargetFormat::RGBA8_SNORM: @@ -61,12 +63,14 @@ u32 RenderTargetBytesPerPixel(RenderTargetFormat format) { case RenderTargetFormat::RG16_FLOAT: case RenderTargetFormat::R32_FLOAT: case RenderTargetFormat::R11G11B10_FLOAT: + case RenderTargetFormat::R32_UINT: return 4; case RenderTargetFormat::R16_UNORM: case RenderTargetFormat::R16_SNORM: case RenderTargetFormat::R16_UINT: case RenderTargetFormat::R16_SINT: case RenderTargetFormat::R16_FLOAT: + case RenderTargetFormat::RG8_UNORM: case RenderTargetFormat::RG8_SNORM: return 2; case RenderTargetFormat::R8_UNORM: diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h index e008d8f26..8a90a3a66 100644 --- a/src/video_core/gpu.h +++ b/src/video_core/gpu.h @@ -20,8 +20,10 @@ enum class RenderTargetFormat : u32 { NONE = 0x0, RGBA32_FLOAT = 0xC0, RGBA32_UINT = 0xC2, + RGBA16_UINT = 0xC9, RGBA16_FLOAT = 0xCA, RG32_FLOAT = 0xCB, + RG32_UINT = 0xCD, BGRA8_UNORM = 0xCF, RGB10_A2_UNORM = 0xD1, RGBA8_UNORM = 0xD5, @@ -33,8 +35,10 @@ enum class RenderTargetFormat : u32 { RG16_UINT = 0xDD, RG16_FLOAT = 0xDE, R11G11B10_FLOAT = 0xE0, + R32_UINT = 0xE4, R32_FLOAT = 0xE5, B5G6R5_UNORM = 0xE8, + RG8_UNORM = 0xEA, RG8_SNORM = 0xEB, R16_UNORM = 0xEE, R16_SNORM = 0xEF, diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp index 84c250c63..4b48ab8e2 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp @@ -101,6 +101,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form {GL_R8, GL_RED, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // R8 {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE, ComponentType::UInt, false}, // R8UI {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, ComponentType::Float, false}, // RGBA16F + {GL_RGBA16UI, GL_RGBA, GL_UNSIGNED_SHORT, ComponentType::UInt, false}, // RGBA16UI {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, ComponentType::Float, false}, // R11FG11FB10F {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // RGBA32UI @@ -134,7 +135,10 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form {GL_RG16_SNORM, GL_RG, GL_SHORT, ComponentType::SNorm, false}, // RG16S {GL_RGB32F, GL_RGB, GL_FLOAT, ComponentType::Float, false}, // RGB32F {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // SRGBA8 + {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // RG8U {GL_RG8, GL_RG, GL_BYTE, ComponentType::SNorm, false}, // RG8S + {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // RG32UI + {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // R32UI // DepthStencil formats {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm, @@ -234,32 +238,59 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, std::vector<u8>& gl_bu static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPUVAddr), SurfaceParams::MaxPixelFormat> morton_to_gl_fns = { - MortonCopy<true, PixelFormat::ABGR8U>, MortonCopy<true, PixelFormat::ABGR8S>, - MortonCopy<true, PixelFormat::B5G6R5>, MortonCopy<true, PixelFormat::A2B10G10R10>, - MortonCopy<true, PixelFormat::A1B5G5R5>, MortonCopy<true, PixelFormat::R8>, - MortonCopy<true, PixelFormat::R8UI>, MortonCopy<true, PixelFormat::RGBA16F>, - 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>, - MortonCopy<true, PixelFormat::R32F>, MortonCopy<true, PixelFormat::R16F>, - MortonCopy<true, PixelFormat::R16UNORM>, MortonCopy<true, PixelFormat::R16S>, - MortonCopy<true, PixelFormat::R16UI>, MortonCopy<true, PixelFormat::R16I>, - MortonCopy<true, PixelFormat::RG16>, MortonCopy<true, PixelFormat::RG16F>, - MortonCopy<true, PixelFormat::RG16UI>, MortonCopy<true, PixelFormat::RG16I>, - MortonCopy<true, PixelFormat::RG16S>, MortonCopy<true, PixelFormat::RGB32F>, - MortonCopy<true, PixelFormat::SRGBA8>, MortonCopy<true, PixelFormat::RG8S>, - MortonCopy<true, PixelFormat::Z24S8>, MortonCopy<true, PixelFormat::S8Z24>, - MortonCopy<true, PixelFormat::Z32F>, MortonCopy<true, PixelFormat::Z16>, + // clang-format off + MortonCopy<true, PixelFormat::ABGR8U>, + MortonCopy<true, PixelFormat::ABGR8S>, + MortonCopy<true, PixelFormat::B5G6R5>, + MortonCopy<true, PixelFormat::A2B10G10R10>, + MortonCopy<true, PixelFormat::A1B5G5R5>, + MortonCopy<true, PixelFormat::R8>, + MortonCopy<true, PixelFormat::R8UI>, + MortonCopy<true, PixelFormat::RGBA16F>, + MortonCopy<true, PixelFormat::RGBA16UI>, + 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>, + MortonCopy<true, PixelFormat::R32F>, + MortonCopy<true, PixelFormat::R16F>, + MortonCopy<true, PixelFormat::R16UNORM>, + MortonCopy<true, PixelFormat::R16S>, + MortonCopy<true, PixelFormat::R16UI>, + MortonCopy<true, PixelFormat::R16I>, + MortonCopy<true, PixelFormat::RG16>, + MortonCopy<true, PixelFormat::RG16F>, + MortonCopy<true, PixelFormat::RG16UI>, + MortonCopy<true, PixelFormat::RG16I>, + MortonCopy<true, PixelFormat::RG16S>, + MortonCopy<true, PixelFormat::RGB32F>, + MortonCopy<true, PixelFormat::SRGBA8>, + MortonCopy<true, PixelFormat::RG8U>, + MortonCopy<true, PixelFormat::RG8S>, + MortonCopy<true, PixelFormat::RG32UI>, + MortonCopy<true, PixelFormat::R32UI>, + MortonCopy<true, PixelFormat::Z24S8>, + MortonCopy<true, PixelFormat::S8Z24>, + MortonCopy<true, PixelFormat::Z32F>, + MortonCopy<true, PixelFormat::Z16>, MortonCopy<true, PixelFormat::Z32FS8>, + // clang-format on }; static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPUVAddr), SurfaceParams::MaxPixelFormat> gl_to_morton_fns = { + // clang-format off MortonCopy<false, PixelFormat::ABGR8U>, MortonCopy<false, PixelFormat::ABGR8S>, MortonCopy<false, PixelFormat::B5G6R5>, @@ -268,6 +299,7 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPU MortonCopy<false, PixelFormat::R8>, MortonCopy<false, PixelFormat::R8UI>, MortonCopy<false, PixelFormat::RGBA16F>, + MortonCopy<false, PixelFormat::RGBA16UI>, MortonCopy<false, PixelFormat::R11FG11FB10F>, MortonCopy<false, PixelFormat::RGBA32UI>, // TODO(Subv): Swizzling DXT1/DXT23/DXT45/DXN1/DXN2/BC7U/ASTC_2D_4X4 formats is not @@ -297,12 +329,16 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPU MortonCopy<false, PixelFormat::RG16S>, MortonCopy<false, PixelFormat::RGB32F>, MortonCopy<false, PixelFormat::SRGBA8>, + MortonCopy<false, PixelFormat::RG8U>, MortonCopy<false, PixelFormat::RG8S>, + MortonCopy<false, PixelFormat::RG32UI>, + MortonCopy<false, PixelFormat::R32UI>, MortonCopy<false, PixelFormat::Z24S8>, MortonCopy<false, PixelFormat::S8Z24>, MortonCopy<false, PixelFormat::Z32F>, MortonCopy<false, PixelFormat::Z16>, MortonCopy<false, PixelFormat::Z32FS8>, + // clang-format on }; // Allocate an uninitialized texture of appropriate size and format for the surface diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h index 202257b58..630b40e77 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h @@ -31,43 +31,47 @@ struct SurfaceParams { R8 = 5, R8UI = 6, RGBA16F = 7, - R11FG11FB10F = 8, - RGBA32UI = 9, - DXT1 = 10, - DXT23 = 11, - DXT45 = 12, - DXN1 = 13, // This is also known as BC4 - DXN2UNORM = 14, - DXN2SNORM = 15, - BC7U = 16, - ASTC_2D_4X4 = 17, - G8R8 = 18, - BGRA8 = 19, - RGBA32F = 20, - RG32F = 21, - R32F = 22, - R16F = 23, - R16UNORM = 24, - R16S = 25, - R16UI = 26, - R16I = 27, - RG16 = 28, - RG16F = 29, - RG16UI = 30, - RG16I = 31, - RG16S = 32, - RGB32F = 33, - SRGBA8 = 34, - RG8S = 35, + RGBA16UI = 8, + R11FG11FB10F = 9, + RGBA32UI = 10, + DXT1 = 11, + DXT23 = 12, + DXT45 = 13, + DXN1 = 14, // This is also known as BC4 + DXN2UNORM = 15, + DXN2SNORM = 16, + BC7U = 17, + ASTC_2D_4X4 = 18, + G8R8 = 19, + BGRA8 = 20, + RGBA32F = 21, + RG32F = 22, + R32F = 23, + R16F = 24, + R16UNORM = 25, + R16S = 26, + R16UI = 27, + R16I = 28, + RG16 = 29, + RG16F = 30, + RG16UI = 31, + RG16I = 32, + RG16S = 33, + RGB32F = 34, + SRGBA8 = 35, + RG8U = 36, + RG8S = 37, + RG32UI = 38, + R32UI = 39, MaxColorFormat, // DepthStencil formats - Z24S8 = 36, - S8Z24 = 37, - Z32F = 38, - Z16 = 39, - Z32FS8 = 40, + Z24S8 = 40, + S8Z24 = 41, + Z32F = 42, + Z16 = 43, + Z32FS8 = 44, MaxDepthStencilFormat, @@ -113,6 +117,7 @@ struct SurfaceParams { 1, // R8 1, // R8UI 1, // RGBA16F + 1, // RGBA16UI 1, // R11FG11FB10F 1, // RGBA32UI 4, // DXT1 @@ -140,7 +145,10 @@ struct SurfaceParams { 1, // RG16S 1, // RGB32F 1, // SRGBA8 + 1, // RG8U 1, // RG8S + 1, // RG32UI + 1, // R32UI 1, // Z24S8 1, // S8Z24 1, // Z32F @@ -165,6 +173,7 @@ struct SurfaceParams { 8, // R8 8, // R8UI 64, // RGBA16F + 64, // RGBA16UI 32, // R11FG11FB10F 128, // RGBA32UI 64, // DXT1 @@ -192,7 +201,10 @@ struct SurfaceParams { 32, // RG16S 96, // RGB32F 32, // SRGBA8 + 16, // RG8U 16, // RG8S + 64, // RG32UI + 32, // R32UI 32, // Z24S8 32, // S8Z24 32, // Z32F @@ -241,6 +253,8 @@ struct SurfaceParams { return PixelFormat::A2B10G10R10; case Tegra::RenderTargetFormat::RGBA16_FLOAT: return PixelFormat::RGBA16F; + case Tegra::RenderTargetFormat::RGBA16_UINT: + return PixelFormat::RGBA16UI; case Tegra::RenderTargetFormat::RGBA32_FLOAT: return PixelFormat::RGBA32F; case Tegra::RenderTargetFormat::RG32_FLOAT: @@ -265,6 +279,8 @@ struct SurfaceParams { return PixelFormat::RG16; case Tegra::RenderTargetFormat::RG16_SNORM: return PixelFormat::RG16S; + case Tegra::RenderTargetFormat::RG8_UNORM: + return PixelFormat::RG8U; case Tegra::RenderTargetFormat::RG8_SNORM: return PixelFormat::RG8S; case Tegra::RenderTargetFormat::R16_FLOAT: @@ -279,6 +295,10 @@ struct SurfaceParams { return PixelFormat::R16I; case Tegra::RenderTargetFormat::R32_FLOAT: return PixelFormat::R32F; + case Tegra::RenderTargetFormat::R32_UINT: + return PixelFormat::R32UI; + case Tegra::RenderTargetFormat::RG32_UINT: + return PixelFormat::RG32UI; default: LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format)); UNREACHABLE(); @@ -332,7 +352,15 @@ struct SurfaceParams { static_cast<u32>(component_type)); UNREACHABLE(); case Tegra::Texture::TextureFormat::R32_G32: - return PixelFormat::RG32F; + switch (component_type) { + case Tegra::Texture::ComponentType::FLOAT: + return PixelFormat::RG32F; + case Tegra::Texture::ComponentType::UINT: + return PixelFormat::RG32UI; + } + LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", + static_cast<u32>(component_type)); + UNREACHABLE(); case Tegra::Texture::TextureFormat::R32_G32_B32: return PixelFormat::RGB32F; case Tegra::Texture::TextureFormat::R16: @@ -352,7 +380,15 @@ struct SurfaceParams { static_cast<u32>(component_type)); UNREACHABLE(); case Tegra::Texture::TextureFormat::R32: - return PixelFormat::R32F; + switch (component_type) { + case Tegra::Texture::ComponentType::FLOAT: + return PixelFormat::R32F; + case Tegra::Texture::ComponentType::UINT: + return PixelFormat::R32UI; + } + LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", + static_cast<u32>(component_type)); + UNREACHABLE(); case Tegra::Texture::TextureFormat::ZF32: return PixelFormat::Z32F; case Tegra::Texture::TextureFormat::Z24S8: @@ -432,6 +468,7 @@ struct SurfaceParams { case Tegra::RenderTargetFormat::RG16_UNORM: case Tegra::RenderTargetFormat::R16_UNORM: case Tegra::RenderTargetFormat::B5G6R5_UNORM: + case Tegra::RenderTargetFormat::RG8_UNORM: return ComponentType::UNorm; case Tegra::RenderTargetFormat::RGBA8_SNORM: case Tegra::RenderTargetFormat::RG16_SNORM: @@ -447,9 +484,12 @@ struct SurfaceParams { case Tegra::RenderTargetFormat::R32_FLOAT: return ComponentType::Float; case Tegra::RenderTargetFormat::RGBA32_UINT: + case Tegra::RenderTargetFormat::RGBA16_UINT: case Tegra::RenderTargetFormat::RG16_UINT: case Tegra::RenderTargetFormat::R8_UINT: case Tegra::RenderTargetFormat::R16_UINT: + case Tegra::RenderTargetFormat::RG32_UINT: + case Tegra::RenderTargetFormat::R32_UINT: return ComponentType::UInt; case Tegra::RenderTargetFormat::RG16_SINT: case Tegra::RenderTargetFormat::R16_SINT: diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h index 679e5ceb2..83ea0cfc0 100644 --- a/src/video_core/renderer_opengl/maxwell_to_gl.h +++ b/src/video_core/renderer_opengl/maxwell_to_gl.h @@ -27,6 +27,7 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) { case Maxwell::VertexAttribute::Type::UnsignedNorm: { switch (attrib.size) { + case Maxwell::VertexAttribute::Size::Size_8: case Maxwell::VertexAttribute::Size::Size_8_8: case Maxwell::VertexAttribute::Size::Size_8_8_8_8: return GL_UNSIGNED_BYTE; |
