summaryrefslogtreecommitdiffstats
path: root/src/video_core/renderer_opengl
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core/renderer_opengl')
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp3
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.cpp59
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.h20
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp77
4 files changed, 105 insertions, 54 deletions
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 84582c777..8d5f277e2 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -286,7 +286,8 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
&ubo, sizeof(ubo), static_cast<std::size_t>(uniform_buffer_alignment));
// Bind the buffer
- glBindBufferRange(GL_UNIFORM_BUFFER, stage, buffer_cache.GetHandle(), offset, sizeof(ubo));
+ glBindBufferRange(GL_UNIFORM_BUFFER, static_cast<GLuint>(stage), buffer_cache.GetHandle(),
+ offset, static_cast<GLsizeiptr>(sizeof(ubo)));
Shader shader{shader_cache.GetStageProgram(program)};
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
index 65a220c41..801d45144 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -231,6 +231,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
{GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // RG32UI
{GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // R32UI
{GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5
+ {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4
// Depth formats
{GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
@@ -277,7 +279,9 @@ static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType
static bool IsPixelFormatASTC(PixelFormat format) {
switch (format) {
case PixelFormat::ASTC_2D_4X4:
+ case PixelFormat::ASTC_2D_5X4:
case PixelFormat::ASTC_2D_8X8:
+ case PixelFormat::ASTC_2D_8X5:
return true;
default:
return false;
@@ -288,8 +292,12 @@ static std::pair<u32, u32> GetASTCBlockSize(PixelFormat format) {
switch (format) {
case PixelFormat::ASTC_2D_4X4:
return {4, 4};
+ case PixelFormat::ASTC_2D_5X4:
+ return {5, 4};
case PixelFormat::ASTC_2D_8X8:
return {8, 8};
+ case PixelFormat::ASTC_2D_8X5:
+ return {8, 5};
default:
LOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format));
UNREACHABLE();
@@ -323,8 +331,8 @@ static bool IsFormatBCn(PixelFormat format) {
}
template <bool morton_to_gl, PixelFormat format>
-void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::size_t gl_buffer_size,
- VAddr addr) {
+void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth, u8* gl_buffer,
+ std::size_t gl_buffer_size, VAddr addr) {
constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
@@ -333,7 +341,7 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::si
// pixel values.
const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
const std::vector<u8> data = Tegra::Texture::UnswizzleTexture(
- addr, tile_size, bytes_per_pixel, stride, height, block_height);
+ addr, tile_size, bytes_per_pixel, stride, height, depth, block_height, block_depth);
const std::size_t size_to_copy{std::min(gl_buffer_size, data.size())};
memcpy(gl_buffer, data.data(), size_to_copy);
} else {
@@ -345,7 +353,7 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::si
}
}
-static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr),
SurfaceParams::MaxPixelFormat>
morton_to_gl_fns = {
// clang-format off
@@ -395,6 +403,8 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
MortonCopy<true, PixelFormat::RG32UI>,
MortonCopy<true, PixelFormat::R32UI>,
MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
+ MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
+ MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
MortonCopy<true, PixelFormat::Z32F>,
MortonCopy<true, PixelFormat::Z16>,
MortonCopy<true, PixelFormat::Z24S8>,
@@ -403,7 +413,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
// clang-format on
};
-static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr),
SurfaceParams::MaxPixelFormat>
gl_to_morton_fns = {
// clang-format off
@@ -455,6 +465,8 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
MortonCopy<false, PixelFormat::RG32UI>,
MortonCopy<false, PixelFormat::R32UI>,
nullptr,
+ nullptr,
+ nullptr,
MortonCopy<false, PixelFormat::Z32F>,
MortonCopy<false, PixelFormat::Z16>,
MortonCopy<false, PixelFormat::Z24S8>,
@@ -790,7 +802,9 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
u32 width, u32 height) {
switch (pixel_format) {
case PixelFormat::ASTC_2D_4X4:
- case PixelFormat::ASTC_2D_8X8: {
+ case PixelFormat::ASTC_2D_8X8:
+ case PixelFormat::ASTC_2D_8X5:
+ case PixelFormat::ASTC_2D_5X4: {
// Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC.
u32 block_width{};
u32 block_height{};
@@ -827,36 +841,23 @@ void CachedSurface::LoadGLBuffer() {
if (params.is_tiled) {
gl_buffer.resize(total_size);
+ u32 depth = params.depth;
+ u32 block_depth = params.block_depth;
ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
params.block_width, static_cast<u32>(params.target));
- ASSERT_MSG(params.block_depth == 1, "Block depth is defined as {} on texture type {}",
- params.block_depth, static_cast<u32>(params.target));
- // TODO(bunnei): This only unswizzles and copies a 2D texture - we do not yet know how to do
- // this for 3D textures, etc.
- switch (params.target) {
- case SurfaceParams::SurfaceTarget::Texture2D:
- // Pass impl. to the fallback code below
- break;
- case SurfaceParams::SurfaceTarget::Texture2DArray:
- case SurfaceParams::SurfaceTarget::TextureCubemap:
- for (std::size_t index = 0; index < params.depth; ++index) {
- const std::size_t offset{index * copy_size};
- morton_to_gl_fns[static_cast<std::size_t>(params.pixel_format)](
- params.width, params.block_height, params.height, gl_buffer.data() + offset,
- copy_size, params.addr + offset);
- }
- break;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented tiled load for target={}",
- static_cast<u32>(params.target));
- UNREACHABLE();
+ if (params.target == SurfaceParams::SurfaceTarget::Texture2D) {
+ // TODO(Blinkhawk): Eliminate this condition once all texture types are implemented.
+ depth = 1U;
+ block_depth = 1U;
}
+ const std::size_t size = copy_size * depth;
+
morton_to_gl_fns[static_cast<std::size_t>(params.pixel_format)](
- params.width, params.block_height, params.height, gl_buffer.data(), copy_size,
- params.addr);
+ params.width, params.block_height, params.height, block_depth, depth, gl_buffer.data(),
+ size, params.addr);
} else {
const u8* const texture_src_data_end{texture_src_data + total_size};
gl_buffer.assign(texture_src_data, texture_src_data_end);
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index 66d98ad4e..0b8ae3eb4 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -74,19 +74,21 @@ struct SurfaceParams {
RG32UI = 43,
R32UI = 44,
ASTC_2D_8X8 = 45,
+ ASTC_2D_8X5 = 46,
+ ASTC_2D_5X4 = 47,
MaxColorFormat,
// Depth formats
- Z32F = 46,
- Z16 = 47,
+ Z32F = 48,
+ Z16 = 49,
MaxDepthFormat,
// DepthStencil formats
- Z24S8 = 48,
- S8Z24 = 49,
- Z32FS8 = 50,
+ Z24S8 = 50,
+ S8Z24 = 51,
+ Z32FS8 = 52,
MaxDepthStencilFormat,
@@ -220,6 +222,8 @@ struct SurfaceParams {
1, // RG32UI
1, // R32UI
4, // ASTC_2D_8X8
+ 4, // ASTC_2D_8X5
+ 4, // ASTC_2D_5X4
1, // Z32F
1, // Z16
1, // Z24S8
@@ -282,6 +286,8 @@ struct SurfaceParams {
64, // RG32UI
32, // R32UI
16, // ASTC_2D_8X8
+ 32, // ASTC_2D_8X5
+ 32, // ASTC_2D_5X4
32, // Z32F
16, // Z16
32, // Z24S8
@@ -553,8 +559,12 @@ struct SurfaceParams {
return PixelFormat::BC6H_SF16;
case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
return PixelFormat::ASTC_2D_4X4;
+ case Tegra::Texture::TextureFormat::ASTC_2D_5X4:
+ return PixelFormat::ASTC_2D_5X4;
case Tegra::Texture::TextureFormat::ASTC_2D_8X8:
return PixelFormat::ASTC_2D_8X8;
+ case Tegra::Texture::TextureFormat::ASTC_2D_8X5:
+ return PixelFormat::ASTC_2D_8X5;
case Tegra::Texture::TextureFormat::R16_G16:
switch (component_type) {
case Tegra::Texture::ComponentType::FLOAT:
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 8dfb49507..28dba0084 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -1436,7 +1436,6 @@ private:
break;
}
-
case OpCode::Type::Shift: {
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, true);
std::string op_b;
@@ -1478,7 +1477,6 @@ private:
}
break;
}
-
case OpCode::Type::ArithmeticIntegerImmediate: {
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
std::string op_b = std::to_string(instr.alu.imm20_32.Value());
@@ -2038,9 +2036,9 @@ private:
break;
}
case OpCode::Id::TEX: {
- ASSERT_MSG(instr.tex.array == 0, "TEX arrays unimplemented");
Tegra::Shader::TextureType texture_type{instr.tex.texture_type};
std::string coord;
+ const bool is_array = instr.tex.array != 0;
ASSERT_MSG(!instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
@@ -2055,21 +2053,59 @@ private:
switch (num_coordinates) {
case 1: {
- const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
- coord = "float coords = " + x + ';';
+ if (is_array) {
+ const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+ coord = "vec2 coords = vec2(" + x + ", " + index + ");";
+ } else {
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+ coord = "float coords = " + x + ';';
+ }
break;
}
case 2: {
- const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
- const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
- coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+ if (is_array) {
+ const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+ const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+ coord = "vec3 coords = vec3(" + x + ", " + y + ", " + index + ");";
+ } else {
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+ const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+ coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+ }
break;
}
case 3: {
- const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
- const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
- const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
- coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+ if (depth_compare) {
+ if (is_array) {
+ const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+ const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+ const std::string z = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+ coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
+ ");";
+ } else {
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+ const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+ const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
+ coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+ }
+ } else {
+ if (is_array) {
+ const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+ const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+ const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 3);
+ coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
+ ");";
+ } else {
+ const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+ const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+ const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+ coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+ }
+ }
break;
}
default:
@@ -2088,7 +2124,7 @@ private:
std::string op_c;
const std::string sampler =
- GetSampler(instr.sampler, texture_type, false, depth_compare);
+ GetSampler(instr.sampler, texture_type, is_array, depth_compare);
// Add an extra scope and declare the texture coords inside to prevent
// overwriting them in case they are used as outputs of the texs instruction.
@@ -2108,10 +2144,13 @@ private:
}
case Tegra::Shader::TextureProcessMode::LB:
case Tegra::Shader::TextureProcessMode::LBA: {
- if (num_coordinates <= 2) {
- op_c = regs.GetRegisterAsFloat(instr.gpr20);
+ if (depth_compare) {
+ if (is_array)
+ op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 2);
+ else
+ op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
} else {
- op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+ op_c = regs.GetRegisterAsFloat(instr.gpr20);
}
// TODO: Figure if A suffix changes the equation at all.
texture = "texture(" + sampler + ", coords, " + op_c + ')';
@@ -2626,14 +2665,14 @@ private:
const std::string pred =
GetPredicateCondition(instr.csetp.pred39, instr.csetp.neg_pred39 != 0);
const std::string combiner = GetPredicateCombiner(instr.csetp.op);
- const std::string controlCode = regs.GetControlCode(instr.csetp.cc);
+ const std::string control_code = regs.GetControlCode(instr.csetp.cc);
if (instr.csetp.pred3 != static_cast<u64>(Pred::UnusedIndex)) {
SetPredicate(instr.csetp.pred3,
- '(' + controlCode + ") " + combiner + " (" + pred + ')');
+ '(' + control_code + ") " + combiner + " (" + pred + ')');
}
if (instr.csetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
SetPredicate(instr.csetp.pred0,
- "!(" + controlCode + ") " + combiner + " (" + pred + ')');
+ "!(" + control_code + ") " + combiner + " (" + pred + ')');
}
break;
}