summaryrefslogtreecommitdiffstats
path: root/src/video_core
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core')
-rw-r--r--src/video_core/CMakeLists.txt4
-rw-r--r--src/video_core/command_processor.cpp2
-rw-r--r--src/video_core/engines/maxwell_3d.cpp10
-rw-r--r--src/video_core/engines/maxwell_3d.h7
-rw-r--r--src/video_core/engines/shader_bytecode.h14
-rw-r--r--src/video_core/macro_interpreter.cpp6
-rw-r--r--src/video_core/macro_interpreter.h5
-rw-r--r--src/video_core/memory_manager.cpp8
-rw-r--r--src/video_core/memory_manager.h7
-rw-r--r--src/video_core/renderer_base.h6
-rw-r--r--src/video_core/renderer_opengl/gl_buffer_cache.cpp2
-rw-r--r--src/video_core/renderer_opengl/gl_primitive_assembler.cpp2
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp32
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h5
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.cpp464
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.h834
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.cpp5
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp265
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.h8
-rw-r--r--src/video_core/renderer_opengl/gl_shader_gen.cpp8
-rw-r--r--src/video_core/renderer_opengl/gl_state.cpp13
-rw-r--r--src/video_core/renderer_opengl/gl_state.h14
-rw-r--r--src/video_core/renderer_opengl/maxwell_to_gl.h27
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp25
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.h3
-rw-r--r--src/video_core/renderer_opengl/utils.cpp38
-rw-r--r--src/video_core/renderer_opengl/utils.h15
-rw-r--r--src/video_core/surface.cpp499
-rw-r--r--src/video_core/surface.h385
-rw-r--r--src/video_core/textures/decoders.h6
-rw-r--r--src/video_core/textures/texture.h5
-rw-r--r--src/video_core/utils.h26
32 files changed, 1633 insertions, 1117 deletions
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index 09ecc5bad..ddb1a1d69 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -51,6 +51,10 @@ add_library(video_core STATIC
renderer_opengl/maxwell_to_gl.h
renderer_opengl/renderer_opengl.cpp
renderer_opengl/renderer_opengl.h
+ renderer_opengl/utils.cpp
+ renderer_opengl/utils.h
+ surface.cpp
+ surface.h
textures/astc.cpp
textures/astc.h
textures/decoders.cpp
diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp
index f1aa6091b..28e8c13aa 100644
--- a/src/video_core/command_processor.cpp
+++ b/src/video_core/command_processor.cpp
@@ -81,7 +81,7 @@ void GPU::ProcessCommandLists(const std::vector<CommandListHeader>& commands) {
for (auto entry : commands) {
Tegra::GPUVAddr address = entry.Address();
u32 size = entry.sz;
- const boost::optional<VAddr> head_address = memory_manager->GpuToCpuAddress(address);
+ const std::optional<VAddr> head_address = memory_manager->GpuToCpuAddress(address);
VAddr current_addr = *head_address;
while (current_addr < *head_address + size * sizeof(CommandHeader)) {
const CommandHeader header = {Memory::Read32(current_addr)};
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 27ef865a2..7357d20d1 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -167,7 +167,7 @@ void Maxwell3D::ProcessQueryGet() {
GPUVAddr sequence_address = regs.query.QueryAddress();
// Since the sequence address is given as a GPU VAddr, we have to convert it to an application
// VAddr before writing.
- boost::optional<VAddr> address = memory_manager.GpuToCpuAddress(sequence_address);
+ std::optional<VAddr> address = memory_manager.GpuToCpuAddress(sequence_address);
// TODO(Subv): Support the other query units.
ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop,
@@ -285,7 +285,7 @@ void Maxwell3D::ProcessCBData(u32 value) {
// Don't allow writing past the end of the buffer.
ASSERT(regs.const_buffer.cb_pos + sizeof(u32) <= regs.const_buffer.cb_size);
- boost::optional<VAddr> address =
+ std::optional<VAddr> address =
memory_manager.GpuToCpuAddress(buffer_address + regs.const_buffer.cb_pos);
Memory::Write32(*address, value);
@@ -298,7 +298,7 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
GPUVAddr tic_base_address = regs.tic.TICAddress();
GPUVAddr tic_address_gpu = tic_base_address + tic_index * sizeof(Texture::TICEntry);
- boost::optional<VAddr> tic_address_cpu = memory_manager.GpuToCpuAddress(tic_address_gpu);
+ std::optional<VAddr> tic_address_cpu = memory_manager.GpuToCpuAddress(tic_address_gpu);
Texture::TICEntry tic_entry;
Memory::ReadBlock(*tic_address_cpu, &tic_entry, sizeof(Texture::TICEntry));
@@ -322,7 +322,7 @@ Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const {
GPUVAddr tsc_base_address = regs.tsc.TSCAddress();
GPUVAddr tsc_address_gpu = tsc_base_address + tsc_index * sizeof(Texture::TSCEntry);
- boost::optional<VAddr> tsc_address_cpu = memory_manager.GpuToCpuAddress(tsc_address_gpu);
+ std::optional<VAddr> tsc_address_cpu = memory_manager.GpuToCpuAddress(tsc_address_gpu);
Texture::TSCEntry tsc_entry;
Memory::ReadBlock(*tsc_address_cpu, &tsc_entry, sizeof(Texture::TSCEntry));
@@ -386,7 +386,7 @@ Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
ASSERT(tex_info_address < tex_info_buffer.address + tex_info_buffer.size);
- boost::optional<VAddr> tex_address_cpu = memory_manager.GpuToCpuAddress(tex_info_address);
+ std::optional<VAddr> tex_address_cpu = memory_manager.GpuToCpuAddress(tex_info_address);
Texture::TextureHandle tex_handle{Memory::Read32(*tex_address_cpu)};
Texture::FullTextureInfo tex_info{};
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index d6978162a..443affc36 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -723,7 +723,11 @@ public:
StencilOp stencil_back_op_zpass;
ComparisonOp stencil_back_func_func;
- INSERT_PADDING_WORDS(0x17);
+ INSERT_PADDING_WORDS(0x4);
+
+ u32 framebuffer_srgb;
+
+ INSERT_PADDING_WORDS(0x12);
union {
BitField<2, 1, u32> coord_origin;
@@ -1086,6 +1090,7 @@ ASSERT_REG_POSITION(stencil_back_op_fail, 0x566);
ASSERT_REG_POSITION(stencil_back_op_zfail, 0x567);
ASSERT_REG_POSITION(stencil_back_op_zpass, 0x568);
ASSERT_REG_POSITION(stencil_back_func_func, 0x569);
+ASSERT_REG_POSITION(framebuffer_srgb, 0x56E);
ASSERT_REG_POSITION(point_coord_replace, 0x581);
ASSERT_REG_POSITION(code_address, 0x582);
ASSERT_REG_POSITION(draw, 0x585);
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index 141b9159b..27c011e6f 100644
--- a/src/video_core/engines/shader_bytecode.h
+++ b/src/video_core/engines/shader_bytecode.h
@@ -5,12 +5,11 @@
#pragma once
#include <bitset>
+#include <optional>
#include <string>
#include <tuple>
#include <vector>
-#include <boost/optional.hpp>
-
#include "common/assert.h"
#include "common/bit_field.h"
#include "common/common_types.h"
@@ -579,6 +578,10 @@ union Instruction {
} fmul32;
union {
+ BitField<52, 1, u64> generates_cc;
+ } op_32;
+
+ union {
BitField<48, 1, u64> is_signed;
} shift;
@@ -1456,7 +1459,7 @@ public:
Type type;
};
- static boost::optional<const Matcher&> Decode(Instruction instr) {
+ static std::optional<std::reference_wrapper<const Matcher>> Decode(Instruction instr) {
static const auto table{GetDecodeTable()};
const auto matches_instruction = [instr](const auto& matcher) {
@@ -1464,7 +1467,8 @@ public:
};
auto iter = std::find_if(table.begin(), table.end(), matches_instruction);
- return iter != table.end() ? boost::optional<const Matcher&>(*iter) : boost::none;
+ return iter != table.end() ? std::optional<std::reference_wrapper<const Matcher>>(*iter)
+ : std::nullopt;
}
private:
@@ -1658,4 +1662,4 @@ private:
}
};
-} // namespace Tegra::Shader \ No newline at end of file
+} // namespace Tegra::Shader
diff --git a/src/video_core/macro_interpreter.cpp b/src/video_core/macro_interpreter.cpp
index 377bd66ab..f6af132fb 100644
--- a/src/video_core/macro_interpreter.cpp
+++ b/src/video_core/macro_interpreter.cpp
@@ -29,7 +29,7 @@ void MacroInterpreter::Execute(const std::vector<u32>& code, std::vector<u32> pa
void MacroInterpreter::Reset() {
registers = {};
pc = 0;
- delayed_pc = boost::none;
+ delayed_pc = {};
method_address.raw = 0;
parameters.clear();
// The next parameter index starts at 1, because $r1 already has the value of the first
@@ -44,10 +44,10 @@ bool MacroInterpreter::Step(const std::vector<u32>& code, bool is_delay_slot) {
pc += 4;
// Update the program counter if we were delayed
- if (delayed_pc != boost::none) {
+ if (delayed_pc) {
ASSERT(is_delay_slot);
pc = *delayed_pc;
- delayed_pc = boost::none;
+ delayed_pc = {};
}
switch (opcode.operation) {
diff --git a/src/video_core/macro_interpreter.h b/src/video_core/macro_interpreter.h
index cee0baaf3..773684bde 100644
--- a/src/video_core/macro_interpreter.h
+++ b/src/video_core/macro_interpreter.h
@@ -5,8 +5,9 @@
#pragma once
#include <array>
+#include <optional>
#include <vector>
-#include <boost/optional.hpp>
+
#include "common/bit_field.h"
#include "common/common_types.h"
@@ -149,7 +150,7 @@ private:
Engines::Maxwell3D& maxwell3d;
u32 pc; ///< Current program counter
- boost::optional<u32>
+ std::optional<u32>
delayed_pc; ///< Program counter to execute at after the delay slot is executed.
static constexpr std::size_t NumMacroRegisters = 8;
diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp
index 022d4ab74..90a8e825d 100644
--- a/src/video_core/memory_manager.cpp
+++ b/src/video_core/memory_manager.cpp
@@ -9,7 +9,7 @@
namespace Tegra {
GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
- boost::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, align);
+ std::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, align);
ASSERT(gpu_addr);
for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
@@ -34,7 +34,7 @@ GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) {
}
GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
- boost::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, PAGE_SIZE);
+ std::optional<GPUVAddr> gpu_addr = FindFreeBlock(size, PAGE_SIZE);
ASSERT(gpu_addr);
for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
@@ -97,7 +97,7 @@ GPUVAddr MemoryManager::GetRegionEnd(GPUVAddr region_start) const {
return {};
}
-boost::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
+std::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
GPUVAddr gpu_addr = 0;
u64 free_space = 0;
align = (align + PAGE_MASK) & ~PAGE_MASK;
@@ -118,7 +118,7 @@ boost::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
return {};
}
-boost::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) {
+std::optional<VAddr> MemoryManager::GpuToCpuAddress(GPUVAddr gpu_addr) {
VAddr base_addr = PageSlot(gpu_addr);
if (base_addr == static_cast<u64>(PageStatus::Allocated) ||
diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h
index caf80093f..b1255fd56 100644
--- a/src/video_core/memory_manager.h
+++ b/src/video_core/memory_manager.h
@@ -6,10 +6,9 @@
#include <array>
#include <memory>
+#include <optional>
#include <vector>
-#include <boost/optional.hpp>
-
#include "common/common_types.h"
namespace Tegra {
@@ -27,7 +26,7 @@ public:
GPUVAddr MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size);
GPUVAddr UnmapBuffer(GPUVAddr gpu_addr, u64 size);
GPUVAddr GetRegionEnd(GPUVAddr region_start) const;
- boost::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
+ std::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
std::vector<GPUVAddr> CpuToGpuAddress(VAddr cpu_addr) const;
static constexpr u64 PAGE_BITS = 16;
@@ -35,7 +34,7 @@ public:
static constexpr u64 PAGE_MASK = PAGE_SIZE - 1;
private:
- boost::optional<GPUVAddr> FindFreeBlock(u64 size, u64 align = 1);
+ std::optional<GPUVAddr> FindFreeBlock(u64 size, u64 align = 1);
bool IsPageMapped(GPUVAddr gpu_addr);
VAddr& PageSlot(GPUVAddr gpu_addr);
diff --git a/src/video_core/renderer_base.h b/src/video_core/renderer_base.h
index 2cd0738ff..669e26e15 100644
--- a/src/video_core/renderer_base.h
+++ b/src/video_core/renderer_base.h
@@ -6,7 +6,8 @@
#include <atomic>
#include <memory>
-#include <boost/optional.hpp>
+#include <optional>
+
#include "common/common_types.h"
#include "video_core/gpu.h"
#include "video_core/rasterizer_interface.h"
@@ -28,7 +29,8 @@ public:
virtual ~RendererBase();
/// Swap buffers (render frame)
- virtual void SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) = 0;
+ virtual void SwapBuffers(
+ std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) = 0;
/// Initialize the renderer
virtual bool Init() = 0;
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
index c142095c5..41a54b3e7 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
@@ -17,7 +17,7 @@ OGLBufferCache::OGLBufferCache(std::size_t size) : stream_buffer(GL_ARRAY_BUFFER
GLintptr OGLBufferCache::UploadMemory(Tegra::GPUVAddr gpu_addr, std::size_t size,
std::size_t alignment, bool cache) {
auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
- const boost::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
+ const std::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
// Cache management is a big overhead, so only cache entries with a given size.
// TODO: Figure out which size is the best for given games.
diff --git a/src/video_core/renderer_opengl/gl_primitive_assembler.cpp b/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
index ee1d9601b..741f14bc3 100644
--- a/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
+++ b/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
@@ -45,7 +45,7 @@ GLintptr PrimitiveAssembler::MakeQuadIndexed(Tegra::GPUVAddr gpu_addr, std::size
auto [dst_pointer, index_offset] = buffer_cache.ReserveMemory(map_size);
auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
- const boost::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
+ const std::optional<VAddr> cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr)};
const u8* source{Memory::GetPointer(*cpu_addr)};
for (u32 primitive = 0; primitive < count / 4; ++primitive) {
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index cd4216c4e..75e31c6de 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -30,8 +30,8 @@
namespace OpenGL {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
-using PixelFormat = SurfaceParams::PixelFormat;
-using SurfaceType = SurfaceParams::SurfaceType;
+using PixelFormat = VideoCore::Surface::PixelFormat;
+using SurfaceType = VideoCore::Surface::SurfaceType;
MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Array Setup", MP_RGB(128, 128, 192));
MICROPROFILE_DEFINE(OpenGL_Shader, "OpenGL", "Shader Setup", MP_RGB(128, 128, 192));
@@ -401,7 +401,7 @@ void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_depth_fb,
bool preserve_contents,
- boost::optional<std::size_t> single_color_target) {
+ std::optional<std::size_t> single_color_target) {
MICROPROFILE_SCOPE(OpenGL_Framebuffer);
const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
@@ -418,6 +418,7 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
// Bind the framebuffer surfaces
state.draw.draw_framebuffer = framebuffer.handle;
state.Apply();
+ state.framebuffer_srgb.enabled = regs.framebuffer_srgb != 0;
if (using_color_fb) {
if (single_color_target) {
@@ -429,6 +430,9 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
// Assume that a surface will be written to if it is used as a framebuffer, even if
// the shader doesn't actually write to it.
color_surface->MarkAsModified(true, res_cache);
+ // Workaround for and issue in nvidia drivers
+ // https://devtalk.nvidia.com/default/topic/776591/opengl/gl_framebuffer_srgb-functions-incorrectly/
+ state.framebuffer_srgb.enabled |= color_surface->GetSurfaceParams().srgb_conversion;
}
glFramebufferTexture2D(
@@ -446,6 +450,11 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
// Assume that a surface will be written to if it is used as a framebuffer, even
// if the shader doesn't actually write to it.
color_surface->MarkAsModified(true, res_cache);
+ // Enable sRGB only for supported formats
+ // Workaround for and issue in nvidia drivers
+ // https://devtalk.nvidia.com/default/topic/776591/opengl/gl_framebuffer_srgb-functions-incorrectly/
+ state.framebuffer_srgb.enabled |=
+ color_surface->GetSurfaceParams().srgb_conversion;
}
buffers[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index);
@@ -537,7 +546,9 @@ void RasterizerOpenGL::Clear() {
ConfigureFramebuffers(use_color, use_depth || use_stencil, false,
regs.clear_buffers.RT.Value());
-
+ // Copy the sRGB setting to the clear state to avoid problem with
+ // specific driver implementations
+ clear_state.framebuffer_srgb.enabled = state.framebuffer_srgb.enabled;
clear_state.Apply();
if (use_color) {
@@ -692,7 +703,8 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
// Verify that the cached surface is the same size and format as the requested framebuffer
const auto& params{surface->GetSurfaceParams()};
- const auto& pixel_format{SurfaceParams::PixelFormatFromGPUPixelFormat(config.pixel_format)};
+ const auto& pixel_format{
+ VideoCore::Surface::PixelFormatFromGPUPixelFormat(config.pixel_format)};
ASSERT_MSG(params.width == config.width, "Framebuffer width is different");
ASSERT_MSG(params.height == config.height, "Framebuffer height is different");
ASSERT_MSG(params.pixel_format == pixel_format, "Framebuffer pixel_format is different");
@@ -720,11 +732,15 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
if (mag_filter != config.mag_filter) {
mag_filter = config.mag_filter;
- glSamplerParameteri(s, GL_TEXTURE_MAG_FILTER, MaxwellToGL::TextureFilterMode(mag_filter));
+ glSamplerParameteri(
+ s, GL_TEXTURE_MAG_FILTER,
+ MaxwellToGL::TextureFilterMode(mag_filter, Tegra::Texture::TextureMipmapFilter::None));
}
- if (min_filter != config.min_filter) {
+ if (min_filter != config.min_filter || mip_filter != config.mip_filter) {
min_filter = config.min_filter;
- glSamplerParameteri(s, GL_TEXTURE_MIN_FILTER, MaxwellToGL::TextureFilterMode(min_filter));
+ mip_filter = config.mip_filter;
+ glSamplerParameteri(s, GL_TEXTURE_MIN_FILTER,
+ MaxwellToGL::TextureFilterMode(min_filter, mip_filter));
}
if (wrap_u != config.wrap_u) {
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index 5020a5392..47097c569 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -8,12 +8,12 @@
#include <cstddef>
#include <map>
#include <memory>
+#include <optional>
#include <tuple>
#include <utility>
#include <vector>
#include <boost/icl/interval_map.hpp>
-#include <boost/optional.hpp>
#include <boost/range/iterator_range.hpp>
#include <glad/glad.h>
@@ -93,6 +93,7 @@ private:
private:
Tegra::Texture::TextureFilter mag_filter;
Tegra::Texture::TextureFilter min_filter;
+ Tegra::Texture::TextureMipmapFilter mip_filter;
Tegra::Texture::WrapMode wrap_u;
Tegra::Texture::WrapMode wrap_v;
Tegra::Texture::WrapMode wrap_p;
@@ -110,7 +111,7 @@ private:
*/
void ConfigureFramebuffers(bool use_color_fb = true, bool using_depth_fb = true,
bool preserve_contents = true,
- boost::optional<std::size_t> single_color_target = {});
+ std::optional<std::size_t> single_color_target = {});
/*
* Configures the current constbuffers to use for the draw command.
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
index 591ec7998..f194a7687 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -16,15 +16,21 @@
#include "core/settings.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_opengl/gl_rasterizer_cache.h"
+#include "video_core/renderer_opengl/utils.h"
+#include "video_core/surface.h"
#include "video_core/textures/astc.h"
#include "video_core/textures/decoders.h"
#include "video_core/utils.h"
namespace OpenGL {
-using SurfaceType = SurfaceParams::SurfaceType;
-using PixelFormat = SurfaceParams::PixelFormat;
-using ComponentType = SurfaceParams::ComponentType;
+using VideoCore::Surface::ComponentTypeFromDepthFormat;
+using VideoCore::Surface::ComponentTypeFromRenderTarget;
+using VideoCore::Surface::ComponentTypeFromTexture;
+using VideoCore::Surface::PixelFormatFromDepthFormat;
+using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
+using VideoCore::Surface::PixelFormatFromTextureFormat;
+using VideoCore::Surface::SurfaceTargetFromTextureType;
struct FormatTuple {
GLint internal_format;
@@ -34,34 +40,6 @@ struct FormatTuple {
bool compressed;
};
-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;
- }
-}
-
-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();
- }
-}
-
void SurfaceParams::InitCacheParameters(Tegra::GPUVAddr gpu_addr_) {
auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
const auto cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr_)};
@@ -78,27 +56,36 @@ void SurfaceParams::InitCacheParameters(Tegra::GPUVAddr gpu_addr_) {
}
}
-std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
+std::size_t SurfaceParams::InnerMipmapMemorySize(u32 mip_level, bool force_gl, bool layer_only,
+ bool uncompressed) const {
const u32 compression_factor{GetCompressionFactor(pixel_format)};
const u32 bytes_per_pixel{GetBytesPerPixel(pixel_format)};
u32 m_depth = (layer_only ? 1U : depth);
- u32 m_width = std::max(1U, width / compression_factor);
- u32 m_height = std::max(1U, height / compression_factor);
- std::size_t size = Tegra::Texture::CalculateSize(is_tiled, bytes_per_pixel, m_width, m_height,
- m_depth, block_height, block_depth);
- u32 m_block_height = block_height;
- u32 m_block_depth = block_depth;
- std::size_t block_size_bytes = 512 * block_height * block_depth; // 512 is GOB size
- for (u32 i = 1; i < max_mip_level; i++) {
- m_width = std::max(1U, m_width / 2);
- m_height = std::max(1U, m_height / 2);
- m_depth = std::max(1U, m_depth / 2);
- m_block_height = std::max(1U, m_block_height / 2);
- m_block_depth = std::max(1U, m_block_depth / 2);
- size += Tegra::Texture::CalculateSize(is_tiled, bytes_per_pixel, m_width, m_height, m_depth,
- m_block_height, m_block_depth);
+ u32 m_width = MipWidth(mip_level);
+ u32 m_height = MipHeight(mip_level);
+ m_width = uncompressed ? m_width
+ : std::max(1U, (m_width + compression_factor - 1) / compression_factor);
+ m_height = uncompressed
+ ? m_height
+ : std::max(1U, (m_height + compression_factor - 1) / compression_factor);
+ m_depth = std::max(1U, m_depth >> mip_level);
+ u32 m_block_height = MipBlockHeight(mip_level);
+ u32 m_block_depth = MipBlockDepth(mip_level);
+ return Tegra::Texture::CalculateSize(force_gl ? false : is_tiled, bytes_per_pixel, m_width,
+ m_height, m_depth, m_block_height, m_block_depth);
+}
+
+std::size_t SurfaceParams::InnerMemorySize(bool force_gl, bool layer_only,
+ bool uncompressed) const {
+ std::size_t block_size_bytes = Tegra::Texture::GetGOBSize() * block_height * block_depth;
+ std::size_t size = 0;
+ for (u32 i = 0; i < max_mip_level; i++) {
+ size += InnerMipmapMemorySize(i, force_gl, layer_only, uncompressed);
+ }
+ if (!force_gl && is_tiled) {
+ size = Common::AlignUp(size, block_size_bytes);
}
- return is_tiled ? Common::AlignUp(size, block_size_bytes) : size;
+ return size;
}
/*static*/ SurfaceParams SurfaceParams::CreateForTexture(
@@ -108,8 +95,9 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
params.block_width = params.is_tiled ? config.tic.BlockWidth() : 0,
params.block_height = params.is_tiled ? config.tic.BlockHeight() : 0,
params.block_depth = params.is_tiled ? config.tic.BlockDepth() : 0,
- params.pixel_format =
- PixelFormatFromTextureFormat(config.tic.format, config.tic.r_type.Value());
+ params.srgb_conversion = config.tic.IsSrgbConversionEnabled();
+ params.pixel_format = PixelFormatFromTextureFormat(config.tic.format, config.tic.r_type.Value(),
+ params.srgb_conversion);
params.component_type = ComponentTypeFromTexture(config.tic.r_type.Value());
params.type = GetFormatType(params.pixel_format);
params.width = Common::AlignUp(config.tic.Width(), GetCompressionFactor(params.pixel_format));
@@ -166,6 +154,8 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
params.block_height = 1 << config.memory_layout.block_height;
params.block_depth = 1 << config.memory_layout.block_depth;
params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
+ params.srgb_conversion = config.format == Tegra::RenderTargetFormat::BGRA8_SRGB ||
+ config.format == Tegra::RenderTargetFormat::RGBA8_SRGB;
params.component_type = ComponentTypeFromRenderTarget(config.format);
params.type = GetFormatType(params.pixel_format);
params.width = config.width;
@@ -173,7 +163,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
params.unaligned_height = config.height;
params.target = SurfaceTarget::Texture2D;
params.depth = 1;
- params.max_mip_level = 0;
+ params.max_mip_level = 1;
params.is_layered = false;
// Render target specific parameters, not used for caching
@@ -201,12 +191,13 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
params.pixel_format = PixelFormatFromDepthFormat(format);
params.component_type = ComponentTypeFromDepthFormat(format);
params.type = GetFormatType(params.pixel_format);
+ params.srgb_conversion = false;
params.width = zeta_width;
params.height = zeta_height;
params.unaligned_height = zeta_height;
params.target = SurfaceTarget::Texture2D;
params.depth = 1;
- params.max_mip_level = 0;
+ params.max_mip_level = 1;
params.is_layered = false;
params.rt = {};
@@ -224,6 +215,8 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
params.block_height = params.is_tiled ? std::min(config.BlockHeight(), 32U) : 0,
params.block_depth = params.is_tiled ? std::min(config.BlockDepth(), 32U) : 0,
params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
+ params.srgb_conversion = config.format == Tegra::RenderTargetFormat::BGRA8_SRGB ||
+ config.format == Tegra::RenderTargetFormat::RGBA8_SRGB;
params.component_type = ComponentTypeFromRenderTarget(config.format);
params.type = GetFormatType(params.pixel_format);
params.width = config.width;
@@ -231,7 +224,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
params.unaligned_height = config.height;
params.target = SurfaceTarget::Texture2D;
params.depth = 1;
- params.max_mip_level = 0;
+ params.max_mip_level = 1;
params.rt = {};
params.InitCacheParameters(config.Address());
@@ -239,7 +232,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
return params;
}
-static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
+static constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format_tuples = {{
{GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // ABGR8U
{GL_RGBA8, GL_RGBA, GL_BYTE, ComponentType::SNorm, false}, // ABGR8S
{GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE, ComponentType::UInt, false}, // ABGR8UI
@@ -255,7 +248,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
{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
- {GL_COMPRESSED_RGB_S3TC_DXT1_EXT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
+ {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
true}, // DXT1
{GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
true}, // DXT23
@@ -289,14 +282,29 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
{GL_RG16I, GL_RG_INTEGER, GL_SHORT, ComponentType::SInt, false}, // RG16I
{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
- {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
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm,
+ false}, // RGBA8_SRGB
+ {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
+ {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
+ {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // BGRA8
+ // Compressed sRGB formats
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
+ true}, // DXT1_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
+ true}, // DXT23_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
+ true}, // DXT45_SRGB
+ {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM_ARB, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8,
+ ComponentType::UNorm, true}, // BC7U_SRGB
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4_SRGB
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8_SRGB
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5_SRGB
+ {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4_SRGB
// Depth formats
{GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
@@ -312,19 +320,19 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
ComponentType::Float, false}, // Z32FS8
}};
-static GLenum SurfaceTargetToGL(SurfaceParams::SurfaceTarget target) {
+static GLenum SurfaceTargetToGL(SurfaceTarget target) {
switch (target) {
- case SurfaceParams::SurfaceTarget::Texture1D:
+ case SurfaceTarget::Texture1D:
return GL_TEXTURE_1D;
- case SurfaceParams::SurfaceTarget::Texture2D:
+ case SurfaceTarget::Texture2D:
return GL_TEXTURE_2D;
- case SurfaceParams::SurfaceTarget::Texture3D:
+ case SurfaceTarget::Texture3D:
return GL_TEXTURE_3D;
- case SurfaceParams::SurfaceTarget::Texture1DArray:
+ case SurfaceTarget::Texture1DArray:
return GL_TEXTURE_1D_ARRAY;
- case SurfaceParams::SurfaceTarget::Texture2DArray:
+ case SurfaceTarget::Texture2DArray:
return GL_TEXTURE_2D_ARRAY;
- case SurfaceParams::SurfaceTarget::TextureCubemap:
+ case SurfaceTarget::TextureCubemap:
return GL_TEXTURE_CUBE_MAP;
}
LOG_CRITICAL(Render_OpenGL, "Unimplemented texture target={}", static_cast<u32>(target));
@@ -340,36 +348,19 @@ static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType
return format;
}
-MathUtil::Rectangle<u32> SurfaceParams::GetRect() const {
- u32 actual_height{unaligned_height};
+MathUtil::Rectangle<u32> SurfaceParams::GetRect(u32 mip_level) const {
+ u32 actual_height{std::max(1U, unaligned_height >> mip_level)};
if (IsPixelFormatASTC(pixel_format)) {
// ASTC formats must stop at the ATSC block size boundary
actual_height = Common::AlignDown(actual_height, GetASTCBlockSize(pixel_format).second);
}
- return {0, actual_height, width, 0};
-}
-
-/// Returns true if the specified PixelFormat is a BCn format, e.g. DXT or DXN
-static bool IsFormatBCn(PixelFormat format) {
- switch (format) {
- case PixelFormat::DXT1:
- case PixelFormat::DXT23:
- case PixelFormat::DXT45:
- case PixelFormat::DXN1:
- case PixelFormat::DXN2SNORM:
- case PixelFormat::DXN2UNORM:
- case PixelFormat::BC7U:
- case PixelFormat::BC6H_UF16:
- case PixelFormat::BC6H_SF16:
- return true;
- }
- return false;
+ return {0, actual_height, MipWidth(mip_level), 0};
}
template <bool morton_to_gl, PixelFormat format>
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::GetBytesPerPixel(format);
+ constexpr u32 bytes_per_pixel = GetBytesPerPixel(format);
// With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
// pixel values.
@@ -388,7 +379,7 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 d
}
using GLConversionArray = std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr),
- SurfaceParams::MaxPixelFormat>;
+ VideoCore::Surface::MaxPixelFormat>;
static constexpr GLConversionArray morton_to_gl_fns = {
// clang-format off
@@ -432,7 +423,7 @@ static constexpr GLConversionArray morton_to_gl_fns = {
MortonCopy<true, PixelFormat::RG16I>,
MortonCopy<true, PixelFormat::RG16S>,
MortonCopy<true, PixelFormat::RGB32F>,
- MortonCopy<true, PixelFormat::SRGBA8>,
+ MortonCopy<true, PixelFormat::RGBA8_SRGB>,
MortonCopy<true, PixelFormat::RG8U>,
MortonCopy<true, PixelFormat::RG8S>,
MortonCopy<true, PixelFormat::RG32UI>,
@@ -440,6 +431,15 @@ static constexpr GLConversionArray morton_to_gl_fns = {
MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
+ MortonCopy<true, PixelFormat::BGRA8_SRGB>,
+ MortonCopy<true, PixelFormat::DXT1_SRGB>,
+ MortonCopy<true, PixelFormat::DXT23_SRGB>,
+ MortonCopy<true, PixelFormat::DXT45_SRGB>,
+ MortonCopy<true, PixelFormat::BC7U_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
+ MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
MortonCopy<true, PixelFormat::Z32F>,
MortonCopy<true, PixelFormat::Z16>,
MortonCopy<true, PixelFormat::Z24S8>,
@@ -491,7 +491,7 @@ static constexpr GLConversionArray gl_to_morton_fns = {
MortonCopy<false, PixelFormat::RG16I>,
MortonCopy<false, PixelFormat::RG16S>,
MortonCopy<false, PixelFormat::RGB32F>,
- MortonCopy<false, PixelFormat::SRGBA8>,
+ MortonCopy<false, PixelFormat::RGBA8_SRGB>,
MortonCopy<false, PixelFormat::RG8U>,
MortonCopy<false, PixelFormat::RG8S>,
MortonCopy<false, PixelFormat::RG32UI>,
@@ -499,6 +499,15 @@ static constexpr GLConversionArray gl_to_morton_fns = {
nullptr,
nullptr,
nullptr,
+ MortonCopy<false, PixelFormat::BGRA8_SRGB>,
+ MortonCopy<false, PixelFormat::DXT1_SRGB>,
+ MortonCopy<false, PixelFormat::DXT23_SRGB>,
+ MortonCopy<false, PixelFormat::DXT45_SRGB>,
+ MortonCopy<false, PixelFormat::BC7U_SRGB>,
+ nullptr,
+ nullptr,
+ nullptr,
+ nullptr,
MortonCopy<false, PixelFormat::Z32F>,
MortonCopy<false, PixelFormat::Z16>,
MortonCopy<false, PixelFormat::Z24S8>,
@@ -508,28 +517,31 @@ static constexpr GLConversionArray gl_to_morton_fns = {
};
void SwizzleFunc(const GLConversionArray& functions, const SurfaceParams& params,
- std::vector<u8>& gl_buffer) {
- u32 depth = params.depth;
- if (params.target == SurfaceParams::SurfaceTarget::Texture2D) {
+ std::vector<u8>& gl_buffer, u32 mip_level) {
+ u32 depth = params.MipDepth(mip_level);
+ if (params.target == SurfaceTarget::Texture2D) {
// TODO(Blinkhawk): Eliminate this condition once all texture types are implemented.
depth = 1U;
}
if (params.is_layered) {
- u64 offset = 0;
+ u64 offset = params.GetMipmapLevelOffset(mip_level);
u64 offset_gl = 0;
u64 layer_size = params.LayerMemorySize();
- u64 gl_size = params.LayerSizeGL();
- for (u32 i = 0; i < depth; i++) {
+ u64 gl_size = params.LayerSizeGL(mip_level);
+ for (u32 i = 0; i < params.depth; i++) {
functions[static_cast<std::size_t>(params.pixel_format)](
- params.width, params.block_height, params.height, params.block_depth, 1,
+ params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
+ params.MipHeight(mip_level), params.MipBlockDepth(mip_level), 1,
gl_buffer.data() + offset_gl, gl_size, params.addr + offset);
offset += layer_size;
offset_gl += gl_size;
}
} else {
+ u64 offset = params.GetMipmapLevelOffset(mip_level);
functions[static_cast<std::size_t>(params.pixel_format)](
- params.width, params.block_height, params.height, params.block_depth, depth,
- gl_buffer.data(), gl_buffer.size(), params.addr);
+ params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
+ params.MipHeight(mip_level), params.MipBlockDepth(mip_level), depth, gl_buffer.data(),
+ gl_buffer.size(), params.addr + offset);
}
}
@@ -546,19 +558,21 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
OpenGLState state;
state.draw.read_framebuffer = read_fb_handle;
state.draw.draw_framebuffer = draw_fb_handle;
+ // Set sRGB enabled if the destination surfaces need it
+ state.framebuffer_srgb.enabled = dst_params.srgb_conversion;
state.Apply();
u32 buffers{};
if (src_params.type == SurfaceType::ColorTexture) {
switch (src_params.target) {
- case SurfaceParams::SurfaceTarget::Texture2D:
+ case SurfaceTarget::Texture2D:
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
GL_TEXTURE_2D, src_surface->Texture().handle, 0);
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
0, 0);
break;
- case SurfaceParams::SurfaceTarget::TextureCubemap:
+ case SurfaceTarget::TextureCubemap:
glFramebufferTexture2D(
GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face),
@@ -567,12 +581,12 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT,
static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face), 0, 0);
break;
- case SurfaceParams::SurfaceTarget::Texture2DArray:
+ case SurfaceTarget::Texture2DArray:
glFramebufferTextureLayer(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
src_surface->Texture().handle, 0, 0);
glFramebufferTextureLayer(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, 0, 0, 0);
break;
- case SurfaceParams::SurfaceTarget::Texture3D:
+ case SurfaceTarget::Texture3D:
glFramebufferTexture3D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + src_attachment,
SurfaceTargetToGL(src_params.target),
src_surface->Texture().handle, 0, 0);
@@ -588,13 +602,13 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
}
switch (dst_params.target) {
- case SurfaceParams::SurfaceTarget::Texture2D:
+ case SurfaceTarget::Texture2D:
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
GL_TEXTURE_2D, dst_surface->Texture().handle, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
0, 0);
break;
- case SurfaceParams::SurfaceTarget::TextureCubemap:
+ case SurfaceTarget::TextureCubemap:
glFramebufferTexture2D(
GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face),
@@ -603,13 +617,13 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT,
static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + cubemap_face), 0, 0);
break;
- case SurfaceParams::SurfaceTarget::Texture2DArray:
+ case SurfaceTarget::Texture2DArray:
glFramebufferTextureLayer(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
dst_surface->Texture().handle, 0, 0);
glFramebufferTextureLayer(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, 0, 0, 0);
break;
- case SurfaceParams::SurfaceTarget::Texture3D:
+ case SurfaceTarget::Texture3D:
glFramebufferTexture3D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + dst_attachment,
SurfaceTargetToGL(dst_params.target),
dst_surface->Texture().handle, 0, 0);
@@ -730,21 +744,21 @@ static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
UNREACHABLE();
} else {
switch (dst_params.target) {
- case SurfaceParams::SurfaceTarget::Texture1D:
+ case SurfaceTarget::Texture1D:
glTextureSubImage1D(dst_surface->Texture().handle, 0, 0, width, dest_format.format,
dest_format.type, nullptr);
break;
- case SurfaceParams::SurfaceTarget::Texture2D:
+ case SurfaceTarget::Texture2D:
glTextureSubImage2D(dst_surface->Texture().handle, 0, 0, 0, width, height,
dest_format.format, dest_format.type, nullptr);
break;
- case SurfaceParams::SurfaceTarget::Texture3D:
- case SurfaceParams::SurfaceTarget::Texture2DArray:
+ case SurfaceTarget::Texture3D:
+ case SurfaceTarget::Texture2DArray:
glTextureSubImage3D(dst_surface->Texture().handle, 0, 0, 0, 0, width, height,
static_cast<GLsizei>(dst_params.depth), dest_format.format,
dest_format.type, nullptr);
break;
- case SurfaceParams::SurfaceTarget::TextureCubemap:
+ case SurfaceTarget::TextureCubemap:
glTextureSubImage3D(dst_surface->Texture().handle, 0, 0, 0,
static_cast<GLint>(cubemap_face), width, height, 1,
dest_format.format, dest_format.type, nullptr);
@@ -781,35 +795,42 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
if (!format_tuple.compressed) {
// Only pre-create the texture for non-compressed textures.
switch (params.target) {
- case SurfaceParams::SurfaceTarget::Texture1D:
- glTexStorage1D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
- rect.GetWidth());
+ case SurfaceTarget::Texture1D:
+ glTexStorage1D(SurfaceTargetToGL(params.target), params.max_mip_level,
+ format_tuple.internal_format, rect.GetWidth());
break;
- case SurfaceParams::SurfaceTarget::Texture2D:
- case SurfaceParams::SurfaceTarget::TextureCubemap:
- glTexStorage2D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
- rect.GetWidth(), rect.GetHeight());
+ case SurfaceTarget::Texture2D:
+ case SurfaceTarget::TextureCubemap:
+ glTexStorage2D(SurfaceTargetToGL(params.target), params.max_mip_level,
+ format_tuple.internal_format, rect.GetWidth(), rect.GetHeight());
break;
- case SurfaceParams::SurfaceTarget::Texture3D:
- case SurfaceParams::SurfaceTarget::Texture2DArray:
- glTexStorage3D(SurfaceTargetToGL(params.target), 1, format_tuple.internal_format,
- rect.GetWidth(), rect.GetHeight(), params.depth);
+ case SurfaceTarget::Texture3D:
+ case SurfaceTarget::Texture2DArray:
+ glTexStorage3D(SurfaceTargetToGL(params.target), params.max_mip_level,
+ format_tuple.internal_format, rect.GetWidth(), rect.GetHeight(),
+ params.depth);
break;
default:
LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
static_cast<u32>(params.target));
UNREACHABLE();
- glTexStorage2D(GL_TEXTURE_2D, 1, format_tuple.internal_format, rect.GetWidth(),
- rect.GetHeight());
+ glTexStorage2D(GL_TEXTURE_2D, params.max_mip_level, format_tuple.internal_format,
+ rect.GetWidth(), rect.GetHeight());
}
}
glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+ glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
+ glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MAX_LEVEL,
+ params.max_mip_level - 1);
+ if (params.max_mip_level == 1) {
+ glTexParameterf(SurfaceTargetToGL(params.target), GL_TEXTURE_LOD_BIAS, 1000.0);
+ }
- VideoCore::LabelGLObject(GL_TEXTURE, texture.handle, params.addr,
- SurfaceParams::SurfaceTargetName(params.target));
+ LabelGLObject(GL_TEXTURE, texture.handle, params.addr,
+ SurfaceParams::SurfaceTargetName(params.target));
// Clamp size to mapped GPU memory region
// TODO(bunnei): Super Mario Odyssey maps a 0x40000 byte region and then uses it for a 0x80000
@@ -839,7 +860,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height, bo
S8Z24 s8z24_pixel{};
Z24S8 z24s8_pixel{};
- constexpr auto bpp{SurfaceParams::GetBytesPerPixel(PixelFormat::S8Z24)};
+ constexpr auto bpp{GetBytesPerPixel(PixelFormat::S8Z24)};
for (std::size_t y = 0; y < height; ++y) {
for (std::size_t x = 0; x < width; ++x) {
const std::size_t offset{bpp * (y * width + x)};
@@ -859,7 +880,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height, bo
}
static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
- constexpr auto bpp{SurfaceParams::GetBytesPerPixel(PixelFormat::G8R8U)};
+ constexpr auto bpp{GetBytesPerPixel(PixelFormat::G8R8U)};
for (std::size_t y = 0; y < height; ++y) {
for (std::size_t x = 0; x < width; ++x) {
const std::size_t offset{bpp * (y * width + x)};
@@ -881,7 +902,11 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
case PixelFormat::ASTC_2D_4X4:
case PixelFormat::ASTC_2D_8X8:
case PixelFormat::ASTC_2D_8X5:
- case PixelFormat::ASTC_2D_5X4: {
+ case PixelFormat::ASTC_2D_5X4:
+ case PixelFormat::ASTC_2D_4X4_SRGB:
+ case PixelFormat::ASTC_2D_8X8_SRGB:
+ case PixelFormat::ASTC_2D_8X5_SRGB:
+ case PixelFormat::ASTC_2D_5X4_SRGB: {
// Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC.
u32 block_width{};
u32 block_height{};
@@ -913,7 +938,9 @@ static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelForm
case PixelFormat::G8R8U:
case PixelFormat::G8R8S:
case PixelFormat::ASTC_2D_4X4:
- case PixelFormat::ASTC_2D_8X8: {
+ case PixelFormat::ASTC_2D_8X8:
+ case PixelFormat::ASTC_2D_4X4_SRGB:
+ case PixelFormat::ASTC_2D_8X8_SRGB: {
LOG_CRITICAL(HW_GPU, "Conversion of format {} after texture flushing is not implemented",
static_cast<u32>(pixel_format));
UNREACHABLE();
@@ -929,20 +956,22 @@ static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelForm
MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
void CachedSurface::LoadGLBuffer() {
MICROPROFILE_SCOPE(OpenGL_SurfaceLoad);
-
- gl_buffer.resize(params.size_in_bytes_gl);
+ gl_buffer.resize(params.max_mip_level);
+ for (u32 i = 0; i < params.max_mip_level; i++)
+ gl_buffer[i].resize(params.GetMipmapSizeGL(i));
if (params.is_tiled) {
ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
params.block_width, static_cast<u32>(params.target));
-
- SwizzleFunc(morton_to_gl_fns, params, gl_buffer);
+ for (u32 i = 0; i < params.max_mip_level; i++)
+ SwizzleFunc(morton_to_gl_fns, params, gl_buffer[i], i);
} else {
const auto texture_src_data{Memory::GetPointer(params.addr)};
const auto texture_src_data_end{texture_src_data + params.size_in_bytes_gl};
- gl_buffer.assign(texture_src_data, texture_src_data_end);
+ gl_buffer[0].assign(texture_src_data, texture_src_data_end);
}
-
- ConvertFormatAsNeeded_LoadGLBuffer(gl_buffer, params.pixel_format, params.width, params.height);
+ for (u32 i = 0; i < params.max_mip_level; i++)
+ ConvertFormatAsNeeded_LoadGLBuffer(gl_buffer[i], params.pixel_format, params.MipWidth(i),
+ params.MipHeight(i));
}
MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
@@ -952,18 +981,19 @@ void CachedSurface::FlushGLBuffer() {
ASSERT_MSG(!IsPixelFormatASTC(params.pixel_format), "Unimplemented");
// OpenGL temporary buffer needs to be big enough to store raw texture size
- gl_buffer.resize(GetSizeInBytes());
+ gl_buffer.resize(1);
+ gl_buffer[0].resize(GetSizeInBytes());
const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
// Ensure no bad interactions with GL_UNPACK_ALIGNMENT
- ASSERT(params.width * SurfaceParams::GetBytesPerPixel(params.pixel_format) % 4 == 0);
+ ASSERT(params.width * GetBytesPerPixel(params.pixel_format) % 4 == 0);
glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(params.width));
ASSERT(!tuple.compressed);
glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
- glGetTextureImage(texture.handle, 0, tuple.format, tuple.type, gl_buffer.size(),
- gl_buffer.data());
+ glGetTextureImage(texture.handle, 0, tuple.format, tuple.type,
+ static_cast<GLsizei>(gl_buffer[0].size()), gl_buffer[0].data());
glPixelStorei(GL_PACK_ROW_LENGTH, 0);
- ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer, params.pixel_format, params.width,
+ ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer[0], params.pixel_format, params.width,
params.height);
ASSERT(params.type != SurfaceType::Fill);
const u8* const texture_src_data = Memory::GetPointer(params.addr);
@@ -972,28 +1002,23 @@ void CachedSurface::FlushGLBuffer() {
ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
params.block_width, static_cast<u32>(params.target));
- SwizzleFunc(gl_to_morton_fns, params, gl_buffer);
+ SwizzleFunc(gl_to_morton_fns, params, gl_buffer[0], 0);
} else {
- std::memcpy(Memory::GetPointer(GetAddr()), gl_buffer.data(), GetSizeInBytes());
+ std::memcpy(Memory::GetPointer(GetAddr()), gl_buffer[0].data(), GetSizeInBytes());
}
}
-MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
-void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
- if (params.type == SurfaceType::Fill)
- return;
-
- MICROPROFILE_SCOPE(OpenGL_TextureUL);
-
- const auto& rect{params.GetRect()};
+void CachedSurface::UploadGLMipmapTexture(u32 mip_map, GLuint read_fb_handle,
+ GLuint draw_fb_handle) {
+ const auto& rect{params.GetRect(mip_map)};
// Load data from memory to the surface
const GLint x0 = static_cast<GLint>(rect.left);
const GLint y0 = static_cast<GLint>(rect.bottom);
std::size_t buffer_offset =
- static_cast<std::size_t>(static_cast<std::size_t>(y0) * params.width +
+ static_cast<std::size_t>(static_cast<std::size_t>(y0) * params.MipWidth(mip_map) +
static_cast<std::size_t>(x0)) *
- SurfaceParams::GetBytesPerPixel(params.pixel_format);
+ GetBytesPerPixel(params.pixel_format);
const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
const GLuint target_tex = texture.handle;
@@ -1009,88 +1034,116 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
cur_state.Apply();
// Ensure no bad interactions with GL_UNPACK_ALIGNMENT
- ASSERT(params.width * SurfaceParams::GetBytesPerPixel(params.pixel_format) % 4 == 0);
- glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(params.width));
+ ASSERT(params.MipWidth(mip_map) * GetBytesPerPixel(params.pixel_format) % 4 == 0);
+ glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(params.MipWidth(mip_map)));
+ GLsizei image_size = static_cast<GLsizei>(params.GetMipmapSizeGL(mip_map, false));
glActiveTexture(GL_TEXTURE0);
if (tuple.compressed) {
switch (params.target) {
- case SurfaceParams::SurfaceTarget::Texture2D:
- glCompressedTexImage2D(
- SurfaceTargetToGL(params.target), 0, tuple.internal_format,
- static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height), 0,
- static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
+ case SurfaceTarget::Texture2D:
+ glCompressedTexImage2D(SurfaceTargetToGL(params.target), mip_map, tuple.internal_format,
+ static_cast<GLsizei>(params.MipWidth(mip_map)),
+ static_cast<GLsizei>(params.MipHeight(mip_map)), 0, image_size,
+ &gl_buffer[mip_map][buffer_offset]);
+ break;
+ case SurfaceTarget::Texture3D:
+ glCompressedTexImage3D(SurfaceTargetToGL(params.target), mip_map, tuple.internal_format,
+ static_cast<GLsizei>(params.MipWidth(mip_map)),
+ static_cast<GLsizei>(params.MipHeight(mip_map)),
+ static_cast<GLsizei>(params.MipDepth(mip_map)), 0, image_size,
+ &gl_buffer[mip_map][buffer_offset]);
break;
- case SurfaceParams::SurfaceTarget::Texture3D:
- case SurfaceParams::SurfaceTarget::Texture2DArray:
- glCompressedTexImage3D(
- SurfaceTargetToGL(params.target), 0, tuple.internal_format,
- static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height),
- static_cast<GLsizei>(params.depth), 0,
- static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
+ case SurfaceTarget::Texture2DArray:
+ glCompressedTexImage3D(SurfaceTargetToGL(params.target), mip_map, tuple.internal_format,
+ static_cast<GLsizei>(params.MipWidth(mip_map)),
+ static_cast<GLsizei>(params.MipHeight(mip_map)),
+ static_cast<GLsizei>(params.depth), 0, image_size,
+ &gl_buffer[mip_map][buffer_offset]);
break;
- case SurfaceParams::SurfaceTarget::TextureCubemap:
+ case SurfaceTarget::TextureCubemap: {
+ GLsizei layer_size = static_cast<GLsizei>(params.LayerSizeGL(mip_map));
for (std::size_t face = 0; face < params.depth; ++face) {
glCompressedTexImage2D(static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face),
- 0, tuple.internal_format, static_cast<GLsizei>(params.width),
- static_cast<GLsizei>(params.height), 0,
- static_cast<GLsizei>(params.SizeInBytesCubeFaceGL()),
- &gl_buffer[buffer_offset]);
- buffer_offset += params.SizeInBytesCubeFace();
+ mip_map, tuple.internal_format,
+ static_cast<GLsizei>(params.MipWidth(mip_map)),
+ static_cast<GLsizei>(params.MipHeight(mip_map)), 0,
+ layer_size, &gl_buffer[mip_map][buffer_offset]);
+ buffer_offset += layer_size;
}
break;
+ }
default:
LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
static_cast<u32>(params.target));
UNREACHABLE();
- glCompressedTexImage2D(
- GL_TEXTURE_2D, 0, tuple.internal_format, static_cast<GLsizei>(params.width),
- static_cast<GLsizei>(params.height), 0,
- static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
+ glCompressedTexImage2D(GL_TEXTURE_2D, mip_map, tuple.internal_format,
+ static_cast<GLsizei>(params.MipWidth(mip_map)),
+ static_cast<GLsizei>(params.MipHeight(mip_map)), 0,
+ static_cast<GLsizei>(params.size_in_bytes_gl),
+ &gl_buffer[mip_map][buffer_offset]);
}
} else {
switch (params.target) {
- case SurfaceParams::SurfaceTarget::Texture1D:
- glTexSubImage1D(SurfaceTargetToGL(params.target), 0, x0,
+ case SurfaceTarget::Texture1D:
+ glTexSubImage1D(SurfaceTargetToGL(params.target), mip_map, x0,
static_cast<GLsizei>(rect.GetWidth()), tuple.format, tuple.type,
- &gl_buffer[buffer_offset]);
+ &gl_buffer[mip_map][buffer_offset]);
break;
- case SurfaceParams::SurfaceTarget::Texture2D:
- glTexSubImage2D(SurfaceTargetToGL(params.target), 0, x0, y0,
+ case SurfaceTarget::Texture2D:
+ glTexSubImage2D(SurfaceTargetToGL(params.target), mip_map, x0, y0,
static_cast<GLsizei>(rect.GetWidth()),
static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
- &gl_buffer[buffer_offset]);
+ &gl_buffer[mip_map][buffer_offset]);
break;
- case SurfaceParams::SurfaceTarget::Texture3D:
- case SurfaceParams::SurfaceTarget::Texture2DArray:
- glTexSubImage3D(SurfaceTargetToGL(params.target), 0, x0, y0, 0,
+ case SurfaceTarget::Texture3D:
+ glTexSubImage3D(SurfaceTargetToGL(params.target), mip_map, x0, y0, 0,
+ static_cast<GLsizei>(rect.GetWidth()),
+ static_cast<GLsizei>(rect.GetHeight()), params.MipDepth(mip_map),
+ tuple.format, tuple.type, &gl_buffer[mip_map][buffer_offset]);
+ break;
+ case SurfaceTarget::Texture2DArray:
+ glTexSubImage3D(SurfaceTargetToGL(params.target), mip_map, x0, y0, 0,
static_cast<GLsizei>(rect.GetWidth()),
static_cast<GLsizei>(rect.GetHeight()), params.depth, tuple.format,
- tuple.type, &gl_buffer[buffer_offset]);
+ tuple.type, &gl_buffer[mip_map][buffer_offset]);
break;
- case SurfaceParams::SurfaceTarget::TextureCubemap:
+ case SurfaceTarget::TextureCubemap: {
+ std::size_t start = buffer_offset;
for (std::size_t face = 0; face < params.depth; ++face) {
- glTexSubImage2D(static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face), 0, x0,
- y0, static_cast<GLsizei>(rect.GetWidth()),
+ glTexSubImage2D(static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face), mip_map,
+ x0, y0, static_cast<GLsizei>(rect.GetWidth()),
static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
- &gl_buffer[buffer_offset]);
- buffer_offset += params.SizeInBytesCubeFace();
+ &gl_buffer[mip_map][buffer_offset]);
+ buffer_offset += params.LayerSizeGL(mip_map);
}
break;
+ }
default:
LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
static_cast<u32>(params.target));
UNREACHABLE();
- glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
+ glTexSubImage2D(GL_TEXTURE_2D, mip_map, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
- &gl_buffer[buffer_offset]);
+ &gl_buffer[mip_map][buffer_offset]);
}
}
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
}
+MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
+void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
+ if (params.type == SurfaceType::Fill)
+ return;
+
+ MICROPROFILE_SCOPE(OpenGL_TextureUL);
+
+ for (u32 i = 0; i < params.max_mip_level; i++)
+ UploadGLMipmapTexture(i, read_fb_handle, draw_fb_handle);
+}
+
RasterizerCacheOpenGL::RasterizerCacheOpenGL() {
read_framebuffer.Create();
draw_framebuffer.Create();
@@ -1231,8 +1284,7 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
// For compatible surfaces, we can just do fast glCopyImageSubData based copy
if (old_params.target == new_params.target && old_params.type == new_params.type &&
old_params.depth == new_params.depth && old_params.depth == 1 &&
- SurfaceParams::GetFormatBpp(old_params.pixel_format) ==
- SurfaceParams::GetFormatBpp(new_params.pixel_format)) {
+ GetFormatBpp(old_params.pixel_format) == GetFormatBpp(new_params.pixel_format)) {
FastCopySurface(old_surface, new_surface);
return new_surface;
}
@@ -1245,15 +1297,15 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
const bool is_blit{old_params.pixel_format == new_params.pixel_format};
switch (new_params.target) {
- case SurfaceParams::SurfaceTarget::Texture2D:
+ case SurfaceTarget::Texture2D:
if (is_blit) {
BlitSurface(old_surface, new_surface, read_framebuffer.handle, draw_framebuffer.handle);
} else {
CopySurface(old_surface, new_surface, copy_pbo.handle);
}
break;
- case SurfaceParams::SurfaceTarget::TextureCubemap:
- case SurfaceParams::SurfaceTarget::Texture3D:
+ case SurfaceTarget::TextureCubemap:
+ case SurfaceTarget::Texture3D:
AccurateCopySurface(old_surface, new_surface);
break;
default:
@@ -1263,7 +1315,7 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
}
return new_surface;
-} // namespace OpenGL
+}
Surface RasterizerCacheOpenGL::TryFindFramebufferSurface(VAddr addr) const {
return TryGet(addr);
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index 181acfc68..f255f4419 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -7,6 +7,7 @@
#include <array>
#include <map>
#include <memory>
+#include <string>
#include <vector>
#include "common/alignment.h"
@@ -18,6 +19,7 @@
#include "video_core/rasterizer_cache.h"
#include "video_core/renderer_opengl/gl_resource_manager.h"
#include "video_core/renderer_opengl/gl_shader_gen.h"
+#include "video_core/surface.h"
#include "video_core/textures/decoders.h"
#include "video_core/textures/texture.h"
@@ -27,126 +29,12 @@ class CachedSurface;
using Surface = std::shared_ptr<CachedSurface>;
using SurfaceSurfaceRect_Tuple = std::tuple<Surface, Surface, MathUtil::Rectangle<u32>>;
-struct SurfaceParams {
- enum class PixelFormat {
- ABGR8U = 0,
- ABGR8S = 1,
- ABGR8UI = 2,
- B5G6R5U = 3,
- A2B10G10R10U = 4,
- A1B5G5R5U = 5,
- R8U = 6,
- R8UI = 7,
- RGBA16F = 8,
- RGBA16U = 9,
- RGBA16UI = 10,
- R11FG11FB10F = 11,
- RGBA32UI = 12,
- DXT1 = 13,
- DXT23 = 14,
- DXT45 = 15,
- DXN1 = 16, // This is also known as BC4
- DXN2UNORM = 17,
- DXN2SNORM = 18,
- BC7U = 19,
- BC6H_UF16 = 20,
- BC6H_SF16 = 21,
- ASTC_2D_4X4 = 22,
- G8R8U = 23,
- G8R8S = 24,
- BGRA8 = 25,
- RGBA32F = 26,
- RG32F = 27,
- R32F = 28,
- R16F = 29,
- R16U = 30,
- R16S = 31,
- R16UI = 32,
- R16I = 33,
- RG16 = 34,
- RG16F = 35,
- RG16UI = 36,
- RG16I = 37,
- RG16S = 38,
- RGB32F = 39,
- SRGBA8 = 40,
- RG8U = 41,
- RG8S = 42,
- RG32UI = 43,
- R32UI = 44,
- ASTC_2D_8X8 = 45,
- ASTC_2D_8X5 = 46,
- ASTC_2D_5X4 = 47,
-
- MaxColorFormat,
-
- // Depth formats
- Z32F = 48,
- Z16 = 49,
-
- MaxDepthFormat,
-
- // DepthStencil formats
- Z24S8 = 50,
- S8Z24 = 51,
- Z32FS8 = 52,
-
- MaxDepthStencilFormat,
-
- Max = MaxDepthStencilFormat,
- Invalid = 255,
- };
-
- static constexpr std::size_t MaxPixelFormat = static_cast<std::size_t>(PixelFormat::Max);
-
- enum class ComponentType {
- Invalid = 0,
- SNorm = 1,
- UNorm = 2,
- SInt = 3,
- UInt = 4,
- Float = 5,
- };
-
- enum class SurfaceType {
- ColorTexture = 0,
- Depth = 1,
- DepthStencil = 2,
- Fill = 3,
- Invalid = 4,
- };
-
- enum class SurfaceTarget {
- Texture1D,
- Texture2D,
- Texture3D,
- Texture1DArray,
- Texture2DArray,
- TextureCubemap,
- };
-
- static SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_type) {
- switch (texture_type) {
- case Tegra::Texture::TextureType::Texture1D:
- return SurfaceTarget::Texture1D;
- case Tegra::Texture::TextureType::Texture2D:
- case Tegra::Texture::TextureType::Texture2DNoMipmap:
- return SurfaceTarget::Texture2D;
- case Tegra::Texture::TextureType::Texture3D:
- return SurfaceTarget::Texture3D;
- case Tegra::Texture::TextureType::TextureCubemap:
- return SurfaceTarget::TextureCubemap;
- case Tegra::Texture::TextureType::Texture1DArray:
- return SurfaceTarget::Texture1DArray;
- case Tegra::Texture::TextureType::Texture2DArray:
- return SurfaceTarget::Texture2DArray;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented texture_type={}", static_cast<u32>(texture_type));
- UNREACHABLE();
- return SurfaceTarget::Texture2D;
- }
- }
+using SurfaceTarget = VideoCore::Surface::SurfaceTarget;
+using SurfaceType = VideoCore::Surface::SurfaceType;
+using PixelFormat = VideoCore::Surface::PixelFormat;
+using ComponentType = VideoCore::Surface::ComponentType;
+struct SurfaceParams {
static std::string SurfaceTargetName(SurfaceTarget target) {
switch (target) {
case SurfaceTarget::Texture1D:
@@ -168,631 +56,12 @@ struct SurfaceParams {
}
}
- static bool SurfaceTargetIsLayered(SurfaceTarget target) {
- switch (target) {
- case SurfaceTarget::Texture1D:
- case SurfaceTarget::Texture2D:
- case SurfaceTarget::Texture3D:
- return false;
- case SurfaceTarget::Texture1DArray:
- case SurfaceTarget::Texture2DArray:
- case SurfaceTarget::TextureCubemap:
- return true;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
- UNREACHABLE();
- return false;
- }
- }
-
- /**
- * Gets the compression factor for the specified PixelFormat. This applies to just the
- * "compressed width" and "compressed height", not the overall compression factor of a
- * compressed image. This is used for maintaining proper surface sizes for compressed
- * texture formats.
- */
- static constexpr u32 GetCompressionFactor(PixelFormat format) {
- if (format == PixelFormat::Invalid)
- return 0;
-
- constexpr std::array<u32, MaxPixelFormat> compression_factor_table = {{
- 1, // ABGR8U
- 1, // ABGR8S
- 1, // ABGR8UI
- 1, // B5G6R5U
- 1, // A2B10G10R10U
- 1, // A1B5G5R5U
- 1, // R8U
- 1, // R8UI
- 1, // RGBA16F
- 1, // RGBA16U
- 1, // RGBA16UI
- 1, // R11FG11FB10F
- 1, // RGBA32UI
- 4, // DXT1
- 4, // DXT23
- 4, // DXT45
- 4, // DXN1
- 4, // DXN2UNORM
- 4, // DXN2SNORM
- 4, // BC7U
- 4, // BC6H_UF16
- 4, // BC6H_SF16
- 4, // ASTC_2D_4X4
- 1, // G8R8U
- 1, // G8R8S
- 1, // BGRA8
- 1, // RGBA32F
- 1, // RG32F
- 1, // R32F
- 1, // R16F
- 1, // R16U
- 1, // R16S
- 1, // R16UI
- 1, // R16I
- 1, // RG16
- 1, // RG16F
- 1, // RG16UI
- 1, // RG16I
- 1, // RG16S
- 1, // RGB32F
- 1, // SRGBA8
- 1, // RG8U
- 1, // RG8S
- 1, // RG32UI
- 1, // R32UI
- 4, // ASTC_2D_8X8
- 4, // ASTC_2D_8X5
- 4, // ASTC_2D_5X4
- 1, // Z32F
- 1, // Z16
- 1, // Z24S8
- 1, // S8Z24
- 1, // Z32FS8
- }};
-
- ASSERT(static_cast<std::size_t>(format) < compression_factor_table.size());
- return compression_factor_table[static_cast<std::size_t>(format)];
- }
-
- static constexpr u32 GetDefaultBlockHeight(PixelFormat format) {
- if (format == PixelFormat::Invalid)
- return 0;
- constexpr std::array<u32, MaxPixelFormat> block_height_table = {{
- 1, // ABGR8U
- 1, // ABGR8S
- 1, // ABGR8UI
- 1, // B5G6R5U
- 1, // A2B10G10R10U
- 1, // A1B5G5R5U
- 1, // R8U
- 1, // R8UI
- 1, // RGBA16F
- 1, // RGBA16U
- 1, // RGBA16UI
- 1, // R11FG11FB10F
- 1, // RGBA32UI
- 4, // DXT1
- 4, // DXT23
- 4, // DXT45
- 4, // DXN1
- 4, // DXN2UNORM
- 4, // DXN2SNORM
- 4, // BC7U
- 4, // BC6H_UF16
- 4, // BC6H_SF16
- 4, // ASTC_2D_4X4
- 1, // G8R8U
- 1, // G8R8S
- 1, // BGRA8
- 1, // RGBA32F
- 1, // RG32F
- 1, // R32F
- 1, // R16F
- 1, // R16U
- 1, // R16S
- 1, // R16UI
- 1, // R16I
- 1, // RG16
- 1, // RG16F
- 1, // RG16UI
- 1, // RG16I
- 1, // RG16S
- 1, // RGB32F
- 1, // SRGBA8
- 1, // RG8U
- 1, // RG8S
- 1, // RG32UI
- 1, // R32UI
- 8, // ASTC_2D_8X8
- 5, // ASTC_2D_8X5
- 4, // ASTC_2D_5X4
- 1, // Z32F
- 1, // Z16
- 1, // Z24S8
- 1, // S8Z24
- 1, // Z32FS8
- }};
- ASSERT(static_cast<std::size_t>(format) < block_height_table.size());
- return block_height_table[static_cast<std::size_t>(format)];
- }
-
- static constexpr u32 GetFormatBpp(PixelFormat format) {
- if (format == PixelFormat::Invalid)
- return 0;
-
- constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
- 32, // ABGR8U
- 32, // ABGR8S
- 32, // ABGR8UI
- 16, // B5G6R5U
- 32, // A2B10G10R10U
- 16, // A1B5G5R5U
- 8, // R8U
- 8, // R8UI
- 64, // RGBA16F
- 64, // RGBA16U
- 64, // RGBA16UI
- 32, // R11FG11FB10F
- 128, // RGBA32UI
- 64, // DXT1
- 128, // DXT23
- 128, // DXT45
- 64, // DXN1
- 128, // DXN2UNORM
- 128, // DXN2SNORM
- 128, // BC7U
- 128, // BC6H_UF16
- 128, // BC6H_SF16
- 32, // ASTC_2D_4X4
- 16, // G8R8U
- 16, // G8R8S
- 32, // BGRA8
- 128, // RGBA32F
- 64, // RG32F
- 32, // R32F
- 16, // R16F
- 16, // R16U
- 16, // R16S
- 16, // R16UI
- 16, // R16I
- 32, // RG16
- 32, // RG16F
- 32, // RG16UI
- 32, // RG16I
- 32, // RG16S
- 96, // RGB32F
- 32, // SRGBA8
- 16, // RG8U
- 16, // RG8S
- 64, // RG32UI
- 32, // R32UI
- 16, // ASTC_2D_8X8
- 32, // ASTC_2D_8X5
- 32, // ASTC_2D_5X4
- 32, // Z32F
- 16, // Z16
- 32, // Z24S8
- 32, // S8Z24
- 64, // Z32FS8
- }};
-
- ASSERT(static_cast<std::size_t>(format) < bpp_table.size());
- return bpp_table[static_cast<std::size_t>(format)];
- }
-
u32 GetFormatBpp() const {
- return GetFormatBpp(pixel_format);
- }
-
- static PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
- switch (format) {
- case Tegra::DepthFormat::S8_Z24_UNORM:
- return PixelFormat::S8Z24;
- case Tegra::DepthFormat::Z24_S8_UNORM:
- return PixelFormat::Z24S8;
- case Tegra::DepthFormat::Z32_FLOAT:
- return PixelFormat::Z32F;
- case Tegra::DepthFormat::Z16_UNORM:
- return PixelFormat::Z16;
- case Tegra::DepthFormat::Z32_S8_X24_FLOAT:
- return PixelFormat::Z32FS8;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
- UNREACHABLE();
- }
- }
-
- static PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) {
- switch (format) {
- // TODO (Hexagon12): Converting SRGBA to RGBA is a hack and doesn't completely correct the
- // gamma.
- case Tegra::RenderTargetFormat::RGBA8_SRGB:
- case Tegra::RenderTargetFormat::RGBA8_UNORM:
- return PixelFormat::ABGR8U;
- case Tegra::RenderTargetFormat::RGBA8_SNORM:
- return PixelFormat::ABGR8S;
- case Tegra::RenderTargetFormat::RGBA8_UINT:
- return PixelFormat::ABGR8UI;
- case Tegra::RenderTargetFormat::BGRA8_SRGB:
- case Tegra::RenderTargetFormat::BGRA8_UNORM:
- return PixelFormat::BGRA8;
- case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
- return PixelFormat::A2B10G10R10U;
- case Tegra::RenderTargetFormat::RGBA16_FLOAT:
- return PixelFormat::RGBA16F;
- case Tegra::RenderTargetFormat::RGBA16_UNORM:
- return PixelFormat::RGBA16U;
- case Tegra::RenderTargetFormat::RGBA16_UINT:
- return PixelFormat::RGBA16UI;
- case Tegra::RenderTargetFormat::RGBA32_FLOAT:
- return PixelFormat::RGBA32F;
- case Tegra::RenderTargetFormat::RG32_FLOAT:
- return PixelFormat::RG32F;
- case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
- return PixelFormat::R11FG11FB10F;
- case Tegra::RenderTargetFormat::B5G6R5_UNORM:
- return PixelFormat::B5G6R5U;
- case Tegra::RenderTargetFormat::BGR5A1_UNORM:
- return PixelFormat::A1B5G5R5U;
- case Tegra::RenderTargetFormat::RGBA32_UINT:
- return PixelFormat::RGBA32UI;
- case Tegra::RenderTargetFormat::R8_UNORM:
- return PixelFormat::R8U;
- case Tegra::RenderTargetFormat::R8_UINT:
- return PixelFormat::R8UI;
- case Tegra::RenderTargetFormat::RG16_FLOAT:
- return PixelFormat::RG16F;
- case Tegra::RenderTargetFormat::RG16_UINT:
- return PixelFormat::RG16UI;
- case Tegra::RenderTargetFormat::RG16_SINT:
- return PixelFormat::RG16I;
- case Tegra::RenderTargetFormat::RG16_UNORM:
- 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:
- return PixelFormat::R16F;
- case Tegra::RenderTargetFormat::R16_UNORM:
- return PixelFormat::R16U;
- case Tegra::RenderTargetFormat::R16_SNORM:
- return PixelFormat::R16S;
- case Tegra::RenderTargetFormat::R16_UINT:
- return PixelFormat::R16UI;
- case Tegra::RenderTargetFormat::R16_SINT:
- 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();
- }
- }
-
- static PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
- Tegra::Texture::ComponentType component_type) {
- // TODO(Subv): Properly implement this
- switch (format) {
- case Tegra::Texture::TextureFormat::A8R8G8B8:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::ABGR8U;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::ABGR8S;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::ABGR8UI;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::B5G6R5:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::B5G6R5U;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::A2B10G10R10:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::A2B10G10R10U;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::A1B5G5R5:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::A1B5G5R5U;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::R8:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::R8U;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::R8UI;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::G8R8:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::G8R8U;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::G8R8S;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
- switch (component_type) {
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::RGBA16U;
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RGBA16F;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::BF10GF11RF11:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::R11FG11FB10F;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::R32_G32_B32_A32:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RGBA32F;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::RGBA32UI;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::R32_G32:
- 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:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::RGB32F;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::R16:
- switch (component_type) {
- case Tegra::Texture::ComponentType::FLOAT:
- return PixelFormat::R16F;
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::R16U;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::R16S;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::R16UI;
- case Tegra::Texture::ComponentType::SINT:
- return PixelFormat::R16I;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- case Tegra::Texture::TextureFormat::R32:
- 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::Z16:
- return PixelFormat::Z16;
- case Tegra::Texture::TextureFormat::Z24S8:
- return PixelFormat::Z24S8;
- case Tegra::Texture::TextureFormat::DXT1:
- return PixelFormat::DXT1;
- case Tegra::Texture::TextureFormat::DXT23:
- return PixelFormat::DXT23;
- case Tegra::Texture::TextureFormat::DXT45:
- 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::BC6H_UF16:
- return PixelFormat::BC6H_UF16;
- case Tegra::Texture::TextureFormat::BC6H_SF16:
- 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:
- return PixelFormat::RG16F;
- case Tegra::Texture::ComponentType::UNORM:
- return PixelFormat::RG16;
- case Tegra::Texture::ComponentType::SNORM:
- return PixelFormat::RG16S;
- case Tegra::Texture::ComponentType::UINT:
- return PixelFormat::RG16UI;
- case Tegra::Texture::ComponentType::SINT:
- return PixelFormat::RG16I;
- }
- LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
- static_cast<u32>(component_type));
- UNREACHABLE();
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}, component_type={}",
- static_cast<u32>(format), static_cast<u32>(component_type));
- UNREACHABLE();
- }
- }
-
- static ComponentType ComponentTypeFromTexture(Tegra::Texture::ComponentType type) {
- // TODO(Subv): Implement more component types
- switch (type) {
- case Tegra::Texture::ComponentType::UNORM:
- return ComponentType::UNorm;
- case Tegra::Texture::ComponentType::FLOAT:
- return ComponentType::Float;
- case Tegra::Texture::ComponentType::SNORM:
- return ComponentType::SNorm;
- case Tegra::Texture::ComponentType::UINT:
- return ComponentType::UInt;
- case Tegra::Texture::ComponentType::SINT:
- return ComponentType::SInt;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented component type={}", static_cast<u32>(type));
- UNREACHABLE();
- }
- }
-
- static ComponentType ComponentTypeFromRenderTarget(Tegra::RenderTargetFormat format) {
- // TODO(Subv): Implement more render targets
- switch (format) {
- case Tegra::RenderTargetFormat::RGBA8_UNORM:
- case Tegra::RenderTargetFormat::RGBA8_SRGB:
- case Tegra::RenderTargetFormat::BGRA8_UNORM:
- case Tegra::RenderTargetFormat::BGRA8_SRGB:
- case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
- case Tegra::RenderTargetFormat::R8_UNORM:
- case Tegra::RenderTargetFormat::RG16_UNORM:
- case Tegra::RenderTargetFormat::R16_UNORM:
- case Tegra::RenderTargetFormat::B5G6R5_UNORM:
- case Tegra::RenderTargetFormat::BGR5A1_UNORM:
- case Tegra::RenderTargetFormat::RG8_UNORM:
- case Tegra::RenderTargetFormat::RGBA16_UNORM:
- return ComponentType::UNorm;
- case Tegra::RenderTargetFormat::RGBA8_SNORM:
- case Tegra::RenderTargetFormat::RG16_SNORM:
- case Tegra::RenderTargetFormat::R16_SNORM:
- case Tegra::RenderTargetFormat::RG8_SNORM:
- return ComponentType::SNorm;
- case Tegra::RenderTargetFormat::RGBA16_FLOAT:
- case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
- case Tegra::RenderTargetFormat::RGBA32_FLOAT:
- case Tegra::RenderTargetFormat::RG32_FLOAT:
- case Tegra::RenderTargetFormat::RG16_FLOAT:
- case Tegra::RenderTargetFormat::R16_FLOAT:
- 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:
- case Tegra::RenderTargetFormat::RGBA8_UINT:
- return ComponentType::UInt;
- case Tegra::RenderTargetFormat::RG16_SINT:
- case Tegra::RenderTargetFormat::R16_SINT:
- return ComponentType::SInt;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
- UNREACHABLE();
- }
- }
-
- static PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
- switch (format) {
- case Tegra::FramebufferConfig::PixelFormat::ABGR8:
- return PixelFormat::ABGR8U;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
- UNREACHABLE();
- }
- }
-
- static ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
- switch (format) {
- case Tegra::DepthFormat::Z16_UNORM:
- case Tegra::DepthFormat::S8_Z24_UNORM:
- case Tegra::DepthFormat::Z24_S8_UNORM:
- return ComponentType::UNorm;
- case Tegra::DepthFormat::Z32_FLOAT:
- case Tegra::DepthFormat::Z32_S8_X24_FLOAT:
- return ComponentType::Float;
- default:
- LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
- UNREACHABLE();
- }
- }
-
- static SurfaceType GetFormatType(PixelFormat pixel_format) {
- if (static_cast<std::size_t>(pixel_format) <
- static_cast<std::size_t>(PixelFormat::MaxColorFormat)) {
- return SurfaceType::ColorTexture;
- }
-
- if (static_cast<std::size_t>(pixel_format) <
- static_cast<std::size_t>(PixelFormat::MaxDepthFormat)) {
- return SurfaceType::Depth;
- }
-
- if (static_cast<std::size_t>(pixel_format) <
- static_cast<std::size_t>(PixelFormat::MaxDepthStencilFormat)) {
- return SurfaceType::DepthStencil;
- }
-
- // TODO(Subv): Implement the other formats
- ASSERT(false);
-
- return SurfaceType::Invalid;
- }
-
- /// Returns the sizer in bytes of the specified pixel format
- static constexpr u32 GetBytesPerPixel(PixelFormat pixel_format) {
- if (pixel_format == SurfaceParams::PixelFormat::Invalid) {
- return 0;
- }
- return GetFormatBpp(pixel_format) / CHAR_BIT;
+ return VideoCore::Surface::GetFormatBpp(pixel_format);
}
/// Returns the rectangle corresponding to this surface
- MathUtil::Rectangle<u32> GetRect() const;
+ MathUtil::Rectangle<u32> GetRect(u32 mip_level = 0) const;
/// Returns the total size of this surface in bytes, adjusted for compression
std::size_t SizeInBytesRaw(bool ignore_tiled = false) const {
@@ -823,7 +92,7 @@ struct SurfaceParams {
/// Returns the exact size of memory occupied by the texture in VRAM, including mipmaps.
std::size_t MemorySize() const {
- std::size_t size = InnerMemorySize(is_layered);
+ std::size_t size = InnerMemorySize(false, is_layered);
if (is_layered)
return size * depth;
return size;
@@ -832,12 +101,78 @@ struct SurfaceParams {
/// Returns the exact size of the memory occupied by a layer in a texture in VRAM, including
/// mipmaps.
std::size_t LayerMemorySize() const {
- return InnerMemorySize(true);
+ return InnerMemorySize(false, true);
}
/// Returns the size of a layer of this surface in OpenGL.
- std::size_t LayerSizeGL() const {
- return SizeInBytesRaw(true) / depth;
+ std::size_t LayerSizeGL(u32 mip_level) const {
+ return InnerMipmapMemorySize(mip_level, true, is_layered, false);
+ }
+
+ std::size_t GetMipmapSizeGL(u32 mip_level, bool ignore_compressed = true) const {
+ std::size_t size = InnerMipmapMemorySize(mip_level, true, is_layered, ignore_compressed);
+ if (is_layered)
+ return size * depth;
+ return size;
+ }
+
+ std::size_t GetMipmapLevelOffset(u32 mip_level) const {
+ std::size_t offset = 0;
+ for (u32 i = 0; i < mip_level; i++)
+ offset += InnerMipmapMemorySize(i, false, is_layered);
+ return offset;
+ }
+
+ std::size_t GetMipmapLevelOffsetGL(u32 mip_level) const {
+ std::size_t offset = 0;
+ for (u32 i = 0; i < mip_level; i++)
+ offset += InnerMipmapMemorySize(i, true, is_layered);
+ return offset;
+ }
+
+ u32 MipWidth(u32 mip_level) const {
+ return std::max(1U, width >> mip_level);
+ }
+
+ u32 MipHeight(u32 mip_level) const {
+ return std::max(1U, height >> mip_level);
+ }
+
+ u32 MipDepth(u32 mip_level) const {
+ return std::max(1U, depth >> mip_level);
+ }
+
+ // Auto block resizing algorithm from:
+ // https://cgit.freedesktop.org/mesa/mesa/tree/src/gallium/drivers/nouveau/nv50/nv50_miptree.c
+ u32 MipBlockHeight(u32 mip_level) const {
+ if (mip_level == 0)
+ return block_height;
+ u32 alt_height = MipHeight(mip_level);
+ u32 h = GetDefaultBlockHeight(pixel_format);
+ u32 blocks_in_y = (alt_height + h - 1) / h;
+ u32 bh = 16;
+ while (bh > 1 && blocks_in_y <= bh * 4) {
+ bh >>= 1;
+ }
+ return bh;
+ }
+
+ u32 MipBlockDepth(u32 mip_level) const {
+ if (mip_level == 0)
+ return block_depth;
+ if (is_layered)
+ return 1;
+ u32 depth = MipDepth(mip_level);
+ u32 bd = 32;
+ while (bd > 1 && depth * 2 <= bd) {
+ bd >>= 1;
+ }
+ if (bd == 32) {
+ u32 bh = MipBlockHeight(mip_level);
+ if (bh >= 4)
+ return 16;
+ }
+ return bd;
}
/// Creates SurfaceParams from a texture configuration
@@ -881,7 +216,7 @@ struct SurfaceParams {
SurfaceTarget target;
u32 max_mip_level;
bool is_layered;
-
+ bool srgb_conversion;
// Parameters used for caching
VAddr addr;
Tegra::GPUVAddr gpu_addr;
@@ -898,7 +233,10 @@ struct SurfaceParams {
} rt;
private:
- std::size_t InnerMemorySize(bool layer_only = false) const;
+ std::size_t InnerMipmapMemorySize(u32 mip_level, bool force_gl = false, bool layer_only = false,
+ bool uncompressed = false) const;
+ std::size_t InnerMemorySize(bool force_gl = false, bool layer_only = false,
+ bool uncompressed = false) const;
};
}; // namespace OpenGL
@@ -960,8 +298,10 @@ public:
void UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle);
private:
+ void UploadGLMipmapTexture(u32 mip_map, GLuint read_fb_handle, GLuint draw_fb_handle);
+
OGLTexture texture;
- std::vector<u8> gl_buffer;
+ std::vector<std::vector<u8>> gl_buffer;
SurfaceParams params;
GLenum gl_target;
std::size_t cached_size_in_bytes;
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp
index 1a03a677f..9522fd344 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp
@@ -8,6 +8,7 @@
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_opengl/gl_shader_cache.h"
#include "video_core/renderer_opengl/gl_shader_manager.h"
+#include "video_core/renderer_opengl/utils.h"
#include "video_core/utils.h"
namespace OpenGL {
@@ -89,7 +90,7 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
shader.Create(program_result.first.c_str(), gl_type);
program.Create(true, shader.handle);
SetShaderUniformBlockBindings(program.handle);
- VideoCore::LabelGLObject(GL_PROGRAM, program.handle, addr);
+ LabelGLObject(GL_PROGRAM, program.handle, addr);
} else {
// Store shader's code to lazily build it on draw
geometry_programs.code = program_result.first;
@@ -130,7 +131,7 @@ GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program,
shader.Create(source.c_str(), GL_GEOMETRY_SHADER);
target_program.Create(true, shader.handle);
SetShaderUniformBlockBindings(target_program.handle);
- VideoCore::LabelGLObject(GL_PROGRAM, target_program.handle, addr, debug_name);
+ LabelGLObject(GL_PROGRAM, target_program.handle, addr, debug_name);
return target_program.handle;
};
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index dec291a7d..601c41f31 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -3,12 +3,12 @@
// Refer to the license.txt file included.
#include <map>
+#include <optional>
#include <set>
#include <string>
#include <string_view>
#include <unordered_set>
-#include <boost/optional.hpp>
#include <fmt/format.h>
#include "common/assert.h"
@@ -144,7 +144,7 @@ private:
for (u32 offset = begin; offset != end && offset != PROGRAM_END; ++offset) {
const Instruction instr = {program_code[offset]};
if (const auto opcode = OpCode::Decode(instr)) {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::EXIT: {
// The EXIT instruction can be predicated, which means that the shader can
// conditionally end on this instruction. We have to consider the case where the
@@ -341,10 +341,10 @@ public:
*/
void SetRegisterToFloat(const Register& reg, u64 elem, const std::string& value,
u64 dest_num_components, u64 value_num_components,
- bool is_saturated = false, u64 dest_elem = 0) {
+ bool is_saturated = false, u64 dest_elem = 0, bool precise = false) {
SetRegister(reg, elem, is_saturated ? "clamp(" + value + ", 0.0, 1.0)" : value,
- dest_num_components, value_num_components, dest_elem);
+ dest_num_components, value_num_components, dest_elem, precise);
}
/**
@@ -368,11 +368,12 @@ public:
const std::string func{is_signed ? "intBitsToFloat" : "uintBitsToFloat"};
SetRegister(reg, elem, func + '(' + ConvertIntegerSize(value, size) + ')',
- dest_num_components, value_num_components, dest_elem);
+ dest_num_components, value_num_components, dest_elem, false);
if (sets_cc) {
const std::string zero_condition = "( " + ConvertIntegerSize(value, size) + " == 0 )";
SetInternalFlag(InternalFlag::ZeroFlag, zero_condition);
+ LOG_WARNING(HW_GPU, "Control Codes Imcomplete.");
}
}
@@ -416,7 +417,7 @@ public:
}
}();
- SetRegister(reg, elem, result, dest_num_components, value_num_components, dest_elem);
+ SetRegister(reg, elem, result, dest_num_components, value_num_components, dest_elem, false);
}
/**
@@ -430,7 +431,7 @@ public:
*/
void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute,
const Tegra::Shader::IpaMode& input_mode,
- boost::optional<Register> vertex = {}) {
+ std::optional<Register> vertex = {}) {
const std::string dest = GetRegisterAsFloat(reg);
const std::string src = GetInputAttribute(attribute, input_mode, vertex) + GetSwizzle(elem);
shader.AddLine(dest + " = " + src + ';');
@@ -757,7 +758,8 @@ private:
* @param dest_elem Optional, the destination element to use for the operation.
*/
void SetRegister(const Register& reg, u64 elem, const std::string& value,
- u64 dest_num_components, u64 value_num_components, u64 dest_elem) {
+ u64 dest_num_components, u64 value_num_components, u64 dest_elem,
+ bool precise) {
if (reg == Register::ZeroIndex) {
LOG_CRITICAL(HW_GPU, "Cannot set Register::ZeroIndex");
UNREACHABLE();
@@ -774,7 +776,18 @@ private:
src += GetSwizzle(elem);
}
- shader.AddLine(dest + " = " + src + ';');
+ if (precise && stage != Maxwell3D::Regs::ShaderStage::Fragment) {
+ shader.AddLine('{');
+ ++shader.scope;
+ // This avoids optimizations of constant propagation and keeps the code as the original
+ // Sadly using the precise keyword causes "linking" errors on fragment shaders.
+ shader.AddLine("precise float tmp = " + src + ';');
+ shader.AddLine(dest + " = tmp;");
+ --shader.scope;
+ shader.AddLine('}');
+ } else {
+ shader.AddLine(dest + " = " + src + ';');
+ }
}
/// Build the GLSL register list.
@@ -795,10 +808,10 @@ private:
/// Generates code representing an input attribute register.
std::string GetInputAttribute(Attribute::Index attribute,
const Tegra::Shader::IpaMode& input_mode,
- boost::optional<Register> vertex = {}) {
+ std::optional<Register> vertex = {}) {
auto GeometryPass = [&](const std::string& name) {
if (stage == Maxwell3D::Regs::ShaderStage::Geometry && vertex) {
- return "gs_" + name + '[' + GetRegisterAsInteger(vertex.value(), 0, false) + ']';
+ return "gs_" + name + '[' + GetRegisterAsInteger(*vertex, 0, false) + ']';
}
return name;
};
@@ -1453,7 +1466,7 @@ private:
}
shader.AddLine(
- fmt::format("// {}: {} (0x{:016x})", offset, opcode->GetName(), instr.value));
+ fmt::format("// {}: {} (0x{:016x})", offset, opcode->get().GetName(), instr.value));
using Tegra::Shader::Pred;
ASSERT_MSG(instr.pred.full_pred != Pred::NeverExecute,
@@ -1461,7 +1474,7 @@ private:
// Some instructions (like SSY) don't have a predicate field, they are always
// unconditionally executed.
- bool can_be_predicated = OpCode::IsPredicatedInstruction(opcode->GetId());
+ bool can_be_predicated = OpCode::IsPredicatedInstruction(opcode->get().GetId());
if (can_be_predicated && instr.pred.pred_index != static_cast<u64>(Pred::UnusedIndex)) {
shader.AddLine("if (" +
@@ -1471,7 +1484,7 @@ private:
++shader.scope;
}
- switch (opcode->GetType()) {
+ switch (opcode->get().GetType()) {
case OpCode::Type::Arithmetic: {
std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
@@ -1488,7 +1501,7 @@ private:
}
}
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::MOV_C:
case OpCode::Id::MOV_R: {
// MOV does not have neither 'abs' nor 'neg' bits.
@@ -1510,8 +1523,13 @@ private:
ASSERT_MSG(instr.fmul.cc == 0, "FMUL cc is not implemented");
op_b = GetOperandAbsNeg(op_b, false, instr.fmul.negate_b);
+
regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b, 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "FMUL Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::FADD_C:
@@ -1519,8 +1537,13 @@ private:
case OpCode::Id::FADD_IMM: {
op_a = GetOperandAbsNeg(op_a, instr.alu.abs_a, instr.alu.negate_a);
op_b = GetOperandAbsNeg(op_b, instr.alu.abs_b, instr.alu.negate_b);
+
regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "FADD Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::MUFU: {
@@ -1528,31 +1551,31 @@ private:
switch (instr.sub_op) {
case SubOp::Cos:
regs.SetRegisterToFloat(instr.gpr0, 0, "cos(" + op_a + ')', 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
break;
case SubOp::Sin:
regs.SetRegisterToFloat(instr.gpr0, 0, "sin(" + op_a + ')', 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
break;
case SubOp::Ex2:
regs.SetRegisterToFloat(instr.gpr0, 0, "exp2(" + op_a + ')', 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
break;
case SubOp::Lg2:
regs.SetRegisterToFloat(instr.gpr0, 0, "log2(" + op_a + ')', 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
break;
case SubOp::Rcp:
regs.SetRegisterToFloat(instr.gpr0, 0, "1.0 / " + op_a, 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
break;
case SubOp::Rsq:
regs.SetRegisterToFloat(instr.gpr0, 0, "inversesqrt(" + op_a + ')', 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
break;
case SubOp::Sqrt:
regs.SetRegisterToFloat(instr.gpr0, 0, "sqrt(" + op_a + ')', 1, 1,
- instr.alu.saturate_d);
+ instr.alu.saturate_d, 0, true);
break;
default:
LOG_CRITICAL(HW_GPU, "Unhandled MUFU sub op: {0:x}",
@@ -1573,7 +1596,11 @@ private:
regs.SetRegisterToFloat(instr.gpr0, 0,
'(' + condition + ") ? min(" + parameters + ") : max(" +
parameters + ')',
- 1, 1);
+ 1, 1, false, 0, true);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "FMNMX Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::RRO_C:
@@ -1586,14 +1613,15 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: {}",
+ opcode->get().GetName());
UNREACHABLE();
}
}
break;
}
case OpCode::Type::ArithmeticImmediate: {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::MOV32_IMM: {
regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
break;
@@ -1602,7 +1630,11 @@ private:
regs.SetRegisterToFloat(instr.gpr0, 0,
regs.GetRegisterAsFloat(instr.gpr8) + " * " +
GetImmediate32(instr),
- 1, 1, instr.fmul32.saturate);
+ 1, 1, instr.fmul32.saturate, 0, true);
+ if (instr.op_32.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "FMUL32 Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::FADD32I: {
@@ -1625,7 +1657,11 @@ private:
op_b = "-(" + op_b + ')';
}
- regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1);
+ regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1, false, 0, true);
+ if (instr.op_32.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "FADD32 Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
}
@@ -1637,7 +1673,7 @@ private:
std::string op_a = instr.bfe.negate_a ? "-" : "";
op_a += regs.GetRegisterAsInteger(instr.gpr8);
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::BFE_IMM: {
std::string inner_shift =
'(' + op_a + " << " + std::to_string(instr.bfe.GetLeftShiftValue()) + ')';
@@ -1646,10 +1682,14 @@ private:
std::to_string(instr.bfe.GetLeftShiftValue() + instr.bfe.shift_position) + ')';
regs.SetRegisterToInteger(instr.gpr0, true, 0, outer_shift, 1, 1);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "BFE Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1671,7 +1711,7 @@ private:
}
}
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::SHR_C:
case OpCode::Id::SHR_R:
case OpCode::Id::SHR_IMM: {
@@ -1683,15 +1723,23 @@ private:
// Cast to int is superfluous for arithmetic shift, it's only for a logical shift
regs.SetRegisterToInteger(instr.gpr0, true, 0, "int(" + op_a + " >> " + op_b + ')',
1, 1);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "SHR Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::SHL_C:
case OpCode::Id::SHL_R:
case OpCode::Id::SHL_IMM:
regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " << " + op_b, 1, 1);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "SHL Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled shift instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled shift instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1701,13 +1749,17 @@ private:
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
std::string op_b = std::to_string(instr.alu.imm20_32.Value());
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::IADD32I:
if (instr.iadd32i.negate_a)
op_a = "-(" + op_a + ')';
regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
instr.iadd32i.saturate != 0);
+ if (instr.op_32.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "IADD32 Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
case OpCode::Id::LOP32I: {
if (instr.alu.lop32i.invert_a)
@@ -1719,11 +1771,15 @@ private:
WriteLogicOperation(instr.gpr0, instr.alu.lop32i.operation, op_a, op_b,
Tegra::Shader::PredicateResultMode::None,
Tegra::Shader::Pred::UnusedIndex);
+ if (instr.op_32.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "LOP32I Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
default: {
LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticIntegerImmediate instruction: {}",
- opcode->GetName());
+ opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1743,7 +1799,7 @@ private:
}
}
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::IADD_C:
case OpCode::Id::IADD_R:
case OpCode::Id::IADD_IMM: {
@@ -1755,6 +1811,10 @@ private:
regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
instr.alu.saturate_d);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "IADD Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::IADD3_C:
@@ -1779,7 +1839,7 @@ private:
}
};
- if (opcode->GetId() == OpCode::Id::IADD3_R) {
+ if (opcode->get().GetId() == OpCode::Id::IADD3_R) {
apply_height(instr.iadd3.height_a, op_a);
apply_height(instr.iadd3.height_b, op_b);
apply_height(instr.iadd3.height_c, op_c);
@@ -1795,7 +1855,7 @@ private:
op_c = "-(" + op_c + ')';
std::string result;
- if (opcode->GetId() == OpCode::Id::IADD3_R) {
+ if (opcode->get().GetId() == OpCode::Id::IADD3_R) {
switch (instr.iadd3.mode) {
case Tegra::Shader::IAdd3Mode::RightShift:
// TODO(tech4me): According to
@@ -1816,6 +1876,11 @@ private:
}
regs.SetRegisterToInteger(instr.gpr0, true, 0, result, 1, 1);
+
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "IADD3 Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::ISCADD_C:
@@ -1831,6 +1896,10 @@ private:
regs.SetRegisterToInteger(instr.gpr0, true, 0,
"((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "ISCADD Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::POPC_C:
@@ -1862,6 +1931,10 @@ private:
WriteLogicOperation(instr.gpr0, instr.alu.lop.operation, op_a, op_b,
instr.alu.lop.pred_result_mode, instr.alu.lop.pred48);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "LOP Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::LOP3_C:
@@ -1870,13 +1943,17 @@ private:
const std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
std::string lut;
- if (opcode->GetId() == OpCode::Id::LOP3_R) {
+ if (opcode->get().GetId() == OpCode::Id::LOP3_R) {
lut = '(' + std::to_string(instr.alu.lop3.GetImmLut28()) + ')';
} else {
lut = '(' + std::to_string(instr.alu.lop3.GetImmLut48()) + ')';
}
WriteLop3Instruction(instr.gpr0, op_a, op_b, op_c, lut);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "LOP3 Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::IMNMX_C:
@@ -1891,6 +1968,10 @@ private:
'(' + condition + ") ? min(" + parameters + ") : max(" +
parameters + ')',
1, 1);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "IMNMX Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::LEA_R2:
@@ -1900,7 +1981,7 @@ private:
case OpCode::Id::LEA_HI: {
std::string op_c;
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::LEA_R2: {
op_a = regs.GetRegisterAsInteger(instr.gpr20);
op_b = regs.GetRegisterAsInteger(instr.gpr39);
@@ -1945,7 +2026,8 @@ private:
op_b = regs.GetRegisterAsInteger(instr.gpr8);
op_a = std::to_string(instr.lea.imm.entry_a);
op_c = std::to_string(instr.lea.imm.entry_b);
- LOG_CRITICAL(HW_GPU, "Unhandled LEA subinstruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled LEA subinstruction: {}",
+ opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1960,7 +2042,7 @@ private:
}
default: {
LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticInteger instruction: {}",
- opcode->GetName());
+ opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1968,20 +2050,21 @@ private:
break;
}
case OpCode::Type::ArithmeticHalf: {
- if (opcode->GetId() == OpCode::Id::HADD2_C || opcode->GetId() == OpCode::Id::HADD2_R) {
+ if (opcode->get().GetId() == OpCode::Id::HADD2_C ||
+ opcode->get().GetId() == OpCode::Id::HADD2_R) {
ASSERT_MSG(instr.alu_half.ftz == 0, "Unimplemented");
}
const bool negate_a =
- opcode->GetId() != OpCode::Id::HMUL2_R && instr.alu_half.negate_a != 0;
+ opcode->get().GetId() != OpCode::Id::HMUL2_R && instr.alu_half.negate_a != 0;
const bool negate_b =
- opcode->GetId() != OpCode::Id::HMUL2_C && instr.alu_half.negate_b != 0;
+ opcode->get().GetId() != OpCode::Id::HMUL2_C && instr.alu_half.negate_b != 0;
const std::string op_a =
GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.alu_half.type_a,
instr.alu_half.abs_a != 0, negate_a);
std::string op_b;
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::HADD2_C:
case OpCode::Id::HMUL2_C:
op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
@@ -1999,7 +2082,7 @@ private:
op_b = GetHalfFloat(op_b, instr.alu_half.type_b, instr.alu_half.abs_b != 0, negate_b);
const std::string result = [&]() {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::HADD2_C:
case OpCode::Id::HADD2_R:
return '(' + op_a + " + " + op_b + ')';
@@ -2007,7 +2090,8 @@ private:
case OpCode::Id::HMUL2_R:
return '(' + op_a + " * " + op_b + ')';
default:
- LOG_CRITICAL(HW_GPU, "Unhandled half float instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled half float instruction: {}",
+ opcode->get().GetName());
UNREACHABLE();
return std::string("0");
}
@@ -2018,7 +2102,7 @@ private:
break;
}
case OpCode::Type::ArithmeticHalfImmediate: {
- if (opcode->GetId() == OpCode::Id::HADD2_IMM) {
+ if (opcode->get().GetId() == OpCode::Id::HADD2_IMM) {
ASSERT_MSG(instr.alu_half_imm.ftz == 0, "Unimplemented");
} else {
ASSERT_MSG(instr.alu_half_imm.precision == Tegra::Shader::HalfPrecision::None,
@@ -2032,7 +2116,7 @@ private:
const std::string op_b = UnpackHalfImmediate(instr, true);
const std::string result = [&]() {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::HADD2_IMM:
return op_a + " + " + op_b;
case OpCode::Id::HMUL2_IMM:
@@ -2058,7 +2142,7 @@ private:
ASSERT_MSG(instr.ffma.tab5980_1 == 0, "FFMA tab5980_1({}) not implemented",
instr.ffma.tab5980_1.Value());
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::FFMA_CR: {
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Float);
@@ -2082,24 +2166,29 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
- regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b + " + " + op_c, 1, 1,
- instr.alu.saturate_d);
+ regs.SetRegisterToFloat(instr.gpr0, 0, "fma(" + op_a + ", " + op_b + ", " + op_c + ')',
+ 1, 1, instr.alu.saturate_d, 0, true);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "FFMA Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
+
break;
}
case OpCode::Type::Hfma2: {
- if (opcode->GetId() == OpCode::Id::HFMA2_RR) {
+ if (opcode->get().GetId() == OpCode::Id::HFMA2_RR) {
ASSERT_MSG(instr.hfma2.rr.precision == Tegra::Shader::HalfPrecision::None,
"Unimplemented");
} else {
ASSERT_MSG(instr.hfma2.precision == Tegra::Shader::HalfPrecision::None,
"Unimplemented");
}
- const bool saturate = opcode->GetId() == OpCode::Id::HFMA2_RR
+ const bool saturate = opcode->get().GetId() == OpCode::Id::HFMA2_RR
? instr.hfma2.rr.saturate != 0
: instr.hfma2.saturate != 0;
@@ -2107,7 +2196,7 @@ private:
GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.hfma2.type_a);
std::string op_b, op_c;
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::HFMA2_CR:
op_b = GetHalfFloat(regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::UnsignedInteger),
@@ -2145,7 +2234,7 @@ private:
break;
}
case OpCode::Type::Conversion: {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::I2I_R: {
ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
@@ -2193,6 +2282,11 @@ private:
}
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
+
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "I2F Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::F2F_R: {
@@ -2231,6 +2325,11 @@ private:
}
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1, instr.alu.saturate_d);
+
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "F2F Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
case OpCode::Id::F2I_R:
@@ -2280,17 +2379,22 @@ private:
regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
1, false, 0, instr.conversion.dest_size);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "F2I Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled conversion instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled conversion instruction: {}",
+ opcode->get().GetName());
UNREACHABLE();
}
}
break;
}
case OpCode::Type::Memory: {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::LD_A: {
// Note: Shouldn't this be interp mode flat? As in no interpolation made.
ASSERT_MSG(instr.gpr8.Value() == Register::ZeroIndex,
@@ -2934,7 +3038,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3028,7 +3132,7 @@ private:
instr.hsetp2.abs_a, instr.hsetp2.negate_a);
const std::string op_b = [&]() {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::HSETP2_R:
return GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.hsetp2.type_b, instr.hsetp2.abs_a,
@@ -3087,10 +3191,15 @@ private:
regs.SetRegisterToFloat(instr.gpr0, 0, value, 1, 1);
}
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "PSET Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
+
break;
}
case OpCode::Type::PredicateSetPredicate: {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::PSETP: {
const std::string op_a =
GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
@@ -3136,7 +3245,8 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled predicate instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled predicate instruction: {}",
+ opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3224,7 +3334,7 @@ private:
instr.hset2.abs_a != 0, instr.hset2.negate_a != 0);
const std::string op_b = [&]() {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::HSET2_R:
return GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.hset2.type_b, instr.hset2.abs_b != 0,
@@ -3273,7 +3383,7 @@ private:
const bool is_signed{instr.xmad.sign_a == 1};
bool is_merge{};
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::XMAD_CR: {
is_merge = instr.xmad.merge_56;
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
@@ -3302,7 +3412,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled XMAD instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled XMAD instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3351,10 +3461,14 @@ private:
}
regs.SetRegisterToInteger(instr.gpr0, is_signed, 0, sum, 1, 1);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "XMAD Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
break;
}
default: {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::EXIT: {
if (stage == Maxwell3D::Regs::ShaderStage::Fragment) {
EmitFragmentOutputsWrite();
@@ -3522,6 +3636,11 @@ private:
regs.SetRegisterToInteger(instr.gpr0, result_signed, 1, result, 1, 1,
instr.vmad.saturate == 1, 0, Register::Size::Word,
instr.vmad.cc);
+ if (instr.generates_cc) {
+ LOG_CRITICAL(HW_GPU, "VMAD Generates an unhandled Control Code");
+ UNREACHABLE();
+ }
+
break;
}
case OpCode::Id::VSETP: {
@@ -3549,7 +3668,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3690,9 +3809,9 @@ std::string GetCommonDeclarations() {
RasterizerOpenGL::MaxConstbufferSize / sizeof(GLvec4));
}
-boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
- Maxwell3D::Regs::ShaderStage stage,
- const std::string& suffix) {
+std::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
+ Maxwell3D::Regs::ShaderStage stage,
+ const std::string& suffix) {
try {
const auto subroutines =
ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
@@ -3701,7 +3820,7 @@ boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code,
} catch (const DecompileFail& exception) {
LOG_ERROR(HW_GPU, "Shader decompilation failed: {}", exception.what());
}
- return boost::none;
+ return {};
}
} // namespace OpenGL::GLShader::Decompiler
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h
index b20cc4bfa..d01a4a7ee 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.h
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h
@@ -6,8 +6,8 @@
#include <array>
#include <functional>
+#include <optional>
#include <string>
-#include <boost/optional.hpp>
#include "common/common_types.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_opengl/gl_shader_gen.h"
@@ -18,8 +18,8 @@ using Tegra::Engines::Maxwell3D;
std::string GetCommonDeclarations();
-boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
- Maxwell3D::Regs::ShaderStage stage,
- const std::string& suffix);
+std::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
+ Maxwell3D::Regs::ShaderStage stage,
+ const std::string& suffix);
} // namespace OpenGL::GLShader::Decompiler
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index dfb562706..9d17edd63 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp
@@ -37,7 +37,7 @@ layout(std140) uniform vs_config {
ProgramResult program =
Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
Maxwell3D::Regs::ShaderStage::Vertex, "vertex")
- .get_value_or({});
+ .value_or(ProgramResult());
out += program.first;
@@ -45,7 +45,7 @@ layout(std140) uniform vs_config {
ProgramResult program_b =
Decompiler::DecompileProgram(setup.program.code_b, PROGRAM_OFFSET,
Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b")
- .get_value_or({});
+ .value_or(ProgramResult());
out += program_b.first;
}
@@ -90,7 +90,7 @@ ProgramResult GenerateGeometryShader(const ShaderSetup& setup) {
ProgramResult program =
Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
Maxwell3D::Regs::ShaderStage::Geometry, "geometry")
- .get_value_or({});
+ .value_or(ProgramResult());
out += R"(
out gl_PerVertex {
vec4 gl_Position;
@@ -124,7 +124,7 @@ ProgramResult GenerateFragmentShader(const ShaderSetup& setup) {
ProgramResult program =
Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
Maxwell3D::Regs::ShaderStage::Fragment, "fragment")
- .get_value_or({});
+ .value_or(ProgramResult());
out += R"(
layout(location = 0) out vec4 FragColor0;
layout(location = 1) out vec4 FragColor1;
diff --git a/src/video_core/renderer_opengl/gl_state.cpp b/src/video_core/renderer_opengl/gl_state.cpp
index f9d41ca24..d8a43cc94 100644
--- a/src/video_core/renderer_opengl/gl_state.cpp
+++ b/src/video_core/renderer_opengl/gl_state.cpp
@@ -11,9 +11,10 @@
namespace OpenGL {
OpenGLState OpenGLState::cur_state;
-
+bool OpenGLState::s_rgb_used;
OpenGLState::OpenGLState() {
// These all match default OpenGL values
+ framebuffer_srgb.enabled = false;
cull.enabled = false;
cull.mode = GL_BACK;
cull.front_face = GL_CCW;
@@ -89,6 +90,16 @@ OpenGLState::OpenGLState() {
}
void OpenGLState::Apply() const {
+ // sRGB
+ if (framebuffer_srgb.enabled != cur_state.framebuffer_srgb.enabled) {
+ if (framebuffer_srgb.enabled) {
+ // Track if sRGB is used
+ s_rgb_used = true;
+ glEnable(GL_FRAMEBUFFER_SRGB);
+ } else {
+ glDisable(GL_FRAMEBUFFER_SRGB);
+ }
+ }
// Culling
if (cull.enabled != cur_state.cull.enabled) {
if (cull.enabled) {
diff --git a/src/video_core/renderer_opengl/gl_state.h b/src/video_core/renderer_opengl/gl_state.h
index 4334b0d35..9e2c573b5 100644
--- a/src/video_core/renderer_opengl/gl_state.h
+++ b/src/video_core/renderer_opengl/gl_state.h
@@ -36,6 +36,10 @@ constexpr TextureUnit ProcTexDiffLUT{9};
class OpenGLState {
public:
struct {
+ bool enabled; // GL_FRAMEBUFFER_SRGB
+ } framebuffer_srgb;
+
+ struct {
bool enabled; // GL_CULL_FACE
GLenum mode; // GL_CULL_FACE_MODE
GLenum front_face; // GL_FRONT_FACE
@@ -161,7 +165,12 @@ public:
static OpenGLState GetCurState() {
return cur_state;
}
-
+ static bool GetsRGBUsed() {
+ return s_rgb_used;
+ }
+ static void ClearsRGBUsed() {
+ s_rgb_used = false;
+ }
/// Apply this state as the current OpenGL state
void Apply() const;
@@ -176,6 +185,9 @@ public:
private:
static OpenGLState cur_state;
+ // Workaround for sRGB problems caused by
+ // QT not supporting srgb output
+ static bool s_rgb_used;
};
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
index 0f6dcab2b..87d511c38 100644
--- a/src/video_core/renderer_opengl/maxwell_to_gl.h
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -135,12 +135,29 @@ inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
return {};
}
-inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode) {
+inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode,
+ Tegra::Texture::TextureMipmapFilter mip_filter_mode) {
switch (filter_mode) {
- case Tegra::Texture::TextureFilter::Linear:
- return GL_LINEAR;
- case Tegra::Texture::TextureFilter::Nearest:
- return GL_NEAREST;
+ case Tegra::Texture::TextureFilter::Linear: {
+ switch (mip_filter_mode) {
+ case Tegra::Texture::TextureMipmapFilter::None:
+ return GL_LINEAR;
+ case Tegra::Texture::TextureMipmapFilter::Nearest:
+ return GL_NEAREST_MIPMAP_LINEAR;
+ case Tegra::Texture::TextureMipmapFilter::Linear:
+ return GL_LINEAR_MIPMAP_LINEAR;
+ }
+ }
+ case Tegra::Texture::TextureFilter::Nearest: {
+ switch (mip_filter_mode) {
+ case Tegra::Texture::TextureMipmapFilter::None:
+ return GL_NEAREST;
+ case Tegra::Texture::TextureMipmapFilter::Nearest:
+ return GL_NEAREST_MIPMAP_NEAREST;
+ case Tegra::Texture::TextureMipmapFilter::Linear:
+ return GL_LINEAR_MIPMAP_NEAREST;
+ }
+ }
}
LOG_CRITICAL(Render_OpenGL, "Unimplemented texture filter mode={}",
static_cast<u32>(filter_mode));
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index 96d916b07..ea38da932 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -115,7 +115,8 @@ RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& window)
RendererOpenGL::~RendererOpenGL() = default;
/// Swap buffers (render frame)
-void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) {
+void RendererOpenGL::SwapBuffers(
+ std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) {
ScopeAcquireGLContext acquire_context{render_window};
Core::System::GetInstance().GetPerfStats().EndSystemFrame();
@@ -124,11 +125,11 @@ void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&
OpenGLState prev_state = OpenGLState::GetCurState();
state.Apply();
- if (framebuffer != boost::none) {
+ if (framebuffer) {
// If framebuffer is provided, reload it from memory to a texture
- if (screen_info.texture.width != (GLsizei)framebuffer->width ||
- screen_info.texture.height != (GLsizei)framebuffer->height ||
- screen_info.texture.pixel_format != framebuffer->pixel_format) {
+ if (screen_info.texture.width != (GLsizei)framebuffer->get().width ||
+ screen_info.texture.height != (GLsizei)framebuffer->get().height ||
+ screen_info.texture.pixel_format != framebuffer->get().pixel_format) {
// Reallocate texture if the framebuffer size has changed.
// This is expected to not happen very often and hence should not be a
// performance problem.
@@ -283,7 +284,8 @@ void RendererOpenGL::CreateRasterizer() {
if (rasterizer) {
return;
}
-
+ // Initialize sRGB Usage
+ OpenGLState::ClearsRGBUsed();
rasterizer = std::make_unique<RasterizerOpenGL>(render_window, screen_info);
}
@@ -356,13 +358,20 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
state.texture_units[0].texture = screen_info.display_texture;
state.texture_units[0].swizzle = {GL_RED, GL_GREEN, GL_BLUE, GL_ALPHA};
+ // Workaround brigthness problems in SMO by enabling sRGB in the final output
+ // if it has been used in the frame
+ // Needed because of this bug in QT
+ // QTBUG-50987
+ state.framebuffer_srgb.enabled = OpenGLState::GetsRGBUsed();
state.Apply();
-
glBufferSubData(GL_ARRAY_BUFFER, 0, sizeof(vertices), vertices.data());
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
-
+ // restore default state
+ state.framebuffer_srgb.enabled = false;
state.texture_units[0].texture = 0;
state.Apply();
+ // Clear sRGB state for the next frame
+ OpenGLState::ClearsRGBUsed();
}
/**
diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h
index 961467a62..c0868c0e4 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.h
+++ b/src/video_core/renderer_opengl/renderer_opengl.h
@@ -51,7 +51,8 @@ public:
~RendererOpenGL() override;
/// Swap buffers (render frame)
- void SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) override;
+ void SwapBuffers(
+ std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;
/// Initialize the renderer
bool Init() override;
diff --git a/src/video_core/renderer_opengl/utils.cpp b/src/video_core/renderer_opengl/utils.cpp
new file mode 100644
index 000000000..d84634cb3
--- /dev/null
+++ b/src/video_core/renderer_opengl/utils.cpp
@@ -0,0 +1,38 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <string>
+#include <fmt/format.h>
+#include <glad/glad.h>
+#include "common/common_types.h"
+#include "video_core/renderer_opengl/utils.h"
+
+namespace OpenGL {
+
+void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string extra_info) {
+ if (!GLAD_GL_KHR_debug) {
+ return; // We don't need to throw an error as this is just for debugging
+ }
+ const std::string nice_addr = fmt::format("0x{:016x}", addr);
+ std::string object_label;
+
+ if (extra_info.empty()) {
+ switch (identifier) {
+ case GL_TEXTURE:
+ object_label = "Texture@" + nice_addr;
+ break;
+ case GL_PROGRAM:
+ object_label = "Shader@" + nice_addr;
+ break;
+ default:
+ object_label = fmt::format("Object(0x{:x})@{}", identifier, nice_addr);
+ break;
+ }
+ } else {
+ object_label = extra_info + '@' + nice_addr;
+ }
+ glObjectLabel(identifier, handle, -1, static_cast<const GLchar*>(object_label.c_str()));
+}
+
+} // namespace OpenGL \ No newline at end of file
diff --git a/src/video_core/renderer_opengl/utils.h b/src/video_core/renderer_opengl/utils.h
new file mode 100644
index 000000000..1fcb6fc11
--- /dev/null
+++ b/src/video_core/renderer_opengl/utils.h
@@ -0,0 +1,15 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+#include <glad/glad.h>
+#include "common/common_types.h"
+
+namespace OpenGL {
+
+void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr, std::string extra_info = "");
+
+} // namespace OpenGL \ No newline at end of file
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
new file mode 100644
index 000000000..d9a97e30b
--- /dev/null
+++ b/src/video_core/surface.cpp
@@ -0,0 +1,499 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/common_types.h"
+#include "common/math_util.h"
+#include "video_core/surface.h"
+
+namespace VideoCore::Surface {
+
+SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_type) {
+ switch (texture_type) {
+ case Tegra::Texture::TextureType::Texture1D:
+ return SurfaceTarget::Texture1D;
+ case Tegra::Texture::TextureType::Texture2D:
+ case Tegra::Texture::TextureType::Texture2DNoMipmap:
+ return SurfaceTarget::Texture2D;
+ case Tegra::Texture::TextureType::Texture3D:
+ return SurfaceTarget::Texture3D;
+ case Tegra::Texture::TextureType::TextureCubemap:
+ return SurfaceTarget::TextureCubemap;
+ case Tegra::Texture::TextureType::Texture1DArray:
+ return SurfaceTarget::Texture1DArray;
+ case Tegra::Texture::TextureType::Texture2DArray:
+ return SurfaceTarget::Texture2DArray;
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented texture_type={}", static_cast<u32>(texture_type));
+ UNREACHABLE();
+ return SurfaceTarget::Texture2D;
+ }
+}
+
+bool SurfaceTargetIsLayered(SurfaceTarget target) {
+ switch (target) {
+ case SurfaceTarget::Texture1D:
+ case SurfaceTarget::Texture2D:
+ case SurfaceTarget::Texture3D:
+ return false;
+ case SurfaceTarget::Texture1DArray:
+ case SurfaceTarget::Texture2DArray:
+ case SurfaceTarget::TextureCubemap:
+ return true;
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
+ UNREACHABLE();
+ return false;
+ }
+}
+
+PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
+ switch (format) {
+ case Tegra::DepthFormat::S8_Z24_UNORM:
+ return PixelFormat::S8Z24;
+ case Tegra::DepthFormat::Z24_S8_UNORM:
+ return PixelFormat::Z24S8;
+ case Tegra::DepthFormat::Z32_FLOAT:
+ return PixelFormat::Z32F;
+ case Tegra::DepthFormat::Z16_UNORM:
+ return PixelFormat::Z16;
+ case Tegra::DepthFormat::Z32_S8_X24_FLOAT:
+ return PixelFormat::Z32FS8;
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
+ UNREACHABLE();
+ }
+}
+
+PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) {
+ switch (format) {
+ // TODO (Hexagon12): Converting SRGBA to RGBA is a hack and doesn't completely correct the
+ // gamma.
+ case Tegra::RenderTargetFormat::RGBA8_SRGB:
+ return PixelFormat::RGBA8_SRGB;
+ case Tegra::RenderTargetFormat::RGBA8_UNORM:
+ return PixelFormat::ABGR8U;
+ case Tegra::RenderTargetFormat::RGBA8_SNORM:
+ return PixelFormat::ABGR8S;
+ case Tegra::RenderTargetFormat::RGBA8_UINT:
+ return PixelFormat::ABGR8UI;
+ case Tegra::RenderTargetFormat::BGRA8_SRGB:
+ return PixelFormat::BGRA8_SRGB;
+ case Tegra::RenderTargetFormat::BGRA8_UNORM:
+ return PixelFormat::BGRA8;
+ case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
+ return PixelFormat::A2B10G10R10U;
+ case Tegra::RenderTargetFormat::RGBA16_FLOAT:
+ return PixelFormat::RGBA16F;
+ case Tegra::RenderTargetFormat::RGBA16_UNORM:
+ return PixelFormat::RGBA16U;
+ case Tegra::RenderTargetFormat::RGBA16_UINT:
+ return PixelFormat::RGBA16UI;
+ case Tegra::RenderTargetFormat::RGBA32_FLOAT:
+ return PixelFormat::RGBA32F;
+ case Tegra::RenderTargetFormat::RG32_FLOAT:
+ return PixelFormat::RG32F;
+ case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
+ return PixelFormat::R11FG11FB10F;
+ case Tegra::RenderTargetFormat::B5G6R5_UNORM:
+ return PixelFormat::B5G6R5U;
+ case Tegra::RenderTargetFormat::BGR5A1_UNORM:
+ return PixelFormat::A1B5G5R5U;
+ case Tegra::RenderTargetFormat::RGBA32_UINT:
+ return PixelFormat::RGBA32UI;
+ case Tegra::RenderTargetFormat::R8_UNORM:
+ return PixelFormat::R8U;
+ case Tegra::RenderTargetFormat::R8_UINT:
+ return PixelFormat::R8UI;
+ case Tegra::RenderTargetFormat::RG16_FLOAT:
+ return PixelFormat::RG16F;
+ case Tegra::RenderTargetFormat::RG16_UINT:
+ return PixelFormat::RG16UI;
+ case Tegra::RenderTargetFormat::RG16_SINT:
+ return PixelFormat::RG16I;
+ case Tegra::RenderTargetFormat::RG16_UNORM:
+ 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:
+ return PixelFormat::R16F;
+ case Tegra::RenderTargetFormat::R16_UNORM:
+ return PixelFormat::R16U;
+ case Tegra::RenderTargetFormat::R16_SNORM:
+ return PixelFormat::R16S;
+ case Tegra::RenderTargetFormat::R16_UINT:
+ return PixelFormat::R16UI;
+ case Tegra::RenderTargetFormat::R16_SINT:
+ 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();
+ }
+}
+
+PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
+ Tegra::Texture::ComponentType component_type,
+ bool is_srgb) {
+ // TODO(Subv): Properly implement this
+ switch (format) {
+ case Tegra::Texture::TextureFormat::A8R8G8B8:
+ if (is_srgb) {
+ return PixelFormat::RGBA8_SRGB;
+ }
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::ABGR8U;
+ case Tegra::Texture::ComponentType::SNORM:
+ return PixelFormat::ABGR8S;
+ case Tegra::Texture::ComponentType::UINT:
+ return PixelFormat::ABGR8UI;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::B5G6R5:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::B5G6R5U;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::A2B10G10R10:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::A2B10G10R10U;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::A1B5G5R5:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::A1B5G5R5U;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::R8:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::R8U;
+ case Tegra::Texture::ComponentType::UINT:
+ return PixelFormat::R8UI;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::G8R8:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::G8R8U;
+ case Tegra::Texture::ComponentType::SNORM:
+ return PixelFormat::G8R8S;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::RGBA16U;
+ case Tegra::Texture::ComponentType::FLOAT:
+ return PixelFormat::RGBA16F;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::BF10GF11RF11:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::FLOAT:
+ return PixelFormat::R11FG11FB10F;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::R32_G32_B32_A32:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::FLOAT:
+ return PixelFormat::RGBA32F;
+ case Tegra::Texture::ComponentType::UINT:
+ return PixelFormat::RGBA32UI;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::R32_G32:
+ 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:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::FLOAT:
+ return PixelFormat::RGB32F;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::R16:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::FLOAT:
+ return PixelFormat::R16F;
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::R16U;
+ case Tegra::Texture::ComponentType::SNORM:
+ return PixelFormat::R16S;
+ case Tegra::Texture::ComponentType::UINT:
+ return PixelFormat::R16UI;
+ case Tegra::Texture::ComponentType::SINT:
+ return PixelFormat::R16I;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ case Tegra::Texture::TextureFormat::R32:
+ 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::Z16:
+ return PixelFormat::Z16;
+ case Tegra::Texture::TextureFormat::Z24S8:
+ return PixelFormat::Z24S8;
+ case Tegra::Texture::TextureFormat::DXT1:
+ return is_srgb ? PixelFormat::DXT1_SRGB : PixelFormat::DXT1;
+ case Tegra::Texture::TextureFormat::DXT23:
+ return is_srgb ? PixelFormat::DXT23_SRGB : PixelFormat::DXT23;
+ case Tegra::Texture::TextureFormat::DXT45:
+ return is_srgb ? PixelFormat::DXT45_SRGB : 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 is_srgb ? PixelFormat::BC7U_SRGB : PixelFormat::BC7U;
+ case Tegra::Texture::TextureFormat::BC6H_UF16:
+ return PixelFormat::BC6H_UF16;
+ case Tegra::Texture::TextureFormat::BC6H_SF16:
+ return PixelFormat::BC6H_SF16;
+ case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
+ return is_srgb ? PixelFormat::ASTC_2D_4X4_SRGB : PixelFormat::ASTC_2D_4X4;
+ case Tegra::Texture::TextureFormat::ASTC_2D_5X4:
+ return is_srgb ? PixelFormat::ASTC_2D_5X4_SRGB : PixelFormat::ASTC_2D_5X4;
+ case Tegra::Texture::TextureFormat::ASTC_2D_8X8:
+ return is_srgb ? PixelFormat::ASTC_2D_8X8_SRGB : PixelFormat::ASTC_2D_8X8;
+ case Tegra::Texture::TextureFormat::ASTC_2D_8X5:
+ return is_srgb ? PixelFormat::ASTC_2D_8X5_SRGB : PixelFormat::ASTC_2D_8X5;
+ case Tegra::Texture::TextureFormat::R16_G16:
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::FLOAT:
+ return PixelFormat::RG16F;
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::RG16;
+ case Tegra::Texture::ComponentType::SNORM:
+ return PixelFormat::RG16S;
+ case Tegra::Texture::ComponentType::UINT:
+ return PixelFormat::RG16UI;
+ case Tegra::Texture::ComponentType::SINT:
+ return PixelFormat::RG16I;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}", static_cast<u32>(component_type));
+ UNREACHABLE();
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented format={}, component_type={}", static_cast<u32>(format),
+ static_cast<u32>(component_type));
+ UNREACHABLE();
+ }
+}
+
+ComponentType ComponentTypeFromTexture(Tegra::Texture::ComponentType type) {
+ // TODO(Subv): Implement more component types
+ switch (type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return ComponentType::UNorm;
+ case Tegra::Texture::ComponentType::FLOAT:
+ return ComponentType::Float;
+ case Tegra::Texture::ComponentType::SNORM:
+ return ComponentType::SNorm;
+ case Tegra::Texture::ComponentType::UINT:
+ return ComponentType::UInt;
+ case Tegra::Texture::ComponentType::SINT:
+ return ComponentType::SInt;
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented component type={}", static_cast<u32>(type));
+ UNREACHABLE();
+ }
+}
+
+ComponentType ComponentTypeFromRenderTarget(Tegra::RenderTargetFormat format) {
+ // TODO(Subv): Implement more render targets
+ switch (format) {
+ case Tegra::RenderTargetFormat::RGBA8_UNORM:
+ case Tegra::RenderTargetFormat::RGBA8_SRGB:
+ case Tegra::RenderTargetFormat::BGRA8_UNORM:
+ case Tegra::RenderTargetFormat::BGRA8_SRGB:
+ case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
+ case Tegra::RenderTargetFormat::R8_UNORM:
+ case Tegra::RenderTargetFormat::RG16_UNORM:
+ case Tegra::RenderTargetFormat::R16_UNORM:
+ case Tegra::RenderTargetFormat::B5G6R5_UNORM:
+ case Tegra::RenderTargetFormat::BGR5A1_UNORM:
+ case Tegra::RenderTargetFormat::RG8_UNORM:
+ case Tegra::RenderTargetFormat::RGBA16_UNORM:
+ return ComponentType::UNorm;
+ case Tegra::RenderTargetFormat::RGBA8_SNORM:
+ case Tegra::RenderTargetFormat::RG16_SNORM:
+ case Tegra::RenderTargetFormat::R16_SNORM:
+ case Tegra::RenderTargetFormat::RG8_SNORM:
+ return ComponentType::SNorm;
+ case Tegra::RenderTargetFormat::RGBA16_FLOAT:
+ case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
+ case Tegra::RenderTargetFormat::RGBA32_FLOAT:
+ case Tegra::RenderTargetFormat::RG32_FLOAT:
+ case Tegra::RenderTargetFormat::RG16_FLOAT:
+ case Tegra::RenderTargetFormat::R16_FLOAT:
+ 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:
+ case Tegra::RenderTargetFormat::RGBA8_UINT:
+ return ComponentType::UInt;
+ case Tegra::RenderTargetFormat::RG16_SINT:
+ case Tegra::RenderTargetFormat::R16_SINT:
+ return ComponentType::SInt;
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
+ UNREACHABLE();
+ }
+}
+
+PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
+ switch (format) {
+ case Tegra::FramebufferConfig::PixelFormat::ABGR8:
+ return PixelFormat::ABGR8U;
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
+ UNREACHABLE();
+ }
+}
+
+ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
+ switch (format) {
+ case Tegra::DepthFormat::Z16_UNORM:
+ case Tegra::DepthFormat::S8_Z24_UNORM:
+ case Tegra::DepthFormat::Z24_S8_UNORM:
+ return ComponentType::UNorm;
+ case Tegra::DepthFormat::Z32_FLOAT:
+ case Tegra::DepthFormat::Z32_S8_X24_FLOAT:
+ return ComponentType::Float;
+ default:
+ LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
+ UNREACHABLE();
+ }
+}
+
+SurfaceType GetFormatType(PixelFormat pixel_format) {
+ if (static_cast<std::size_t>(pixel_format) <
+ static_cast<std::size_t>(PixelFormat::MaxColorFormat)) {
+ return SurfaceType::ColorTexture;
+ }
+
+ if (static_cast<std::size_t>(pixel_format) <
+ static_cast<std::size_t>(PixelFormat::MaxDepthFormat)) {
+ return SurfaceType::Depth;
+ }
+
+ if (static_cast<std::size_t>(pixel_format) <
+ static_cast<std::size_t>(PixelFormat::MaxDepthStencilFormat)) {
+ return SurfaceType::DepthStencil;
+ }
+
+ // TODO(Subv): Implement the other formats
+ ASSERT(false);
+
+ return SurfaceType::Invalid;
+}
+
+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:
+ case PixelFormat::ASTC_2D_4X4_SRGB:
+ case PixelFormat::ASTC_2D_5X4_SRGB:
+ case PixelFormat::ASTC_2D_8X8_SRGB:
+ case PixelFormat::ASTC_2D_8X5_SRGB:
+ return true;
+ default:
+ return false;
+ }
+}
+
+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};
+ case PixelFormat::ASTC_2D_4X4_SRGB:
+ return {4, 4};
+ case PixelFormat::ASTC_2D_5X4_SRGB:
+ return {5, 4};
+ case PixelFormat::ASTC_2D_8X8_SRGB:
+ return {8, 8};
+ case PixelFormat::ASTC_2D_8X5_SRGB:
+ return {8, 5};
+ default:
+ LOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format));
+ UNREACHABLE();
+ }
+}
+
+bool IsFormatBCn(PixelFormat format) {
+ switch (format) {
+ case PixelFormat::DXT1:
+ case PixelFormat::DXT23:
+ case PixelFormat::DXT45:
+ case PixelFormat::DXN1:
+ case PixelFormat::DXN2SNORM:
+ case PixelFormat::DXN2UNORM:
+ case PixelFormat::BC7U:
+ case PixelFormat::BC6H_UF16:
+ case PixelFormat::BC6H_SF16:
+ case PixelFormat::DXT1_SRGB:
+ case PixelFormat::DXT23_SRGB:
+ case PixelFormat::DXT45_SRGB:
+ case PixelFormat::BC7U_SRGB:
+ return true;
+ }
+ return false;
+}
+
+} // namespace VideoCore::Surface
diff --git a/src/video_core/surface.h b/src/video_core/surface.h
new file mode 100644
index 000000000..3232e437f
--- /dev/null
+++ b/src/video_core/surface.h
@@ -0,0 +1,385 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <climits>
+#include <utility>
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "video_core/gpu.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCore::Surface {
+
+enum class PixelFormat {
+ ABGR8U = 0,
+ ABGR8S = 1,
+ ABGR8UI = 2,
+ B5G6R5U = 3,
+ A2B10G10R10U = 4,
+ A1B5G5R5U = 5,
+ R8U = 6,
+ R8UI = 7,
+ RGBA16F = 8,
+ RGBA16U = 9,
+ RGBA16UI = 10,
+ R11FG11FB10F = 11,
+ RGBA32UI = 12,
+ DXT1 = 13,
+ DXT23 = 14,
+ DXT45 = 15,
+ DXN1 = 16, // This is also known as BC4
+ DXN2UNORM = 17,
+ DXN2SNORM = 18,
+ BC7U = 19,
+ BC6H_UF16 = 20,
+ BC6H_SF16 = 21,
+ ASTC_2D_4X4 = 22,
+ G8R8U = 23,
+ G8R8S = 24,
+ BGRA8 = 25,
+ RGBA32F = 26,
+ RG32F = 27,
+ R32F = 28,
+ R16F = 29,
+ R16U = 30,
+ R16S = 31,
+ R16UI = 32,
+ R16I = 33,
+ RG16 = 34,
+ RG16F = 35,
+ RG16UI = 36,
+ RG16I = 37,
+ RG16S = 38,
+ RGB32F = 39,
+ RGBA8_SRGB = 40,
+ RG8U = 41,
+ RG8S = 42,
+ RG32UI = 43,
+ R32UI = 44,
+ ASTC_2D_8X8 = 45,
+ ASTC_2D_8X5 = 46,
+ ASTC_2D_5X4 = 47,
+ BGRA8_SRGB = 48,
+ DXT1_SRGB = 49,
+ DXT23_SRGB = 50,
+ DXT45_SRGB = 51,
+ BC7U_SRGB = 52,
+ ASTC_2D_4X4_SRGB = 53,
+ ASTC_2D_8X8_SRGB = 54,
+ ASTC_2D_8X5_SRGB = 55,
+ ASTC_2D_5X4_SRGB = 56,
+
+ MaxColorFormat,
+
+ // Depth formats
+ Z32F = 57,
+ Z16 = 58,
+
+ MaxDepthFormat,
+
+ // DepthStencil formats
+ Z24S8 = 59,
+ S8Z24 = 60,
+ Z32FS8 = 61,
+
+ MaxDepthStencilFormat,
+
+ Max = MaxDepthStencilFormat,
+ Invalid = 255,
+};
+
+static constexpr std::size_t MaxPixelFormat = static_cast<std::size_t>(PixelFormat::Max);
+
+enum class ComponentType {
+ Invalid = 0,
+ SNorm = 1,
+ UNorm = 2,
+ SInt = 3,
+ UInt = 4,
+ Float = 5,
+};
+
+enum class SurfaceType {
+ ColorTexture = 0,
+ Depth = 1,
+ DepthStencil = 2,
+ Fill = 3,
+ Invalid = 4,
+};
+
+enum class SurfaceTarget {
+ Texture1D,
+ Texture2D,
+ Texture3D,
+ Texture1DArray,
+ Texture2DArray,
+ TextureCubemap,
+};
+
+/**
+ * Gets the compression factor for the specified PixelFormat. This applies to just the
+ * "compressed width" and "compressed height", not the overall compression factor of a
+ * compressed image. This is used for maintaining proper surface sizes for compressed
+ * texture formats.
+ */
+static constexpr u32 GetCompressionFactor(PixelFormat format) {
+ if (format == PixelFormat::Invalid)
+ return 0;
+
+ constexpr std::array<u32, MaxPixelFormat> compression_factor_table = {{
+ 1, // ABGR8U
+ 1, // ABGR8S
+ 1, // ABGR8UI
+ 1, // B5G6R5U
+ 1, // A2B10G10R10U
+ 1, // A1B5G5R5U
+ 1, // R8U
+ 1, // R8UI
+ 1, // RGBA16F
+ 1, // RGBA16U
+ 1, // RGBA16UI
+ 1, // R11FG11FB10F
+ 1, // RGBA32UI
+ 4, // DXT1
+ 4, // DXT23
+ 4, // DXT45
+ 4, // DXN1
+ 4, // DXN2UNORM
+ 4, // DXN2SNORM
+ 4, // BC7U
+ 4, // BC6H_UF16
+ 4, // BC6H_SF16
+ 4, // ASTC_2D_4X4
+ 1, // G8R8U
+ 1, // G8R8S
+ 1, // BGRA8
+ 1, // RGBA32F
+ 1, // RG32F
+ 1, // R32F
+ 1, // R16F
+ 1, // R16U
+ 1, // R16S
+ 1, // R16UI
+ 1, // R16I
+ 1, // RG16
+ 1, // RG16F
+ 1, // RG16UI
+ 1, // RG16I
+ 1, // RG16S
+ 1, // RGB32F
+ 1, // RGBA8_SRGB
+ 1, // RG8U
+ 1, // RG8S
+ 1, // RG32UI
+ 1, // R32UI
+ 4, // ASTC_2D_8X8
+ 4, // ASTC_2D_8X5
+ 4, // ASTC_2D_5X4
+ 1, // BGRA8_SRGB
+ 4, // DXT1_SRGB
+ 4, // DXT23_SRGB
+ 4, // DXT45_SRGB
+ 4, // BC7U_SRGB
+ 4, // ASTC_2D_4X4_SRGB
+ 4, // ASTC_2D_8X8_SRGB
+ 4, // ASTC_2D_8X5_SRGB
+ 4, // ASTC_2D_5X4_SRGB
+ 1, // Z32F
+ 1, // Z16
+ 1, // Z24S8
+ 1, // S8Z24
+ 1, // Z32FS8
+ }};
+
+ ASSERT(static_cast<std::size_t>(format) < compression_factor_table.size());
+ return compression_factor_table[static_cast<std::size_t>(format)];
+}
+
+static constexpr u32 GetDefaultBlockHeight(PixelFormat format) {
+ if (format == PixelFormat::Invalid)
+ return 0;
+
+ constexpr std::array<u32, MaxPixelFormat> block_height_table = {{
+ 1, // ABGR8U
+ 1, // ABGR8S
+ 1, // ABGR8UI
+ 1, // B5G6R5U
+ 1, // A2B10G10R10U
+ 1, // A1B5G5R5U
+ 1, // R8U
+ 1, // R8UI
+ 1, // RGBA16F
+ 1, // RGBA16U
+ 1, // RGBA16UI
+ 1, // R11FG11FB10F
+ 1, // RGBA32UI
+ 4, // DXT1
+ 4, // DXT23
+ 4, // DXT45
+ 4, // DXN1
+ 4, // DXN2UNORM
+ 4, // DXN2SNORM
+ 4, // BC7U
+ 4, // BC6H_UF16
+ 4, // BC6H_SF16
+ 4, // ASTC_2D_4X4
+ 1, // G8R8U
+ 1, // G8R8S
+ 1, // BGRA8
+ 1, // RGBA32F
+ 1, // RG32F
+ 1, // R32F
+ 1, // R16F
+ 1, // R16U
+ 1, // R16S
+ 1, // R16UI
+ 1, // R16I
+ 1, // RG16
+ 1, // RG16F
+ 1, // RG16UI
+ 1, // RG16I
+ 1, // RG16S
+ 1, // RGB32F
+ 1, // RGBA8_SRGB
+ 1, // RG8U
+ 1, // RG8S
+ 1, // RG32UI
+ 1, // R32UI
+ 8, // ASTC_2D_8X8
+ 5, // ASTC_2D_8X5
+ 4, // ASTC_2D_5X4
+ 1, // BGRA8_SRGB
+ 4, // DXT1_SRGB
+ 4, // DXT23_SRGB
+ 4, // DXT45_SRGB
+ 4, // BC7U_SRGB
+ 4, // ASTC_2D_4X4_SRGB
+ 8, // ASTC_2D_8X8_SRGB
+ 5, // ASTC_2D_8X5_SRGB
+ 4, // ASTC_2D_5X4_SRGB
+ 1, // Z32F
+ 1, // Z16
+ 1, // Z24S8
+ 1, // S8Z24
+ 1, // Z32FS8
+ }};
+
+ ASSERT(static_cast<std::size_t>(format) < block_height_table.size());
+ return block_height_table[static_cast<std::size_t>(format)];
+}
+
+static constexpr u32 GetFormatBpp(PixelFormat format) {
+ if (format == PixelFormat::Invalid)
+ return 0;
+
+ constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
+ 32, // ABGR8U
+ 32, // ABGR8S
+ 32, // ABGR8UI
+ 16, // B5G6R5U
+ 32, // A2B10G10R10U
+ 16, // A1B5G5R5U
+ 8, // R8U
+ 8, // R8UI
+ 64, // RGBA16F
+ 64, // RGBA16U
+ 64, // RGBA16UI
+ 32, // R11FG11FB10F
+ 128, // RGBA32UI
+ 64, // DXT1
+ 128, // DXT23
+ 128, // DXT45
+ 64, // DXN1
+ 128, // DXN2UNORM
+ 128, // DXN2SNORM
+ 128, // BC7U
+ 128, // BC6H_UF16
+ 128, // BC6H_SF16
+ 32, // ASTC_2D_4X4
+ 16, // G8R8U
+ 16, // G8R8S
+ 32, // BGRA8
+ 128, // RGBA32F
+ 64, // RG32F
+ 32, // R32F
+ 16, // R16F
+ 16, // R16U
+ 16, // R16S
+ 16, // R16UI
+ 16, // R16I
+ 32, // RG16
+ 32, // RG16F
+ 32, // RG16UI
+ 32, // RG16I
+ 32, // RG16S
+ 96, // RGB32F
+ 32, // RGBA8_SRGB
+ 16, // RG8U
+ 16, // RG8S
+ 64, // RG32UI
+ 32, // R32UI
+ 16, // ASTC_2D_8X8
+ 16, // ASTC_2D_8X5
+ 32, // ASTC_2D_5X4
+ 32, // BGRA8_SRGB
+ 64, // DXT1_SRGB
+ 128, // DXT23_SRGB
+ 128, // DXT45_SRGB
+ 128, // BC7U
+ 32, // ASTC_2D_4X4_SRGB
+ 16, // ASTC_2D_8X8_SRGB
+ 16, // ASTC_2D_8X5_SRGB
+ 32, // ASTC_2D_5X4_SRGB
+ 32, // Z32F
+ 16, // Z16
+ 32, // Z24S8
+ 32, // S8Z24
+ 64, // Z32FS8
+ }};
+
+ ASSERT(static_cast<std::size_t>(format) < bpp_table.size());
+ return bpp_table[static_cast<std::size_t>(format)];
+}
+
+/// Returns the sizer in bytes of the specified pixel format
+static constexpr u32 GetBytesPerPixel(PixelFormat pixel_format) {
+ if (pixel_format == PixelFormat::Invalid) {
+ return 0;
+ }
+ return GetFormatBpp(pixel_format) / CHAR_BIT;
+}
+
+SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_type);
+
+bool SurfaceTargetIsLayered(SurfaceTarget target);
+
+PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format);
+
+PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format);
+
+PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
+ Tegra::Texture::ComponentType component_type,
+ bool is_srgb);
+
+ComponentType ComponentTypeFromTexture(Tegra::Texture::ComponentType type);
+
+ComponentType ComponentTypeFromRenderTarget(Tegra::RenderTargetFormat format);
+
+PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format);
+
+ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format);
+
+SurfaceType GetFormatType(PixelFormat pixel_format);
+
+bool IsPixelFormatASTC(PixelFormat format);
+
+std::pair<u32, u32> GetASTCBlockSize(PixelFormat format);
+
+/// Returns true if the specified PixelFormat is a BCn format, e.g. DXT or DXN
+bool IsFormatBCn(PixelFormat format);
+
+} // namespace VideoCore::Surface
diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h
index 4726f54a5..b390219e4 100644
--- a/src/video_core/textures/decoders.h
+++ b/src/video_core/textures/decoders.h
@@ -10,6 +10,12 @@
namespace Tegra::Texture {
+// GOBSize constant. Calculated by 64 bytes in x multiplied by 8 y coords, represents
+// an small rect of (64/bytes_per_pixel)X8.
+inline std::size_t GetGOBSize() {
+ return 512;
+}
+
/**
* Unswizzles a swizzled texture without changing its format.
*/
diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h
index 5947bd2b9..d12d2ecb8 100644
--- a/src/video_core/textures/texture.h
+++ b/src/video_core/textures/texture.h
@@ -173,6 +173,7 @@ struct TICEntry {
};
union {
BitField<0, 16, u32> width_minus_1;
+ BitField<22, 1, u32> srgb_conversion;
BitField<23, 4, TextureType> texture_type;
};
union {
@@ -227,6 +228,10 @@ struct TICEntry {
return header_version == TICHeaderVersion::BlockLinear ||
header_version == TICHeaderVersion::BlockLinearColorKey;
}
+
+ bool IsSrgbConversionEnabled() const {
+ return srgb_conversion != 0;
+ }
};
static_assert(sizeof(TICEntry) == 0x20, "TICEntry has wrong size");
diff --git a/src/video_core/utils.h b/src/video_core/utils.h
index 237cc1307..e0a14d48f 100644
--- a/src/video_core/utils.h
+++ b/src/video_core/utils.h
@@ -161,30 +161,4 @@ static inline void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixe
}
}
-static void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr,
- std::string extra_info = "") {
- if (!GLAD_GL_KHR_debug) {
- return; // We don't need to throw an error as this is just for debugging
- }
- const std::string nice_addr = fmt::format("0x{:016x}", addr);
- std::string object_label;
-
- if (extra_info.empty()) {
- switch (identifier) {
- case GL_TEXTURE:
- object_label = "Texture@" + nice_addr;
- break;
- case GL_PROGRAM:
- object_label = "Shader@" + nice_addr;
- break;
- default:
- object_label = fmt::format("Object(0x{:x})@{}", identifier, nice_addr);
- break;
- }
- } else {
- object_label = extra_info + '@' + nice_addr;
- }
- glObjectLabel(identifier, handle, -1, static_cast<const GLchar*>(object_label.c_str()));
-}
-
} // namespace VideoCore