summaryrefslogtreecommitdiffstats
path: root/src/video_core
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core')
-rw-r--r--src/video_core/CMakeLists.txt2
-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/shader_bytecode.h8
-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.cpp12
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h5
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.cpp243
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.h89
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.cpp5
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp112
-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/maxwell_to_gl.h27
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp11
-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/textures/decoders.h6
-rw-r--r--src/video_core/utils.h26
26 files changed, 426 insertions, 240 deletions
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index 09ecc5bad..c5f7128ec 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -51,6 +51,8 @@ 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
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/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index 141b9159b..b84da512f 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"
@@ -1456,7 +1455,7 @@ 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 +1463,8 @@ 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:
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 cb180b93c..bf381271e 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -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;
@@ -731,11 +731,15 @@ 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 b057e2efa..1d43a419d 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -16,6 +16,7 @@
#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/textures/astc.h"
#include "video_core/textures/decoders.h"
#include "video_core/utils.h"
@@ -90,27 +91,36 @@ 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(
@@ -188,7 +198,7 @@ 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
@@ -222,7 +232,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool layer_only) const {
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 = {};
@@ -249,7 +259,7 @@ 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());
@@ -273,7 +283,7 @@ 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
@@ -318,7 +328,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
{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_S3TC_DXT1_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
+ {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
@@ -373,13 +383,13 @@ 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};
+ return {0, actual_height, MipWidth(mip_level), 0};
}
/// Returns true if the specified PixelFormat is a BCn format, e.g. DXT or DXN
@@ -563,28 +573,31 @@ static constexpr GLConversionArray gl_to_morton_fns = {
};
void SwizzleFunc(const GLConversionArray& functions, const SurfaceParams& params,
- std::vector<u8>& gl_buffer) {
- u32 depth = params.depth;
+ std::vector<u8>& gl_buffer, u32 mip_level) {
+ u32 depth = params.MipDepth(mip_level);
if (params.target == SurfaceParams::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);
}
}
@@ -839,34 +852,41 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
// 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());
+ 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());
+ 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);
+ 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
@@ -992,20 +1012,22 @@ 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));
@@ -1015,7 +1037,8 @@ 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
@@ -1024,9 +1047,9 @@ void CachedSurface::FlushGLBuffer() {
ASSERT(!tuple.compressed);
glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
glGetTextureImage(texture.handle, 0, tuple.format, tuple.type,
- static_cast<GLsizei>(gl_buffer.size()), gl_buffer.data());
+ 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);
@@ -1035,26 +1058,21 @@ 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);
@@ -1072,88 +1090,117 @@ 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) * SurfaceParams::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]);
+ 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 SurfaceParams::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::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]);
+ 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 SurfaceParams::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,
+ 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,
+ 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:
+ 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 SurfaceParams::SurfaceTarget::Texture2DArray:
- glTexSubImage3D(SurfaceTargetToGL(params.target), 0, x0, y0, 0,
+ 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 SurfaceParams::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();
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index b4701a616..e72f4f2d2 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -395,7 +395,7 @@ struct SurfaceParams {
64, // RG32UI
32, // R32UI
16, // ASTC_2D_8X8
- 32, // ASTC_2D_8X5
+ 16, // ASTC_2D_8X5
32, // ASTC_2D_5X4
32, // BGRA8_SRGB
64, // DXT1_SRGB
@@ -404,7 +404,7 @@ struct SurfaceParams {
128, // BC7U
32, // ASTC_2D_4X4_SRGB
16, // ASTC_2D_8X8_SRGB
- 32, // ASTC_2D_8X5_SRGB
+ 16, // ASTC_2D_8X5_SRGB
32, // ASTC_2D_5X4_SRGB
32, // Z32F
16, // Z16
@@ -834,7 +834,7 @@ struct SurfaceParams {
}
/// 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 {
@@ -865,7 +865,7 @@ 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;
@@ -874,12 +874,78 @@ 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
@@ -940,7 +1006,10 @@ 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
@@ -1002,8 +1071,10 @@ 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 dcf6941b0..d1f6ffe40 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
@@ -430,7 +430,7 @@ 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 + ';');
@@ -807,10 +807,10 @@ 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;
};
@@ -1465,7 +1465,7 @@ 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,
@@ -1473,7 +1473,7 @@ 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 (" +
@@ -1483,7 +1483,7 @@ private:
++shader.scope;
}
- switch (opcode->GetType()) {
+ switch (opcode->get().GetType()) {
case OpCode::Type::Arithmetic: {
std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
@@ -1500,7 +1500,7 @@ 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.
@@ -1600,14 +1600,15 @@ 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;
@@ -1651,7 +1652,7 @@ 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()) + ')';
@@ -1663,7 +1664,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1685,7 +1686,7 @@ private:
}
}
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::SHR_C:
case OpCode::Id::SHR_R:
case OpCode::Id::SHR_IMM: {
@@ -1705,7 +1706,7 @@ private:
regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " << " + op_b, 1, 1);
break;
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled shift instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled shift instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1715,7 +1716,7 @@ 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 + ')';
@@ -1737,7 +1738,7 @@ private:
}
default: {
LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticIntegerImmediate instruction: {}",
- opcode->GetName());
+ opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1757,7 +1758,7 @@ private:
}
}
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::IADD_C:
case OpCode::Id::IADD_R:
case OpCode::Id::IADD_IMM: {
@@ -1793,7 +1794,7 @@ 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);
@@ -1809,7 +1810,7 @@ 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
@@ -1884,7 +1885,7 @@ 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()) + ')';
@@ -1914,7 +1915,7 @@ 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);
@@ -1959,7 +1960,8 @@ 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();
}
}
@@ -1974,7 +1976,7 @@ private:
}
default: {
LOG_CRITICAL(HW_GPU, "Unhandled ArithmeticInteger instruction: {}",
- opcode->GetName());
+ opcode->get().GetName());
UNREACHABLE();
}
}
@@ -1982,20 +1984,21 @@ 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,
@@ -2013,7 +2016,7 @@ 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 + ')';
@@ -2021,7 +2024,8 @@ 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");
}
@@ -2032,7 +2036,7 @@ 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,
@@ -2046,7 +2050,7 @@ 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:
@@ -2072,7 +2076,7 @@ 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);
@@ -2096,7 +2100,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -2107,14 +2111,14 @@ private:
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;
@@ -2122,7 +2126,7 @@ 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),
@@ -2160,7 +2164,7 @@ private:
break;
}
case OpCode::Type::Conversion: {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::I2I_R: {
ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
@@ -2298,14 +2302,15 @@ private:
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,
@@ -2949,7 +2954,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3043,7 +3048,7 @@ 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,
@@ -3105,7 +3110,7 @@ private:
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);
@@ -3151,7 +3156,8 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled predicate instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled predicate instruction: {}",
+ opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3239,7 +3245,7 @@ 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,
@@ -3288,7 +3294,7 @@ 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,
@@ -3317,7 +3323,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled XMAD instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled XMAD instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3369,7 +3375,7 @@ private:
break;
}
default: {
- switch (opcode->GetId()) {
+ switch (opcode->get().GetId()) {
case OpCode::Id::EXIT: {
if (stage == Maxwell3D::Regs::ShaderStage::Fragment) {
EmitFragmentOutputsWrite();
@@ -3564,7 +3570,7 @@ private:
break;
}
default: {
- LOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->GetName());
+ LOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->get().GetName());
UNREACHABLE();
}
}
@@ -3705,9 +3711,9 @@ 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();
@@ -3716,7 +3722,7 @@ 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/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 90b68943d..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.
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/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/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