hdkv/yuzu
1
Fork 0
Code Issues Pull Requests Releases 1 Activity

early-access version 2790

Browse Source
This commit is contained in:
pineappleEA
2022-06-16 03:46:18 +02:00
parent a67a0e1eb5
commit 518fdfccad
155 changed files with 9311 additions and 2805 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
src/video_core/renderer_opengl/gl_compute_pipeline.cpp
View File

@@ -28,12 +28,11 @@ bool ComputePipelineKey::operator==(const ComputePipelineKey& rhs) const noexcep
}
ComputePipeline::ComputePipeline(const Device& device, TextureCache& texture_cache_,
BufferCache& buffer_cache_, Tegra::MemoryManager& gpu_memory_,
Tegra::Engines::KeplerCompute& kepler_compute_,
ProgramManager& program_manager_, const Shader::Info& info_,
std::string code, std::vector<u32> code_v)
: texture_cache{texture_cache_}, buffer_cache{buffer_cache_}, gpu_memory{gpu_memory_},
kepler_compute{kepler_compute_}, program_manager{program_manager_}, info{info_} {
BufferCache& buffer_cache_, ProgramManager& program_manager_,
const Shader::Info& info_, std::string code,
std::vector<u32> code_v)
: texture_cache{texture_cache_}, buffer_cache{buffer_cache_},
program_manager{program_manager_}, info{info_} {
switch (device.GetShaderBackend()) {
case Settings::ShaderBackend::GLSL:
source_program = CreateProgram(code, GL_COMPUTE_SHADER);
@@ -86,7 +85,7 @@ void ComputePipeline::Configure() {
GLsizei texture_binding{};
GLsizei image_binding{};
const auto& qmd{kepler_compute.launch_description};
const auto& qmd{kepler_compute->launch_description};
const auto& cbufs{qmd.const_buffer_config};
const bool via_header_index{qmd.linked_tsc != 0};
const auto read_handle{[&](const auto& desc, u32 index) {
@@ -101,12 +100,13 @@ void ComputePipeline::Configure() {
const u32 secondary_offset{desc.secondary_cbuf_offset + index_offset};
const GPUVAddr separate_addr{cbufs[desc.secondary_cbuf_index].Address() +
secondary_offset};
const u32 lhs_raw{gpu_memory.Read<u32>(addr)};
const u32 rhs_raw{gpu_memory.Read<u32>(separate_addr)};
const u32 lhs_raw{gpu_memory->Read<u32>(addr) << desc.shift_left};
const u32 rhs_raw{gpu_memory->Read<u32>(separate_addr)
<< desc.secondary_shift_left};
return TexturePair(lhs_raw | rhs_raw, via_header_index);
}
}
return TexturePair(gpu_memory.Read<u32>(addr), via_header_index);
return TexturePair(gpu_memory->Read<u32>(addr), via_header_index);
}};
const auto add_image{[&](const auto& desc, bool blacklist) {
for (u32 index = 0; index < desc.count; ++index) {

16
src/video_core/renderer_opengl/gl_compute_pipeline.h
View File

@@ -49,10 +49,8 @@ static_assert(std::is_trivially_constructible_v<ComputePipelineKey>);
class ComputePipeline {
public:
explicit ComputePipeline(const Device& device, TextureCache& texture_cache_,
BufferCache& buffer_cache_, Tegra::MemoryManager& gpu_memory_,
Tegra::Engines::KeplerCompute& kepler_compute_,
ProgramManager& program_manager_, const Shader::Info& info_,
std::string code, std::vector<u32> code_v);
BufferCache& buffer_cache_, ProgramManager& program_manager_,
const Shader::Info& info_, std::string code, std::vector<u32> code_v);
void Configure();
@@ -60,11 +58,17 @@ public:
return writes_global_memory;
}
void SetEngine(Tegra::Engines::KeplerCompute* kepler_compute_,
Tegra::MemoryManager* gpu_memory_) {
kepler_compute = kepler_compute_;
gpu_memory = gpu_memory_;
}
private:
TextureCache& texture_cache;
BufferCache& buffer_cache;
Tegra::MemoryManager& gpu_memory;
Tegra::Engines::KeplerCompute& kepler_compute;
Tegra::MemoryManager* gpu_memory;
Tegra::Engines::KeplerCompute* kepler_compute;
ProgramManager& program_manager;
Shader::Info info;

13
src/video_core/renderer_opengl/gl_fence_manager.cpp
View File

@@ -10,10 +10,7 @@
namespace OpenGL {
GLInnerFence::GLInnerFence(u32 payload_, bool is_stubbed_) : FenceBase{payload_, is_stubbed_} {}
GLInnerFence::GLInnerFence(GPUVAddr address_, u32 payload_, bool is_stubbed_)
: FenceBase{address_, payload_, is_stubbed_} {}
GLInnerFence::GLInnerFence(bool is_stubbed_) : FenceBase{is_stubbed_} {}
GLInnerFence::~GLInnerFence() = default;
@@ -48,12 +45,8 @@ FenceManagerOpenGL::FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterize
BufferCache& buffer_cache_, QueryCache& query_cache_)
: GenericFenceManager{rasterizer_, gpu_, texture_cache_, buffer_cache_, query_cache_} {}
Fence FenceManagerOpenGL::CreateFence(u32 value, bool is_stubbed) {
return std::make_shared<GLInnerFence>(value, is_stubbed);
}
Fence FenceManagerOpenGL::CreateFence(GPUVAddr addr, u32 value, bool is_stubbed) {
return std::make_shared<GLInnerFence>(addr, value, is_stubbed);
Fence FenceManagerOpenGL::CreateFence(bool is_stubbed) {
return std::make_shared<GLInnerFence>(is_stubbed);
}
void FenceManagerOpenGL::QueueFence(Fence& fence) {

6
src/video_core/renderer_opengl/gl_fence_manager.h
View File

@@ -16,8 +16,7 @@ namespace OpenGL {
class GLInnerFence : public VideoCommon::FenceBase {
public:
explicit GLInnerFence(u32 payload_, bool is_stubbed_);
explicit GLInnerFence(GPUVAddr address_, u32 payload_, bool is_stubbed_);
explicit GLInnerFence(bool is_stubbed_);
~GLInnerFence();
void Queue();
@@ -40,8 +39,7 @@ public:
QueryCache& query_cache);
protected:
Fence CreateFence(u32 value, bool is_stubbed) override;
Fence CreateFence(GPUVAddr addr, u32 value, bool is_stubbed) override;
Fence CreateFence(bool is_stubbed) override;
void QueueFence(Fence& fence) override;
bool IsFenceSignaled(Fence& fence) const override;
void WaitFence(Fence& fence) override;

29
src/video_core/renderer_opengl/gl_graphics_pipeline.cpp
View File

@@ -169,15 +169,15 @@ ConfigureFuncPtr ConfigureFunc(const std::array<Shader::Info, 5>& infos, u32 ena
}
} // Anonymous namespace
GraphicsPipeline::GraphicsPipeline(
const Device& device, TextureCache& texture_cache_, BufferCache& buffer_cache_,
Tegra::MemoryManager& gpu_memory_, Tegra::Engines::Maxwell3D& maxwell3d_,
ProgramManager& program_manager_, StateTracker& state_tracker_, ShaderWorker* thread_worker,
VideoCore::ShaderNotify* shader_notify, std::array<std::string, 5> sources,
std::array<std::vector<u32>, 5> sources_spirv, const std::array<const Shader::Info*, 5>& infos,
const GraphicsPipelineKey& key_)
: texture_cache{texture_cache_}, buffer_cache{buffer_cache_},
gpu_memory{gpu_memory_}, maxwell3d{maxwell3d_}, program_manager{program_manager_},
GraphicsPipeline::GraphicsPipeline(const Device& device, TextureCache& texture_cache_,
BufferCache& buffer_cache_, ProgramManager& program_manager_,
StateTracker& state_tracker_, ShaderWorker* thread_worker,
VideoCore::ShaderNotify* shader_notify,
std::array<std::string, 5> sources,
std::array<std::vector<u32>, 5> sources_spirv,
const std::array<const Shader::Info*, 5>& infos,
const GraphicsPipelineKey& key_)
: texture_cache{texture_cache_}, buffer_cache{buffer_cache_}, program_manager{program_manager_},
state_tracker{state_tracker_}, key{key_} {
if (shader_notify) {
shader_notify->MarkShaderBuilding();
@@ -285,7 +285,7 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) {
buffer_cache.runtime.SetBaseStorageBindings(base_storage_bindings);
buffer_cache.runtime.SetEnableStorageBuffers(use_storage_buffers);
const auto& regs{maxwell3d.regs};
const auto& regs{maxwell3d->regs};
const bool via_header_index{regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex};
const auto config_stage{[&](size_t stage) LAMBDA_FORCEINLINE {
const Shader::Info& info{stage_infos[stage]};
@@ -299,7 +299,7 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) {
++ssbo_index;
}
}
const auto& cbufs{maxwell3d.state.shader_stages[stage].const_buffers};
const auto& cbufs{maxwell3d->state.shader_stages[stage].const_buffers};
const auto read_handle{[&](const auto& desc, u32 index) {
ASSERT(cbufs[desc.cbuf_index].enabled);
const u32 index_offset{index << desc.size_shift};
@@ -312,13 +312,14 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) {
const u32 second_offset{desc.secondary_cbuf_offset + index_offset};
const GPUVAddr separate_addr{cbufs[desc.secondary_cbuf_index].address +
second_offset};
const u32 lhs_raw{gpu_memory.Read<u32>(addr)};
const u32 rhs_raw{gpu_memory.Read<u32>(separate_addr)};
const u32 lhs_raw{gpu_memory->Read<u32>(addr) << desc.shift_left};
const u32 rhs_raw{gpu_memory->Read<u32>(separate_addr)
<< desc.secondary_shift_left};
const u32 raw{lhs_raw | rhs_raw};
return TexturePair(raw, via_header_index);
}
}
return TexturePair(gpu_memory.Read<u32>(addr), via_header_index);
return TexturePair(gpu_memory->Read<u32>(addr), via_header_index);
}};
const auto add_image{[&](const auto& desc, bool blacklist) LAMBDA_FORCEINLINE {
for (u32 index = 0; index < desc.count; ++index) {

16
src/video_core/renderer_opengl/gl_graphics_pipeline.h
View File

@@ -71,10 +71,9 @@ static_assert(std::is_trivially_constructible_v<GraphicsPipelineKey>);
class GraphicsPipeline {
public:
explicit GraphicsPipeline(const Device& device, TextureCache& texture_cache_,
BufferCache& buffer_cache_, Tegra::MemoryManager& gpu_memory_,
Tegra::Engines::Maxwell3D& maxwell3d_,
ProgramManager& program_manager_, StateTracker& state_tracker_,
ShaderWorker* thread_worker, VideoCore::ShaderNotify* shader_notify,
BufferCache& buffer_cache_, ProgramManager& program_manager_,
StateTracker& state_tracker_, ShaderWorker* thread_worker,
VideoCore::ShaderNotify* shader_notify,
std::array<std::string, 5> sources,
std::array<std::vector<u32>, 5> sources_spirv,
const std::array<const Shader::Info*, 5>& infos,
@@ -107,6 +106,11 @@ public:
};
}
void SetEngine(Tegra::Engines::Maxwell3D* maxwell3d_, Tegra::MemoryManager* gpu_memory_) {
maxwell3d = maxwell3d_;
gpu_memory = gpu_memory_;
}
private:
template <typename Spec>
void ConfigureImpl(bool is_indexed);
@@ -119,8 +123,8 @@ private:
TextureCache& texture_cache;
BufferCache& buffer_cache;
Tegra::MemoryManager& gpu_memory;
Tegra::Engines::Maxwell3D& maxwell3d;
Tegra::MemoryManager* gpu_memory;
Tegra::Engines::Maxwell3D* maxwell3d;
ProgramManager& program_manager;
StateTracker& state_tracker;
const GraphicsPipelineKey key;

5
src/video_core/renderer_opengl/gl_query_cache.cpp
View File

@@ -26,9 +26,8 @@ constexpr GLenum GetTarget(VideoCore::QueryType type) {
} // Anonymous namespace
QueryCache::QueryCache(RasterizerOpenGL& rasterizer_, Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::MemoryManager& gpu_memory_)
: QueryCacheBase(rasterizer_, maxwell3d_, gpu_memory_), gl_rasterizer{rasterizer_} {}
QueryCache::QueryCache(RasterizerOpenGL& rasterizer_)
: QueryCacheBase(rasterizer_), gl_rasterizer{rasterizer_} {}
QueryCache::~QueryCache() = default;

3
src/video_core/renderer_opengl/gl_query_cache.h
View File

@@ -28,8 +28,7 @@ using CounterStream = VideoCommon::CounterStreamBase<QueryCache, HostCounter>;
class QueryCache final
: public VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream, HostCounter> {
public:
explicit QueryCache(RasterizerOpenGL& rasterizer_, Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::MemoryManager& gpu_memory_);
explicit QueryCache(RasterizerOpenGL& rasterizer_);
~QueryCache();
OGLQuery AllocateQuery(VideoCore::QueryType type);

217
src/video_core/renderer_opengl/gl_rasterizer.cpp
View File

@@ -17,7 +17,7 @@
#include "common/microprofile.h"
#include "common/scope_exit.h"
#include "common/settings.h"
#include "video_core/control/channel_state.h"
#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/memory_manager.h"
@@ -57,22 +57,20 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra
Core::Memory::Memory& cpu_memory_, const Device& device_,
ScreenInfo& screen_info_, ProgramManager& program_manager_,
StateTracker& state_tracker_)
: RasterizerAccelerated(cpu_memory_), gpu(gpu_), maxwell3d(gpu.Maxwell3D()),
kepler_compute(gpu.KeplerCompute()), gpu_memory(gpu.MemoryManager()), device(device_),
screen_info(screen_info_), program_manager(program_manager_), state_tracker(state_tracker_),
: RasterizerAccelerated(cpu_memory_), gpu(gpu_), device(device_), screen_info(screen_info_),
program_manager(program_manager_), state_tracker(state_tracker_),
texture_cache_runtime(device, program_manager, state_tracker),
texture_cache(texture_cache_runtime, *this, maxwell3d, kepler_compute, gpu_memory),
buffer_cache_runtime(device),
buffer_cache(*this, maxwell3d, kepler_compute, gpu_memory, cpu_memory_, buffer_cache_runtime),
shader_cache(*this, emu_window_, maxwell3d, kepler_compute, gpu_memory, device, texture_cache,
buffer_cache, program_manager, state_tracker, gpu.ShaderNotify()),
query_cache(*this, maxwell3d, gpu_memory), accelerate_dma(buffer_cache),
texture_cache(texture_cache_runtime, *this), buffer_cache_runtime(device),
buffer_cache(*this, cpu_memory_, buffer_cache_runtime),
shader_cache(*this, emu_window_, device, texture_cache, buffer_cache, program_manager,
state_tracker, gpu.ShaderNotify()),
query_cache(*this), accelerate_dma(buffer_cache),
fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache) {}
RasterizerOpenGL::~RasterizerOpenGL() = default;
void RasterizerOpenGL::SyncVertexFormats() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::VertexFormats]) {
return;
}
@@ -90,7 +88,7 @@ void RasterizerOpenGL::SyncVertexFormats() {
}
flags[Dirty::VertexFormat0 + index] = false;
const auto attrib = maxwell3d.regs.vertex_attrib_format[index];
const auto attrib = maxwell3d->regs.vertex_attrib_format[index];
const auto gl_index = static_cast<GLuint>(index);
// Disable constant attributes.
@@ -114,13 +112,13 @@ void RasterizerOpenGL::SyncVertexFormats() {
}
void RasterizerOpenGL::SyncVertexInstances() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::VertexInstances]) {
return;
}
flags[Dirty::VertexInstances] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_ATTRIBUTES; ++index) {
if (!flags[Dirty::VertexInstance0 + index]) {
continue;
@@ -141,11 +139,11 @@ void RasterizerOpenGL::LoadDiskResources(u64 title_id, std::stop_token stop_load
void RasterizerOpenGL::Clear() {
MICROPROFILE_SCOPE(OpenGL_Clears);
if (!maxwell3d.ShouldExecute()) {
if (!maxwell3d->ShouldExecute()) {
return;
}
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
bool use_color{};
bool use_depth{};
bool use_stencil{};
@@ -218,22 +216,26 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
if (!pipeline) {
return;
}
gpu.TickWork();
std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
pipeline->SetEngine(maxwell3d, gpu_memory);
pipeline->Configure(is_indexed);
SyncState();
const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(maxwell3d.regs.draw.topology);
const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(maxwell3d->regs.draw.topology);
BeginTransformFeedback(pipeline, primitive_mode);
const GLuint base_instance = static_cast<GLuint>(maxwell3d.regs.vb_base_instance);
const GLuint base_instance = static_cast<GLuint>(maxwell3d->regs.vb_base_instance);
const GLsizei num_instances =
static_cast<GLsizei>(is_instanced ? maxwell3d.mme_draw.instance_count : 1);
static_cast<GLsizei>(is_instanced ? maxwell3d->mme_draw.instance_count : 1);
if (is_indexed) {
const GLint base_vertex = static_cast<GLint>(maxwell3d.regs.vb_element_base);
const GLsizei num_vertices = static_cast<GLsizei>(maxwell3d.regs.index_array.count);
const GLint base_vertex = static_cast<GLint>(maxwell3d->regs.vb_element_base);
const GLsizei num_vertices = static_cast<GLsizei>(maxwell3d->regs.index_array.count);
const GLvoid* const offset = buffer_cache_runtime.IndexOffset();
const GLenum format = MaxwellToGL::IndexFormat(maxwell3d.regs.index_array.format);
const GLenum format = MaxwellToGL::IndexFormat(maxwell3d->regs.index_array.format);
if (num_instances == 1 && base_instance == 0 && base_vertex == 0) {
glDrawElements(primitive_mode, num_vertices, format, offset);
} else if (num_instances == 1 && base_instance == 0) {
@@ -252,8 +254,8 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
base_instance);
}
} else {
const GLint base_vertex = static_cast<GLint>(maxwell3d.regs.vertex_buffer.first);
const GLsizei num_vertices = static_cast<GLsizei>(maxwell3d.regs.vertex_buffer.count);
const GLint base_vertex = static_cast<GLint>(maxwell3d->regs.vertex_buffer.first);
const GLsizei num_vertices = static_cast<GLsizei>(maxwell3d->regs.vertex_buffer.count);
if (num_instances == 1 && base_instance == 0) {
glDrawArrays(primitive_mode, base_vertex, num_vertices);
} else if (base_instance == 0) {
@@ -274,8 +276,9 @@ void RasterizerOpenGL::DispatchCompute() {
if (!pipeline) {
return;
}
pipeline->SetEngine(kepler_compute, gpu_memory);
pipeline->Configure();
const auto& qmd{kepler_compute.launch_description};
const auto& qmd{kepler_compute->launch_description};
glDispatchCompute(qmd.grid_dim_x, qmd.grid_dim_y, qmd.grid_dim_z);
++num_queued_commands;
has_written_global_memory |= pipeline->WritesGlobalMemory();
@@ -360,7 +363,7 @@ void RasterizerOpenGL::OnCPUWrite(VAddr addr, u64 size) {
}
}
void RasterizerOpenGL::SyncGuestHost() {
void RasterizerOpenGL::InvalidateGPUCache() {
MICROPROFILE_SCOPE(OpenGL_CacheManagement);
shader_cache.SyncGuestHost();
{
@@ -381,40 +384,30 @@ void RasterizerOpenGL::UnmapMemory(VAddr addr, u64 size) {
shader_cache.OnCPUWrite(addr, size);
}
void RasterizerOpenGL::ModifyGPUMemory(GPUVAddr addr, u64 size) {
void RasterizerOpenGL::ModifyGPUMemory(size_t as_id, GPUVAddr addr, u64 size) {
{
std::scoped_lock lock{texture_cache.mutex};
texture_cache.UnmapGPUMemory(addr, size);
texture_cache.UnmapGPUMemory(as_id, addr, size);
}
}
void RasterizerOpenGL::SignalSemaphore(GPUVAddr addr, u32 value) {
if (!gpu.IsAsync()) {
gpu_memory.Write<u32>(addr, value);
return;
}
fence_manager.SignalSemaphore(addr, value);
void RasterizerOpenGL::SignalFence(std::function<void()>&& func) {
fence_manager.SignalFence(std::move(func));
}
void RasterizerOpenGL::SyncOperation(std::function<void()>&& func) {
fence_manager.SyncOperation(std::move(func));
}
void RasterizerOpenGL::SignalSyncPoint(u32 value) {
if (!gpu.IsAsync()) {
gpu.IncrementSyncPoint(value);
return;
}
fence_manager.SignalSyncPoint(value);
}
void RasterizerOpenGL::SignalReference() {
if (!gpu.IsAsync()) {
return;
}
fence_manager.SignalOrdering();
}
void RasterizerOpenGL::ReleaseFences() {
if (!gpu.IsAsync()) {
return;
}
fence_manager.WaitPendingFences();
}
@@ -431,6 +424,7 @@ void RasterizerOpenGL::WaitForIdle() {
}
void RasterizerOpenGL::FragmentBarrier() {
glTextureBarrier();
glMemoryBarrier(GL_FRAMEBUFFER_BARRIER_BIT | GL_TEXTURE_FETCH_BARRIER_BIT);
}
@@ -483,13 +477,13 @@ Tegra::Engines::AccelerateDMAInterface& RasterizerOpenGL::AccessAccelerateDMA()
}
void RasterizerOpenGL::AccelerateInlineToMemory(GPUVAddr address, size_t copy_size,
std::span<u8> memory) {
auto cpu_addr = gpu_memory.GpuToCpuAddress(address);
std::span<const u8> memory) {
auto cpu_addr = gpu_memory->GpuToCpuAddress(address);
if (!cpu_addr) [[unlikely]] {
gpu_memory.WriteBlock(address, memory.data(), copy_size);
gpu_memory->WriteBlock(address, memory.data(), copy_size);
return;
}
gpu_memory.WriteBlockUnsafe(address, memory.data(), copy_size);
gpu_memory->WriteBlockUnsafe(address, memory.data(), copy_size);
{
std::unique_lock<std::mutex> lock{buffer_cache.mutex};
if (!buffer_cache.InlineMemory(*cpu_addr, copy_size, memory)) {
@@ -552,8 +546,8 @@ void RasterizerOpenGL::SyncState() {
}
void RasterizerOpenGL::SyncViewport() {
auto& flags = maxwell3d.dirty.flags;
const auto& regs = maxwell3d.regs;
auto& flags = maxwell3d->dirty.flags;
const auto& regs = maxwell3d->regs;
const bool rescale_viewports = flags[VideoCommon::Dirty::RescaleViewports];
const bool dirty_viewport = flags[Dirty::Viewports] || rescale_viewports;
@@ -658,23 +652,23 @@ void RasterizerOpenGL::SyncViewport() {
}
void RasterizerOpenGL::SyncDepthClamp() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::DepthClampEnabled]) {
return;
}
flags[Dirty::DepthClampEnabled] = false;
oglEnable(GL_DEPTH_CLAMP, maxwell3d.regs.view_volume_clip_control.depth_clamp_disabled == 0);
oglEnable(GL_DEPTH_CLAMP, maxwell3d->regs.view_volume_clip_control.depth_clamp_disabled == 0);
}
void RasterizerOpenGL::SyncClipEnabled(u32 clip_mask) {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::ClipDistances] && !flags[VideoCommon::Dirty::Shaders]) {
return;
}
flags[Dirty::ClipDistances] = false;
clip_mask &= maxwell3d.regs.clip_distance_enabled;
clip_mask &= maxwell3d->regs.clip_distance_enabled;
if (clip_mask == last_clip_distance_mask) {
return;
}
@@ -690,8 +684,8 @@ void RasterizerOpenGL::SyncClipCoef() {
}
void RasterizerOpenGL::SyncCullMode() {
auto& flags = maxwell3d.dirty.flags;
const auto& regs = maxwell3d.regs;
auto& flags = maxwell3d->dirty.flags;
const auto& regs = maxwell3d->regs;
if (flags[Dirty::CullTest]) {
flags[Dirty::CullTest] = false;
@@ -706,23 +700,23 @@ void RasterizerOpenGL::SyncCullMode() {
}
void RasterizerOpenGL::SyncPrimitiveRestart() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::PrimitiveRestart]) {
return;
}
flags[Dirty::PrimitiveRestart] = false;
if (maxwell3d.regs.primitive_restart.enabled) {
if (maxwell3d->regs.primitive_restart.enabled) {
glEnable(GL_PRIMITIVE_RESTART);
glPrimitiveRestartIndex(maxwell3d.regs.primitive_restart.index);
glPrimitiveRestartIndex(maxwell3d->regs.primitive_restart.index);
} else {
glDisable(GL_PRIMITIVE_RESTART);
}
}
void RasterizerOpenGL::SyncDepthTestState() {
auto& flags = maxwell3d.dirty.flags;
const auto& regs = maxwell3d.regs;
auto& flags = maxwell3d->dirty.flags;
const auto& regs = maxwell3d->regs;
if (flags[Dirty::DepthMask]) {
flags[Dirty::DepthMask] = false;
@@ -741,13 +735,13 @@ void RasterizerOpenGL::SyncDepthTestState() {
}
void RasterizerOpenGL::SyncStencilTestState() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::StencilTest]) {
return;
}
flags[Dirty::StencilTest] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
oglEnable(GL_STENCIL_TEST, regs.stencil_enable);
glStencilFuncSeparate(GL_FRONT, MaxwellToGL::ComparisonOp(regs.stencil_front_func_func),
@@ -772,23 +766,23 @@ void RasterizerOpenGL::SyncStencilTestState() {
}
void RasterizerOpenGL::SyncRasterizeEnable() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::RasterizeEnable]) {
return;
}
flags[Dirty::RasterizeEnable] = false;
oglEnable(GL_RASTERIZER_DISCARD, maxwell3d.regs.rasterize_enable == 0);
oglEnable(GL_RASTERIZER_DISCARD, maxwell3d->regs.rasterize_enable == 0);
}
void RasterizerOpenGL::SyncPolygonModes() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::PolygonModes]) {
return;
}
flags[Dirty::PolygonModes] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
if (regs.fill_rectangle) {
if (!GLAD_GL_NV_fill_rectangle) {
LOG_ERROR(Render_OpenGL, "GL_NV_fill_rectangle used and not supported");
@@ -821,7 +815,7 @@ void RasterizerOpenGL::SyncPolygonModes() {
}
void RasterizerOpenGL::SyncColorMask() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::ColorMasks]) {
return;
}
@@ -830,7 +824,7 @@ void RasterizerOpenGL::SyncColorMask() {
const bool force = flags[Dirty::ColorMaskCommon];
flags[Dirty::ColorMaskCommon] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
if (regs.color_mask_common) {
if (!force && !flags[Dirty::ColorMask0]) {
return;
@@ -855,30 +849,30 @@ void RasterizerOpenGL::SyncColorMask() {
}
void RasterizerOpenGL::SyncMultiSampleState() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::MultisampleControl]) {
return;
}
flags[Dirty::MultisampleControl] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
oglEnable(GL_SAMPLE_ALPHA_TO_COVERAGE, regs.multisample_control.alpha_to_coverage);
oglEnable(GL_SAMPLE_ALPHA_TO_ONE, regs.multisample_control.alpha_to_one);
}
void RasterizerOpenGL::SyncFragmentColorClampState() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::FragmentClampColor]) {
return;
}
flags[Dirty::FragmentClampColor] = false;
glClampColor(GL_CLAMP_FRAGMENT_COLOR, maxwell3d.regs.frag_color_clamp ? GL_TRUE : GL_FALSE);
glClampColor(GL_CLAMP_FRAGMENT_COLOR, maxwell3d->regs.frag_color_clamp ? GL_TRUE : GL_FALSE);
}
void RasterizerOpenGL::SyncBlendState() {
auto& flags = maxwell3d.dirty.flags;
const auto& regs = maxwell3d.regs;
auto& flags = maxwell3d->dirty.flags;
const auto& regs = maxwell3d->regs;
if (flags[Dirty::BlendColor]) {
flags[Dirty::BlendColor] = false;
@@ -935,13 +929,13 @@ void RasterizerOpenGL::SyncBlendState() {
}
void RasterizerOpenGL::SyncLogicOpState() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::LogicOp]) {
return;
}
flags[Dirty::LogicOp] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
if (regs.logic_op.enable) {
glEnable(GL_COLOR_LOGIC_OP);
glLogicOp(MaxwellToGL::LogicOp(regs.logic_op.operation));
@@ -951,7 +945,7 @@ void RasterizerOpenGL::SyncLogicOpState() {
}
void RasterizerOpenGL::SyncScissorTest() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::Scissors] && !flags[VideoCommon::Dirty::RescaleScissors]) {
return;
}
@@ -960,7 +954,7 @@ void RasterizerOpenGL::SyncScissorTest() {
const bool force = flags[VideoCommon::Dirty::RescaleScissors];
flags[VideoCommon::Dirty::RescaleScissors] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
const auto& resolution = Settings::values.resolution_info;
const bool is_rescaling{texture_cache.IsRescaling()};
@@ -996,39 +990,39 @@ void RasterizerOpenGL::SyncScissorTest() {
}
void RasterizerOpenGL::SyncPointState() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::PointSize]) {
return;
}
flags[Dirty::PointSize] = false;
oglEnable(GL_POINT_SPRITE, maxwell3d.regs.point_sprite_enable);
oglEnable(GL_PROGRAM_POINT_SIZE, maxwell3d.regs.vp_point_size.enable);
oglEnable(GL_POINT_SPRITE, maxwell3d->regs.point_sprite_enable);
oglEnable(GL_PROGRAM_POINT_SIZE, maxwell3d->regs.vp_point_size.enable);
const bool is_rescaling{texture_cache.IsRescaling()};
const float scale = is_rescaling ? Settings::values.resolution_info.up_factor : 1.0f;
glPointSize(std::max(1.0f, maxwell3d.regs.point_size * scale));
glPointSize(std::max(1.0f, maxwell3d->regs.point_size * scale));
}
void RasterizerOpenGL::SyncLineState() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::LineWidth]) {
return;
}
flags[Dirty::LineWidth] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
oglEnable(GL_LINE_SMOOTH, regs.line_smooth_enable);
glLineWidth(regs.line_smooth_enable ? regs.line_width_smooth : regs.line_width_aliased);
}
void RasterizerOpenGL::SyncPolygonOffset() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::PolygonOffset]) {
return;
}
flags[Dirty::PolygonOffset] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
oglEnable(GL_POLYGON_OFFSET_FILL, regs.polygon_offset_fill_enable);
oglEnable(GL_POLYGON_OFFSET_LINE, regs.polygon_offset_line_enable);
oglEnable(GL_POLYGON_OFFSET_POINT, regs.polygon_offset_point_enable);
@@ -1042,13 +1036,13 @@ void RasterizerOpenGL::SyncPolygonOffset() {
}
void RasterizerOpenGL::SyncAlphaTest() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::AlphaTest]) {
return;
}
flags[Dirty::AlphaTest] = false;
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
if (regs.alpha_test_enabled) {
glEnable(GL_ALPHA_TEST);
glAlphaFunc(MaxwellToGL::ComparisonOp(regs.alpha_test_func), regs.alpha_test_ref);
@@ -1058,17 +1052,17 @@ void RasterizerOpenGL::SyncAlphaTest() {
}
void RasterizerOpenGL::SyncFramebufferSRGB() {
auto& flags = maxwell3d.dirty.flags;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::FramebufferSRGB]) {
return;
}
flags[Dirty::FramebufferSRGB] = false;
oglEnable(GL_FRAMEBUFFER_SRGB, maxwell3d.regs.framebuffer_srgb);
oglEnable(GL_FRAMEBUFFER_SRGB, maxwell3d->regs.framebuffer_srgb);
}
void RasterizerOpenGL::BeginTransformFeedback(GraphicsPipeline* program, GLenum primitive_mode) {
const auto& regs = maxwell3d.regs;
const auto& regs = maxwell3d->regs;
if (regs.tfb_enabled == 0) {
return;
}
@@ -1087,11 +1081,48 @@ void RasterizerOpenGL::BeginTransformFeedback(GraphicsPipeline* program, GLenum
}
void RasterizerOpenGL::EndTransformFeedback() {
if (maxwell3d.regs.tfb_enabled != 0) {
if (maxwell3d->regs.tfb_enabled != 0) {
glEndTransformFeedback();
}
}
void RasterizerOpenGL::InitializeChannel(Tegra::Control::ChannelState& channel) {
CreateChannel(channel);
{
std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
texture_cache.CreateChannel(channel);
buffer_cache.CreateChannel(channel);
}
shader_cache.CreateChannel(channel);
query_cache.CreateChannel(channel);
state_tracker.SetupTables(channel);
}
void RasterizerOpenGL::BindChannel(Tegra::Control::ChannelState& channel) {
const s32 channel_id = channel.bind_id;
BindToChannel(channel_id);
{
std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
texture_cache.BindToChannel(channel_id);
buffer_cache.BindToChannel(channel_id);
}
shader_cache.BindToChannel(channel_id);
query_cache.BindToChannel(channel_id);
state_tracker.ChangeChannel(channel);
state_tracker.InvalidateState();
}
void RasterizerOpenGL::ReleaseChannel(s32 channel_id) {
EraseChannel(channel_id);
{
std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
texture_cache.EraseChannel(channel_id);
buffer_cache.EraseChannel(channel_id);
}
shader_cache.EraseChannel(channel_id);
query_cache.EraseChannel(channel_id);
}
AccelerateDMA::AccelerateDMA(BufferCache& buffer_cache_) : buffer_cache{buffer_cache_} {}
bool AccelerateDMA::BufferCopy(GPUVAddr src_address, GPUVAddr dest_address, u64 amount) {

22
src/video_core/renderer_opengl/gl_rasterizer.h
View File

@@ -13,6 +13,7 @@
#include <glad/glad.h>
#include "common/common_types.h"
#include "video_core/control/channel_state_cache.h"
#include "video_core/engines/maxwell_dma.h"
#include "video_core/rasterizer_accelerated.h"
#include "video_core/rasterizer_interface.h"
@@ -59,7 +60,8 @@ private:
BufferCache& buffer_cache;
};
class RasterizerOpenGL : public VideoCore::RasterizerAccelerated {
class RasterizerOpenGL : public VideoCore::RasterizerAccelerated,
protected VideoCommon::ChannelSetupCaches<VideoCommon::ChannelInfo> {
public:
explicit RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
Core::Memory::Memory& cpu_memory_, const Device& device_,
@@ -79,10 +81,11 @@ public:
bool MustFlushRegion(VAddr addr, u64 size) override;
void InvalidateRegion(VAddr addr, u64 size) override;
void OnCPUWrite(VAddr addr, u64 size) override;
void SyncGuestHost() override;
void InvalidateGPUCache() override;
void UnmapMemory(VAddr addr, u64 size) override;
void ModifyGPUMemory(GPUVAddr addr, u64 size) override;
void SignalSemaphore(GPUVAddr addr, u32 value) override;
void ModifyGPUMemory(size_t as_id, GPUVAddr addr, u64 size) override;
void SignalFence(std::function<void()>&& func) override;
void SyncOperation(std::function<void()>&& func) override;
void SignalSyncPoint(u32 value) override;
void SignalReference() override;
void ReleaseFences() override;
@@ -97,7 +100,7 @@ public:
const Tegra::Engines::Fermi2D::Config& copy_config) override;
Tegra::Engines::AccelerateDMAInterface& AccessAccelerateDMA() override;
void AccelerateInlineToMemory(GPUVAddr address, size_t copy_size,
std::span<u8> memory) override;
std::span<const u8> memory) override;
bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
u32 pixel_stride) override;
void LoadDiskResources(u64 title_id, std::stop_token stop_loading,
@@ -108,6 +111,12 @@ public:
return num_queued_commands > 0;
}
void InitializeChannel(Tegra::Control::ChannelState& channel) override;
void BindChannel(Tegra::Control::ChannelState& channel) override;
void ReleaseChannel(s32 channel_id) override;
private:
static constexpr size_t MAX_TEXTURES = 192;
static constexpr size_t MAX_IMAGES = 48;
@@ -192,9 +201,6 @@ private:
void EndTransformFeedback();
Tegra::GPU& gpu;
Tegra::Engines::Maxwell3D& maxwell3d;
Tegra::Engines::KeplerCompute& kepler_compute;
Tegra::MemoryManager& gpu_memory;
const Device& device;
ScreenInfo& screen_info;

42
src/video_core/renderer_opengl/gl_shader_cache.cpp
View File

@@ -151,16 +151,13 @@ void SetXfbState(VideoCommon::TransformFeedbackState& state, const Maxwell& regs
} // Anonymous namespace
ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindow& emu_window_,
Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::Engines::KeplerCompute& kepler_compute_,
Tegra::MemoryManager& gpu_memory_, const Device& device_,
TextureCache& texture_cache_, BufferCache& buffer_cache_,
ProgramManager& program_manager_, StateTracker& state_tracker_,
VideoCore::ShaderNotify& shader_notify_)
: VideoCommon::ShaderCache{rasterizer_, gpu_memory_, maxwell3d_, kepler_compute_},
emu_window{emu_window_}, device{device_}, texture_cache{texture_cache_},
buffer_cache{buffer_cache_}, program_manager{program_manager_}, state_tracker{state_tracker_},
shader_notify{shader_notify_}, use_asynchronous_shaders{device.UseAsynchronousShaders()},
const Device& device_, TextureCache& texture_cache_,
BufferCache& buffer_cache_, ProgramManager& program_manager_,
StateTracker& state_tracker_, VideoCore::ShaderNotify& shader_notify_)
: VideoCommon::ShaderCache{rasterizer_}, emu_window{emu_window_}, device{device_},
texture_cache{texture_cache_}, buffer_cache{buffer_cache_}, program_manager{program_manager_},
state_tracker{state_tracker_}, shader_notify{shader_notify_},
use_asynchronous_shaders{device.UseAsynchronousShaders()},
profile{
.supported_spirv = 0x00010000,
@@ -310,7 +307,7 @@ GraphicsPipeline* ShaderCache::CurrentGraphicsPipeline() {
current_pipeline = nullptr;
return nullptr;
}
const auto& regs{maxwell3d.regs};
const auto& regs{maxwell3d->regs};
graphics_key.raw = 0;
graphics_key.early_z.Assign(regs.force_early_fragment_tests != 0 ? 1 : 0);
graphics_key.gs_input_topology.Assign(graphics_key.unique_hashes[4] != 0
@@ -351,13 +348,13 @@ GraphicsPipeline* ShaderCache::BuiltPipeline(GraphicsPipeline* pipeline) const n
}
// If something is using depth, we can assume that games are not rendering anything which
// will be used one time.
if (maxwell3d.regs.zeta_enable) {
if (maxwell3d->regs.zeta_enable) {
return nullptr;
}
// If games are using a small index count, we can assume these are full screen quads.
// Usually these shaders are only used once for building textures so we can assume they
// can't be built async
if (maxwell3d.regs.index_array.count <= 6 || maxwell3d.regs.vertex_buffer.count <= 6) {
if (maxwell3d->regs.index_array.count <= 6 || maxwell3d->regs.vertex_buffer.count <= 6) {
return pipeline;
}
return nullptr;
@@ -368,7 +365,7 @@ ComputePipeline* ShaderCache::CurrentComputePipeline() {
if (!shader) {
return nullptr;
}
const auto& qmd{kepler_compute.launch_description};
const auto& qmd{kepler_compute->launch_description};
const ComputePipelineKey key{
.unique_hash = shader->unique_hash,
.shared_memory_size = qmd.shared_alloc,
@@ -480,9 +477,9 @@ std::unique_ptr<GraphicsPipeline> ShaderCache::CreateGraphicsPipeline(
previous_program = &program;
}
auto* const thread_worker{build_in_parallel ? workers.get() : nullptr};
return std::make_unique<GraphicsPipeline>(
device, texture_cache, buffer_cache, gpu_memory, maxwell3d, program_manager, state_tracker,
thread_worker, &shader_notify, sources, sources_spirv, infos, key);
return std::make_unique<GraphicsPipeline>(device, texture_cache, buffer_cache, program_manager,
state_tracker, thread_worker, &shader_notify, sources,
sources_spirv, infos, key);
} catch (Shader::Exception& exception) {
LOG_ERROR(Render_OpenGL, "{}", exception.what());
@@ -491,9 +488,9 @@ std::unique_ptr<GraphicsPipeline> ShaderCache::CreateGraphicsPipeline(
std::unique_ptr<ComputePipeline> ShaderCache::CreateComputePipeline(
const ComputePipelineKey& key, const VideoCommon::ShaderInfo* shader) {
const GPUVAddr program_base{kepler_compute.regs.code_loc.Address()};
const auto& qmd{kepler_compute.launch_description};
ComputeEnvironment env{kepler_compute, gpu_memory, program_base, qmd.program_start};
const GPUVAddr program_base{kepler_compute->regs.code_loc.Address()};
const auto& qmd{kepler_compute->launch_description};
ComputeEnvironment env{*kepler_compute, *gpu_memory, program_base, qmd.program_start};
env.SetCachedSize(shader->size_bytes);
main_pools.ReleaseContents();
@@ -536,9 +533,8 @@ std::unique_ptr<ComputePipeline> ShaderCache::CreateComputePipeline(
break;
}
return std::make_unique<ComputePipeline>(device, texture_cache, buffer_cache, gpu_memory,
kepler_compute, program_manager, program.info, code,
code_spirv);
return std::make_unique<ComputePipeline>(device, texture_cache, buffer_cache, program_manager,
program.info, code, code_spirv);
} catch (Shader::Exception& exception) {
LOG_ERROR(Render_OpenGL, "{}", exception.what());
return nullptr;

9
src/video_core/renderer_opengl/gl_shader_cache.h
View File

@@ -30,12 +30,9 @@ using ShaderWorker = Common::StatefulThreadWorker<ShaderContext::Context>;
class ShaderCache : public VideoCommon::ShaderCache {
public:
explicit ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindow& emu_window_,
Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::Engines::KeplerCompute& kepler_compute_,
Tegra::MemoryManager& gpu_memory_, const Device& device_,
TextureCache& texture_cache_, BufferCache& buffer_cache_,
ProgramManager& program_manager_, StateTracker& state_tracker_,
VideoCore::ShaderNotify& shader_notify_);
const Device& device_, TextureCache& texture_cache_,
BufferCache& buffer_cache_, ProgramManager& program_manager_,
StateTracker& state_tracker_, VideoCore::ShaderNotify& shader_notify_);
~ShaderCache();
void LoadDiskResources(u64 title_id, std::stop_token stop_loading,

17
src/video_core/renderer_opengl/gl_state_tracker.cpp
View File

@@ -7,8 +7,8 @@
#include "common/common_types.h"
#include "core/core.h"
#include "video_core/control/channel_state.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/gpu.h"
#include "video_core/renderer_opengl/gl_state_tracker.h"
#define OFF(field_name) MAXWELL3D_REG_INDEX(field_name)
@@ -202,9 +202,8 @@ void SetupDirtyMisc(Tables& tables) {
} // Anonymous namespace
StateTracker::StateTracker(Tegra::GPU& gpu) : flags{gpu.Maxwell3D().dirty.flags} {
auto& dirty = gpu.Maxwell3D().dirty;
auto& tables = dirty.tables;
void StateTracker::SetupTables(Tegra::Control::ChannelState& channel_state) {
auto& tables{channel_state.maxwell_3d->dirty.tables};
SetupDirtyFlags(tables);
SetupDirtyColorMasks(tables);
SetupDirtyViewports(tables);
@@ -230,4 +229,14 @@ StateTracker::StateTracker(Tegra::GPU& gpu) : flags{gpu.Maxwell3D().dirty.flags}
SetupDirtyMisc(tables);
}
void StateTracker::ChangeChannel(Tegra::Control::ChannelState& channel_state) {
flags = &channel_state.maxwell_3d->dirty.flags;
}
void StateTracker::InvalidateState() {
flags->set();
}
StateTracker::StateTracker() : flags{} {}
} // namespace OpenGL

82
src/video_core/renderer_opengl/gl_state_tracker.h
View File

@@ -12,8 +12,10 @@
#include "video_core/engines/maxwell_3d.h"
namespace Tegra {
class GPU;
namespace Control {
struct ChannelState;
}
} // namespace Tegra
namespace OpenGL {
@@ -83,7 +85,7 @@ static_assert(Last <= std::numeric_limits<u8>::max());
class StateTracker {
public:
explicit StateTracker(Tegra::GPU& gpu);
explicit StateTracker();
void BindIndexBuffer(GLuint new_index_buffer) {
if (index_buffer == new_index_buffer) {
@@ -121,94 +123,106 @@ public:
}
void NotifyScreenDrawVertexArray() {
flags[OpenGL::Dirty::VertexFormats] = true;
flags[OpenGL::Dirty::VertexFormat0 + 0] = true;
flags[OpenGL::Dirty::VertexFormat0 + 1] = true;
(*flags)[OpenGL::Dirty::VertexFormats] = true;
(*flags)[OpenGL::Dirty::VertexFormat0 + 0] = true;
(*flags)[OpenGL::Dirty::VertexFormat0 + 1] = true;
flags[VideoCommon::Dirty::VertexBuffers] = true;
flags[VideoCommon::Dirty::VertexBuffer0] = true;
(*flags)[VideoCommon::Dirty::VertexBuffers] = true;
(*flags)[VideoCommon::Dirty::VertexBuffer0] = true;
flags[OpenGL::Dirty::VertexInstances] = true;
flags[OpenGL::Dirty::VertexInstance0 + 0] = true;
flags[OpenGL::Dirty::VertexInstance0 + 1] = true;
(*flags)[OpenGL::Dirty::VertexInstances] = true;
(*flags)[OpenGL::Dirty::VertexInstance0 + 0] = true;
(*flags)[OpenGL::Dirty::VertexInstance0 + 1] = true;
}
void NotifyPolygonModes() {
flags[OpenGL::Dirty::PolygonModes] = true;
flags[OpenGL::Dirty::PolygonModeFront] = true;
flags[OpenGL::Dirty::PolygonModeBack] = true;
(*flags)[OpenGL::Dirty::PolygonModes] = true;
(*flags)[OpenGL::Dirty::PolygonModeFront] = true;
(*flags)[OpenGL::Dirty::PolygonModeBack] = true;
}
void NotifyViewport0() {
flags[OpenGL::Dirty::Viewports] = true;
flags[OpenGL::Dirty::Viewport0] = true;
(*flags)[OpenGL::Dirty::Viewports] = true;
(*flags)[OpenGL::Dirty::Viewport0] = true;
}
void NotifyScissor0() {
flags[OpenGL::Dirty::Scissors] = true;
flags[OpenGL::Dirty::Scissor0] = true;
(*flags)[OpenGL::Dirty::Scissors] = true;
(*flags)[OpenGL::Dirty::Scissor0] = true;
}
void NotifyColorMask(size_t index) {
flags[OpenGL::Dirty::ColorMasks] = true;
flags[OpenGL::Dirty::ColorMask0 + index] = true;
(*flags)[OpenGL::Dirty::ColorMasks] = true;
(*flags)[OpenGL::Dirty::ColorMask0 + index] = true;
}
void NotifyBlend0() {
flags[OpenGL::Dirty::BlendStates] = true;
flags[OpenGL::Dirty::BlendState0] = true;
(*flags)[OpenGL::Dirty::BlendStates] = true;
(*flags)[OpenGL::Dirty::BlendState0] = true;
}
void NotifyFramebuffer() {
flags[VideoCommon::Dirty::RenderTargets] = true;
(*flags)[VideoCommon::Dirty::RenderTargets] = true;
}
void NotifyFrontFace() {
flags[OpenGL::Dirty::FrontFace] = true;
(*flags)[OpenGL::Dirty::FrontFace] = true;
}
void NotifyCullTest() {
flags[OpenGL::Dirty::CullTest] = true;
(*flags)[OpenGL::Dirty::CullTest] = true;
}
void NotifyDepthMask() {
flags[OpenGL::Dirty::DepthMask] = true;
(*flags)[OpenGL::Dirty::DepthMask] = true;
}
void NotifyDepthTest() {
flags[OpenGL::Dirty::DepthTest] = true;
(*flags)[OpenGL::Dirty::DepthTest] = true;
}
void NotifyStencilTest() {
flags[OpenGL::Dirty::StencilTest] = true;
(*flags)[OpenGL::Dirty::StencilTest] = true;
}
void NotifyPolygonOffset() {
flags[OpenGL::Dirty::PolygonOffset] = true;
(*flags)[OpenGL::Dirty::PolygonOffset] = true;
}
void NotifyRasterizeEnable() {
flags[OpenGL::Dirty::RasterizeEnable] = true;
(*flags)[OpenGL::Dirty::RasterizeEnable] = true;
}
void NotifyFramebufferSRGB() {
flags[OpenGL::Dirty::FramebufferSRGB] = true;
(*flags)[OpenGL::Dirty::FramebufferSRGB] = true;
}
void NotifyLogicOp() {
flags[OpenGL::Dirty::LogicOp] = true;
(*flags)[OpenGL::Dirty::LogicOp] = true;
}
void NotifyClipControl() {
flags[OpenGL::Dirty::ClipControl] = true;
(*flags)[OpenGL::Dirty::ClipControl] = true;
}
void NotifyAlphaTest() {
flags[OpenGL::Dirty::AlphaTest] = true;
(*flags)[OpenGL::Dirty::AlphaTest] = true;
}
void NotifyRange(u8 start, u8 end) {
for (auto flag = start; flag <= end; flag++) {
(*flags)[flag] = true;
}
}
void SetupTables(Tegra::Control::ChannelState& channel_state);
void ChangeChannel(Tegra::Control::ChannelState& channel_state);
void InvalidateState();
private:
Tegra::Engines::Maxwell3D::DirtyState::Flags& flags;
Tegra::Engines::Maxwell3D::DirtyState::Flags* flags;
GLuint framebuffer = 0;
GLuint index_buffer = 0;

2
src/video_core/renderer_opengl/maxwell_to_gl.h
View File

@@ -87,7 +87,7 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> FORMAT_TAB
{GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT}, // BC3_SRGB
{GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM}, // BC7_SRGB
{GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV}, // A4B4G4R4_UNORM
{GL_R8, GL_RED, GL_UNSIGNED_BYTE}, // R4G4_UNORM
{GL_R8, GL_RED, GL_UNSIGNED_BYTE}, // G4R4_UNORM
{GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR}, // ASTC_2D_4X4_SRGB
{GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x8_KHR}, // ASTC_2D_8X8_SRGB
{GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x5_KHR}, // ASTC_2D_8X5_SRGB

2
src/video_core/renderer_opengl/renderer_opengl.cpp
View File

@@ -132,7 +132,7 @@ RendererOpenGL::RendererOpenGL(Core::TelemetrySession& telemetry_session_,
Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_,
std::unique_ptr<Core::Frontend::GraphicsContext> context_)
: RendererBase{emu_window_, std::move(context_)}, telemetry_session{telemetry_session_},
emu_window{emu_window_}, cpu_memory{cpu_memory_}, gpu{gpu_}, state_tracker{gpu},
emu_window{emu_window_}, cpu_memory{cpu_memory_}, gpu{gpu_}, state_tracker{},
program_manager{device},
rasterizer(emu_window, gpu, cpu_memory, device, screen_info, program_manager, state_tracker) {
if (Settings::values.renderer_debug && GLAD_GL_KHR_debug) {