aboutsummaryrefslogtreecommitdiff
path: root/src/video_core
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core')
-rw-r--r--src/video_core/gpu.cpp15
-rw-r--r--src/video_core/gpu.h6
-rw-r--r--src/video_core/gpu_asynch.cpp4
-rw-r--r--src/video_core/gpu_asynch.h1
-rw-r--r--src/video_core/gpu_synch.h1
-rw-r--r--src/video_core/gpu_thread.cpp19
-rw-r--r--src/video_core/gpu_thread.h9
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp1
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h3
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp64
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp5
-rw-r--r--src/video_core/shader/node.h2
-rw-r--r--src/video_core/surface.cpp1
13 files changed, 72 insertions, 59 deletions
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index 76cfe8107..095660115 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -3,6 +3,7 @@
// Refer to the license.txt file included.
#include "common/assert.h"
+#include "common/microprofile.h"
#include "core/core.h"
#include "core/core_timing.h"
#include "core/memory.h"
@@ -17,6 +18,8 @@
namespace Tegra {
+MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
+
GPU::GPU(Core::System& system, VideoCore::RendererBase& renderer, bool is_async)
: system{system}, renderer{renderer}, is_async{is_async} {
auto& rasterizer{renderer.Rasterizer()};
@@ -63,6 +66,16 @@ const DmaPusher& GPU::DmaPusher() const {
return *dma_pusher;
}
+void GPU::WaitFence(u32 syncpoint_id, u32 value) const {
+ // Synced GPU, is always in sync
+ if (!is_async) {
+ return;
+ }
+ MICROPROFILE_SCOPE(GPU_wait);
+ while (syncpoints[syncpoint_id].load(std::memory_order_relaxed) < value) {
+ }
+}
+
void GPU::IncrementSyncPoint(const u32 syncpoint_id) {
syncpoints[syncpoint_id]++;
std::lock_guard lock{sync_mutex};
@@ -326,7 +339,7 @@ void GPU::ProcessSemaphoreTriggerMethod() {
block.sequence = regs.semaphore_sequence;
// TODO(Kmather73): Generate a real GPU timestamp and write it here instead of
// CoreTiming
- block.timestamp = Core::System::GetInstance().CoreTiming().GetTicks();
+ block.timestamp = system.CoreTiming().GetTicks();
memory_manager->WriteBlock(regs.semaphore_address.SemaphoreAddress(), &block,
sizeof(block));
} else {
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index 29fa8e95b..dbca19f35 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -177,6 +177,12 @@ public:
/// Returns a reference to the GPU DMA pusher.
Tegra::DmaPusher& DmaPusher();
+ // Waits for the GPU to finish working
+ virtual void WaitIdle() const = 0;
+
+ /// Allows the CPU/NvFlinger to wait on the GPU before presenting a frame.
+ void WaitFence(u32 syncpoint_id, u32 value) const;
+
void IncrementSyncPoint(u32 syncpoint_id);
u32 GetSyncpointValue(u32 syncpoint_id) const;
diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp
index f2a3a390e..04222d060 100644
--- a/src/video_core/gpu_asynch.cpp
+++ b/src/video_core/gpu_asynch.cpp
@@ -44,4 +44,8 @@ void GPUAsynch::TriggerCpuInterrupt(const u32 syncpoint_id, const u32 value) con
interrupt_manager.GPUInterruptSyncpt(syncpoint_id, value);
}
+void GPUAsynch::WaitIdle() const {
+ gpu_thread.WaitIdle();
+}
+
} // namespace VideoCommon
diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h
index a12f9bac4..1241ade1d 100644
--- a/src/video_core/gpu_asynch.h
+++ b/src/video_core/gpu_asynch.h
@@ -25,6 +25,7 @@ public:
void FlushRegion(CacheAddr addr, u64 size) override;
void InvalidateRegion(CacheAddr addr, u64 size) override;
void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
+ void WaitIdle() const override;
protected:
void TriggerCpuInterrupt(u32 syncpoint_id, u32 value) const override;
diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h
index 5eb1c461c..c71baee89 100644
--- a/src/video_core/gpu_synch.h
+++ b/src/video_core/gpu_synch.h
@@ -24,6 +24,7 @@ public:
void FlushRegion(CacheAddr addr, u64 size) override;
void InvalidateRegion(CacheAddr addr, u64 size) override;
void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
+ void WaitIdle() const override {}
protected:
void TriggerCpuInterrupt([[maybe_unused]] u32 syncpoint_id,
diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp
index 5f039e4fd..758a37f14 100644
--- a/src/video_core/gpu_thread.cpp
+++ b/src/video_core/gpu_thread.cpp
@@ -5,8 +5,6 @@
#include "common/assert.h"
#include "common/microprofile.h"
#include "core/core.h"
-#include "core/core_timing.h"
-#include "core/core_timing_util.h"
#include "core/frontend/scope_acquire_window_context.h"
#include "video_core/dma_pusher.h"
#include "video_core/gpu.h"
@@ -68,14 +66,10 @@ ThreadManager::~ThreadManager() {
void ThreadManager::StartThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher) {
thread = std::thread{RunThread, std::ref(renderer), std::ref(dma_pusher), std::ref(state)};
- synchronization_event = system.CoreTiming().RegisterEvent(
- "GPUThreadSynch", [this](u64 fence, s64) { state.WaitForSynchronization(fence); });
}
void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
- const u64 fence{PushCommand(SubmitListCommand(std::move(entries)))};
- const s64 synchronization_ticks{Core::Timing::usToCycles(std::chrono::microseconds{9000})};
- system.CoreTiming().ScheduleEvent(synchronization_ticks, synchronization_event, fence);
+ PushCommand(SubmitListCommand(std::move(entries)));
}
void ThreadManager::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
@@ -96,16 +90,15 @@ void ThreadManager::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
InvalidateRegion(addr, size);
}
+void ThreadManager::WaitIdle() const {
+ while (state.last_fence > state.signaled_fence.load(std::memory_order_relaxed)) {
+ }
+}
+
u64 ThreadManager::PushCommand(CommandData&& command_data) {
const u64 fence{++state.last_fence};
state.queue.Push(CommandDataContainer(std::move(command_data), fence));
return fence;
}
-MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
-void SynchState::WaitForSynchronization(u64 fence) {
- while (signaled_fence.load() < fence)
- ;
-}
-
} // namespace VideoCommon::GPUThread
diff --git a/src/video_core/gpu_thread.h b/src/video_core/gpu_thread.h
index 3ae0ec9f3..08dc96bb3 100644
--- a/src/video_core/gpu_thread.h
+++ b/src/video_core/gpu_thread.h
@@ -21,9 +21,6 @@ class DmaPusher;
namespace Core {
class System;
-namespace Timing {
-struct EventType;
-} // namespace Timing
} // namespace Core
namespace VideoCommon::GPUThread {
@@ -89,8 +86,6 @@ struct CommandDataContainer {
struct SynchState final {
std::atomic_bool is_running{true};
- void WaitForSynchronization(u64 fence);
-
using CommandQueue = Common::SPSCQueue<CommandDataContainer>;
CommandQueue queue;
u64 last_fence{};
@@ -121,6 +116,9 @@ public:
/// Notify rasterizer that any caches of the specified region should be flushed and invalidated
void FlushAndInvalidateRegion(CacheAddr addr, u64 size);
+ // Wait until the gpu thread is idle.
+ void WaitIdle() const;
+
private:
/// Pushes a command to be executed by the GPU thread
u64 PushCommand(CommandData&& command_data);
@@ -128,7 +126,6 @@ private:
private:
SynchState state;
Core::System& system;
- Core::Timing::EventType* synchronization_event{};
std::thread thread;
std::thread::id thread_id;
};
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index a85f730a8..cbcf81414 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -348,6 +348,7 @@ static constexpr auto RangeFromInterval(Map& map, const Interval& interval) {
}
void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
+ std::lock_guard lock{pages_mutex};
const u64 page_start{addr >> Memory::PAGE_BITS};
const u64 page_end{(addr + size + Memory::PAGE_SIZE - 1) >> Memory::PAGE_BITS};
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index 9c10ebda3..c24a02d71 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -9,6 +9,7 @@
#include <cstddef>
#include <map>
#include <memory>
+#include <mutex>
#include <optional>
#include <tuple>
#include <utility>
@@ -230,6 +231,8 @@ private:
using CachedPageMap = boost::icl::interval_map<u64, int>;
CachedPageMap cached_pages;
+
+ std::mutex pages_mutex;
};
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 6a610a3bc..a3524a6a9 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -1148,7 +1148,7 @@ private:
for (const auto& variant : extras) {
if (const auto argument = std::get_if<TextureArgument>(&variant)) {
expr += GenerateTextureArgument(*argument);
- } else if (std::get_if<TextureAoffi>(&variant)) {
+ } else if (std::holds_alternative<TextureAoffi>(variant)) {
expr += GenerateTextureAoffi(meta->aoffi);
} else {
UNREACHABLE();
@@ -1158,8 +1158,8 @@ private:
return expr + ')';
}
- std::string GenerateTextureArgument(TextureArgument argument) {
- const auto [type, operand] = argument;
+ std::string GenerateTextureArgument(const TextureArgument& argument) {
+ const auto& [type, operand] = argument;
if (operand == nullptr) {
return {};
}
@@ -1235,7 +1235,7 @@ private:
std::string BuildImageValues(Operation operation) {
constexpr std::array constructors{"uint", "uvec2", "uvec3", "uvec4"};
- const auto meta{std::get<MetaImage>(operation.GetMeta())};
+ const auto& meta{std::get<MetaImage>(operation.GetMeta())};
const std::size_t values_count{meta.values.size()};
std::string expr = fmt::format("{}(", constructors.at(values_count - 1));
@@ -1780,14 +1780,14 @@ private:
return {"0", Type::Int};
}
- const auto meta{std::get<MetaImage>(operation.GetMeta())};
+ const auto& meta{std::get<MetaImage>(operation.GetMeta())};
return {fmt::format("imageLoad({}, {}){}", GetImage(meta.image),
BuildIntegerCoordinates(operation), GetSwizzle(meta.element)),
Type::Uint};
}
Expression ImageStore(Operation operation) {
- const auto meta{std::get<MetaImage>(operation.GetMeta())};
+ const auto& meta{std::get<MetaImage>(operation.GetMeta())};
code.AddLine("imageStore({}, {}, {});", GetImage(meta.image),
BuildIntegerCoordinates(operation), BuildImageValues(operation));
return {};
@@ -1795,7 +1795,7 @@ private:
template <const std::string_view& opname>
Expression AtomicImage(Operation operation) {
- const auto meta{std::get<MetaImage>(operation.GetMeta())};
+ const auto& meta{std::get<MetaImage>(operation.GetMeta())};
ASSERT(meta.values.size() == 1);
return {fmt::format("imageAtomic{}({}, {}, {})", opname, GetImage(meta.image),
@@ -2246,7 +2246,7 @@ private:
code.AddLine("#ifdef SAMPLER_{}_IS_BUFFER", sampler.GetIndex());
}
- std::string GetDeclarationWithSuffix(u32 index, const std::string& name) const {
+ std::string GetDeclarationWithSuffix(u32 index, std::string_view name) const {
return fmt::format("{}_{}_{}", name, index, suffix);
}
@@ -2271,17 +2271,15 @@ private:
ShaderWriter code;
};
-static constexpr std::string_view flow_var = "flow_var_";
-
std::string GetFlowVariable(u32 i) {
- return fmt::format("{}{}", flow_var, i);
+ return fmt::format("flow_var_{}", i);
}
class ExprDecompiler {
public:
explicit ExprDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {}
- void operator()(VideoCommon::Shader::ExprAnd& expr) {
+ void operator()(const ExprAnd& expr) {
inner += "( ";
std::visit(*this, *expr.operand1);
inner += " && ";
@@ -2289,7 +2287,7 @@ public:
inner += ')';
}
- void operator()(VideoCommon::Shader::ExprOr& expr) {
+ void operator()(const ExprOr& expr) {
inner += "( ";
std::visit(*this, *expr.operand1);
inner += " || ";
@@ -2297,17 +2295,17 @@ public:
inner += ')';
}
- void operator()(VideoCommon::Shader::ExprNot& expr) {
+ void operator()(const ExprNot& expr) {
inner += '!';
std::visit(*this, *expr.operand1);
}
- void operator()(VideoCommon::Shader::ExprPredicate& expr) {
+ void operator()(const ExprPredicate& expr) {
const auto pred = static_cast<Tegra::Shader::Pred>(expr.predicate);
inner += decomp.GetPredicate(pred);
}
- void operator()(VideoCommon::Shader::ExprCondCode& expr) {
+ void operator()(const ExprCondCode& expr) {
const Node cc = decomp.ir.GetConditionCode(expr.cc);
std::string target;
@@ -2329,15 +2327,15 @@ public:
inner += target;
}
- void operator()(VideoCommon::Shader::ExprVar& expr) {
+ void operator()(const ExprVar& expr) {
inner += GetFlowVariable(expr.var_index);
}
- void operator()(VideoCommon::Shader::ExprBoolean& expr) {
+ void operator()(const ExprBoolean& expr) {
inner += expr.value ? "true" : "false";
}
- std::string& GetResult() {
+ const std::string& GetResult() const {
return inner;
}
@@ -2350,7 +2348,7 @@ class ASTDecompiler {
public:
explicit ASTDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {}
- void operator()(VideoCommon::Shader::ASTProgram& ast) {
+ void operator()(const ASTProgram& ast) {
ASTNode current = ast.nodes.GetFirst();
while (current) {
Visit(current);
@@ -2358,7 +2356,7 @@ public:
}
}
- void operator()(VideoCommon::Shader::ASTIfThen& ast) {
+ void operator()(const ASTIfThen& ast) {
ExprDecompiler expr_parser{decomp};
std::visit(expr_parser, *ast.condition);
decomp.code.AddLine("if ({}) {{", expr_parser.GetResult());
@@ -2372,7 +2370,7 @@ public:
decomp.code.AddLine("}}");
}
- void operator()(VideoCommon::Shader::ASTIfElse& ast) {
+ void operator()(const ASTIfElse& ast) {
decomp.code.AddLine("else {{");
decomp.code.scope++;
ASTNode current = ast.nodes.GetFirst();
@@ -2384,29 +2382,29 @@ public:
decomp.code.AddLine("}}");
}
- void operator()(VideoCommon::Shader::ASTBlockEncoded& ast) {
+ void operator()([[maybe_unused]] const ASTBlockEncoded& ast) {
UNREACHABLE();
}
- void operator()(VideoCommon::Shader::ASTBlockDecoded& ast) {
+ void operator()(const ASTBlockDecoded& ast) {
decomp.VisitBlock(ast.nodes);
}
- void operator()(VideoCommon::Shader::ASTVarSet& ast) {
+ void operator()(const ASTVarSet& ast) {
ExprDecompiler expr_parser{decomp};
std::visit(expr_parser, *ast.condition);
decomp.code.AddLine("{} = {};", GetFlowVariable(ast.index), expr_parser.GetResult());
}
- void operator()(VideoCommon::Shader::ASTLabel& ast) {
+ void operator()(const ASTLabel& ast) {
decomp.code.AddLine("// Label_{}:", ast.index);
}
- void operator()(VideoCommon::Shader::ASTGoto& ast) {
+ void operator()([[maybe_unused]] const ASTGoto& ast) {
UNREACHABLE();
}
- void operator()(VideoCommon::Shader::ASTDoWhile& ast) {
+ void operator()(const ASTDoWhile& ast) {
ExprDecompiler expr_parser{decomp};
std::visit(expr_parser, *ast.condition);
decomp.code.AddLine("do {{");
@@ -2420,7 +2418,7 @@ public:
decomp.code.AddLine("}} while({});", expr_parser.GetResult());
}
- void operator()(VideoCommon::Shader::ASTReturn& ast) {
+ void operator()(const ASTReturn& ast) {
const bool is_true = VideoCommon::Shader::ExprIsTrue(ast.condition);
if (!is_true) {
ExprDecompiler expr_parser{decomp};
@@ -2440,7 +2438,7 @@ public:
}
}
- void operator()(VideoCommon::Shader::ASTBreak& ast) {
+ void operator()(const ASTBreak& ast) {
const bool is_true = VideoCommon::Shader::ExprIsTrue(ast.condition);
if (!is_true) {
ExprDecompiler expr_parser{decomp};
@@ -2455,7 +2453,7 @@ public:
}
}
- void Visit(VideoCommon::Shader::ASTNode& node) {
+ void Visit(const ASTNode& node) {
std::visit(*this, *node->GetInnerData());
}
@@ -2468,9 +2466,9 @@ void GLSLDecompiler::DecompileAST() {
for (u32 i = 0; i < num_flow_variables; i++) {
code.AddLine("bool {} = false;", GetFlowVariable(i));
}
+
ASTDecompiler decompiler{*this};
- VideoCommon::Shader::ASTNode program = ir.GetASTProgram();
- decompiler.Visit(program);
+ decompiler.Visit(ir.GetASTProgram());
}
} // Anonymous namespace
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index 1e6ef66ab..4bbd17b12 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -102,8 +102,6 @@ RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::Syst
RendererOpenGL::~RendererOpenGL() = default;
void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
- system.GetPerfStats().EndSystemFrame();
-
// Maintain the rasterizer's state as a priority
OpenGLState prev_state = OpenGLState::GetCurState();
state.AllDirty();
@@ -135,9 +133,6 @@ void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
render_window.PollEvents();
- system.FrameLimiter().DoFrameLimiting(system.CoreTiming().GetGlobalTimeUs());
- system.GetPerfStats().BeginSystemFrame();
-
// Restore the rasterizer state
prev_state.AllDirty();
prev_state.Apply();
diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h
index 338bab17c..447fb5c1d 100644
--- a/src/video_core/shader/node.h
+++ b/src/video_core/shader/node.h
@@ -410,7 +410,7 @@ public:
explicit OperationNode(OperationCode code) : OperationNode(code, Meta{}) {}
explicit OperationNode(OperationCode code, Meta meta)
- : OperationNode(code, meta, std::vector<Node>{}) {}
+ : OperationNode(code, std::move(meta), std::vector<Node>{}) {}
explicit OperationNode(OperationCode code, std::vector<Node> operands)
: OperationNode(code, Meta{}, std::move(operands)) {}
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
index 250afc6d6..ef6b3592e 100644
--- a/src/video_core/surface.cpp
+++ b/src/video_core/surface.cpp
@@ -252,6 +252,7 @@ PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
default:
break;
}
+ break;
case Tegra::Texture::TextureFormat::R32_G32_B32_A32:
switch (component_type) {
case Tegra::Texture::ComponentType::FLOAT:
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-27 05:30:59 +0000