Compare commits

...

10 Commits

Author SHA1 Message Date
lizzie faa771646c fix 2026-07-03 17:44:22 +00:00
lizzie de59fa1ab4 fix 2026-07-03 17:44:12 +00:00
lizzie b79295c528 windows sucks 2026-07-03 17:44:02 +00:00
lizzie c47bae4c4e license 2026-07-03 17:44:02 +00:00
lizzie ca0eebcfe8 [core] move event creation to attached Core::System, remove unused atomics/prevent false share on Core::Timing
Signed-off-by: lizzie <lizzie@eden-emu.dev>
2026-07-03 17:44:02 +00:00
lizzie 4c65780f11 Revert "[common/dynamic_library] fix AUR build error (#4156)" (#4157)
This reverts commit a9c4c8aefd.

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4157
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
2026-07-02 22:07:37 +02:00
lizzie a9c4c8aefd [common/dynamic_library] fix AUR build error (#4156)
not gonna question why there was an ifdef there

Signed-off-by: lizzie <lizzie@eden-emu.dev>

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4156
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
2026-07-02 19:49:19 +02:00
lizzie a769505a45 [vk, ogl] Remove dedicated precomputed swizzle table; use shorter inline table in block_linear swizzle shaders (#4146)
old table = 64 * 8 * sizeof(u32) = 512 * 4 = 4096 bytes
new table = 8 * sizeof(u32) = 8 * 4 = 32 bytes

the expression
```glsl
    return ((pos & 0x0180) >> 1)
         | ((pos & 0x0040) >> 2)
         | ((pos & 0x0020) << 3)
         | ((pos & 0x0010) << 1)
         | ((pos & 0x000f) << 0);
```
is equivalent for generating the table but idk if we'd want that

Signed-off-by: lizzie <lizzie@eden-emu.dev>

Co-authored-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4146
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
2026-07-02 19:24:07 +02:00
xbzk 54c3d10b86 [vk, renderdoc] (VUID-02997) avoid vk_image_view as VK_NULL_HANDLE when feature nullDescriptor is unavailable (#4056)
This minor change in src\video_core\renderer_vulkan\pipeline_helper.h allows renderdoc capture on mhr sunbreak.
Maybe it sanitizes some crashes on old vulkan GPUs.

Device log (nvidia kepler gpu vulkan 1.1.117):
[  17.605262] Render.Vulkan <Info> video_core\vulkan_common\vulkan_device.cpp:1041:GetSuitability: Device doesn't support feature nullDescriptor

VUID-VkWriteDescriptorSet-descriptorType-02997
For sampled/combined/storage image descriptors, Khronos says imageView must be valid or VK_NULL_HANDLE
(https://docs.vulkan.org/refpages/latest/refpages/source/VkPhysicalDeviceRobustness2FeaturesKHR.html)

but then it says: if nullDescriptor is not enabled, imageView must not be VK_NULL_HANDLE.
(https://docs.vulkan.org/spec/latest/chapters/descriptorsets.html)
nullDescriptor is exactly the feature that allows descriptors to be written with null handles.

The fix relies on conditionally replacing the null sampled view with Eden’s dummy null-image-view object (ImageView constructor checks HasNullDescriptor which checks features.robustness2.nullDescriptor. cheap checks, and only for null handles, so no notable cost).

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4056
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
2026-07-02 03:23:08 +02:00
xbzk bdf60d79a0 [video_core] gate Mesa EXIT terminator detection behind non-proprietary driver (#4151)
FIXES A REGRESSION FROM 4012.

TryFindSize's Mesa fallback matched the unconditional @PT EXIT (0xE30000000007000F) that nv50_ir emits at program end.
That exact word is also emitted by NVN mid-program for unconditional early-outs, so on retail titles (e.g. Super Mario 3D World) the scan truncated at the first early EXIT and dropped the rest of the shader, making textures vanish.

Mesa's terminal EXIT and NVN's mid-program EXIT are bit-identical, so no mask can separate them. Gate the heuristic on !is_proprietary_driver:

NVN binaries (bindless cbuf slot 2) keep self-branch-only sizing, while nouveau/Mesa homebrew (single terminal EXIT, no self-branch trailer) still uses the EXIT terminator. nv50_ir emits exactly one EXIT (early returns lower to BRA-to-exit, discard to DISCARD), so a single terminal match is correct for that path.

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4151
Reviewed-by: Lizzie <lizzie@eden-emu.dev>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
2026-07-01 13:06:00 +02:00
29 changed files with 239 additions and 396 deletions
+5 -3
View File
@@ -22,9 +22,11 @@ using namespace std::literals;
constexpr auto INCREMENT_TIME{5ms};
DeviceSession::DeviceSession(Core::System& system_)
: system{system_}, thread_event{Core::Timing::CreateEvent(
"AudioOutSampleTick",
[this](s64 time, std::chrono::nanoseconds) { return ThreadFunc(); })} {}
: system{system_}
, thread_event{system_.CreateTimingEvent("AudioOutSampleTick", [this](s64 time, std::chrono::nanoseconds) {
return ThreadFunc();
})}
{}
DeviceSession::~DeviceSession() {
Finalize();
+4
View File
@@ -969,4 +969,8 @@ void System::ApplySettings() {
}
}
std::shared_ptr<Core::Timing::EventType> System::CreateTimingEvent(std::string name, Core::Timing::TimedCallback&& callback) {
return std::make_shared<Core::Timing::EventType>(std::move(callback), std::move(name));
}
} // namespace Core
+4
View File
@@ -19,6 +19,7 @@
#include "core/file_sys/vfs/vfs_types.h"
#include "core/hle/service/os/event.h"
#include "core/hle/service/kernel_helpers.h"
#include "core/core_timing.h"
namespace Core::Frontend {
class EmuWindow;
@@ -438,6 +439,9 @@ public:
/// Applies any changes to settings to this core instance.
void ApplySettings();
std::shared_ptr<Core::Timing::EventType> CreateTimingEvent(std::string name, Core::Timing::TimedCallback&& callback);
private:
struct Impl;
std::unique_ptr<Impl> impl;
};
+14 -45
View File
@@ -23,10 +23,6 @@ namespace Core::Timing {
constexpr s64 MAX_SLICE_LENGTH = 10000;
std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback) {
return std::make_shared<EventType>(std::move(callback), std::move(name));
}
struct CoreTiming::Event {
s64 time;
u64 fifo_order;
@@ -36,11 +32,10 @@ struct CoreTiming::Event {
// Sort by time, unless the times are the same, in which case sort by
// the order added to the queue
friend bool operator>(const Event& left, const Event& right) {
friend bool operator>(const Event& left, const Event& right) noexcept {
return std::tie(left.time, left.fifo_order) > std::tie(right.time, right.fifo_order);
}
friend bool operator<(const Event& left, const Event& right) {
friend bool operator<(const Event& left, const Event& right) noexcept {
return std::tie(left.time, left.fifo_order) < std::tie(right.time, right.fifo_order);
}
};
@@ -60,8 +55,6 @@ void CoreTiming::Initialize(std::function<void()>&& on_thread_init_) {
Common::SetCurrentThreadName("HostTiming");
Common::SetCurrentThreadPriority(Common::ThreadPriority::High);
on_thread_init();
has_started = true;
// base frequency in MHz: 1ns (10^-9) = 1GHz (10^9)
while (!stop_token.stop_requested()) {
while (!paused && !stop_token.stop_requested()) {
@@ -77,8 +70,8 @@ void CoreTiming::Initialize(std::function<void()>&& on_thread_init_) {
// continue.
wait_set = true;
event.Wait();
wait_set = false;
}
wait_set = false;
}
paused_set = true;
pause_event.Wait();
@@ -144,39 +137,28 @@ void CoreTiming::ScheduleEvent(std::chrono::nanoseconds ns_into_future,
event.Set();
}
void CoreTiming::ScheduleLoopingEvent(std::chrono::nanoseconds start_time,
std::chrono::nanoseconds resched_time,
const std::shared_ptr<EventType>& event_type,
bool absolute_time) {
void CoreTiming::ScheduleLoopingEvent(std::chrono::nanoseconds start_time, std::chrono::nanoseconds resched_time, const std::shared_ptr<EventType>& event_type, bool absolute_time) {
{
std::scoped_lock scope{basic_lock};
const auto next_time{absolute_time ? start_time : GetGlobalTimeNs() + start_time};
auto h{event_queue.emplace(
Event{next_time.count(), event_fifo_id++, event_type, resched_time.count()})};
auto h = event_queue.emplace(Event{next_time.count(), event_fifo_id++, event_type, resched_time.count()});
(*h).handle = h;
}
event.Set();
}
void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type,
UnscheduleEventType type) {
void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, UnscheduleEventType type) {
{
std::scoped_lock lk{basic_lock};
std::vector<heap_t::handle_type> to_remove;
for (auto itr = event_queue.begin(); itr != event_queue.end(); itr++) {
const Event& e = *itr;
for (auto it = event_queue.begin(); it != event_queue.end(); it++) {
auto const& e = *it;
if (e.type.lock().get() == event_type.get()) {
to_remove.push_back(itr->handle);
to_remove.push_back(it->handle);
}
}
for (auto& h : to_remove) {
for (auto& h : to_remove)
event_queue.erase(h);
}
event_type->sequence_number++;
}
@@ -187,16 +169,15 @@ void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type,
}
static u64 GetNextTickCount(u64 next_ticks) {
if (Settings::values.use_custom_cpu_ticks.GetValue()) {
if (Settings::values.use_custom_cpu_ticks.GetValue())
return Settings::values.cpu_ticks.GetValue();
}
return next_ticks;
}
void CoreTiming::AddTicks(u64 ticks_to_add) {
const u64 ticks = GetNextTickCount(ticks_to_add);
cpu_ticks += ticks;
downcount -= static_cast<s64>(ticks);
downcount -= s64(ticks);
}
void CoreTiming::Idle() {
@@ -270,19 +251,14 @@ std::optional<s64> CoreTiming::Advance() {
next_time = pause_end_time + next_schedule_time;
}
event_queue.update(evt.handle, Event{next_time, event_fifo_id++, evt.type,
next_schedule_time, evt.handle});
event_queue.update(evt.handle, Event{next_time, event_fifo_id++, evt.type, next_schedule_time, evt.handle});
}
}
global_timer = GetGlobalTimeNs().count();
}
if (!event_queue.empty()) {
return event_queue.top().time;
} else {
return std::nullopt;
}
return event_queue.empty() ? std::optional<s64>{} : event_queue.top().time;
}
void CoreTiming::Reset() {
@@ -293,7 +269,6 @@ void CoreTiming::Reset() {
timer_thread.request_stop();
timer_thread.join();
}
has_started = false;
}
/// @brief Returns current time in nanoseconds.
@@ -310,10 +285,4 @@ std::chrono::microseconds CoreTiming::GetGlobalTimeUs() const noexcept {
: std::chrono::microseconds{Common::WallClock::CPUTickToUS(cpu_ticks)};
}
#ifdef _WIN32
void CoreTiming::SetTimerResolutionNs(std::chrono::nanoseconds ns) {
timer_resolution_ns = ns.count();
}
#endif
} // namespace Core::Timing
+11 -33
View File
@@ -29,8 +29,7 @@ using TimedCallback = std::function<std::optional<std::chrono::nanoseconds>(
/// Contains the characteristics of a particular event.
struct EventType {
explicit EventType(TimedCallback&& callback_, std::string&& name_)
: callback{std::move(callback_)}, name{std::move(name_)}, sequence_number{0} {}
explicit EventType(TimedCallback&& callback_, std::string&& name_) : callback{std::move(callback_)}, name{std::move(name_)}, sequence_number{0} {}
/// The event's callback function.
TimedCallback callback;
@@ -90,11 +89,6 @@ public:
/// Checks if core timing is running.
bool IsRunning() const;
/// Checks if the timer thread has started.
bool HasStarted() const {
return has_started;
}
/// Checks if there are any pending time events.
bool HasPendingEvents() const;
@@ -134,45 +128,29 @@ public:
/// Checks for events manually and returns time in nanoseconds for next event, threadsafe.
std::optional<s64> Advance();
#ifdef _WIN32
void SetTimerResolutionNs(std::chrono::nanoseconds ns);
#endif
struct Event;
void Reset();
using heap_t = boost::heap::fibonacci_heap<CoreTiming::Event, boost::heap::compare<std::greater<>>>;
Common::Event event{};
Common::Event pause_event{};
alignas(64) mutable std::mutex basic_lock;
alignas(64) std::mutex advance_lock;
alignas(64) std::atomic<bool> paused{};
alignas(64) std::atomic<bool> paused_set{};
alignas(64) std::atomic<bool> wait_set{};
std::function<void()> on_thread_init{};
heap_t event_queue;
std::jthread timer_thread;
s64 global_timer = 0;
#ifdef _WIN32
s64 timer_resolution_ns;
#endif
u64 event_fifo_id = 0;
s64 pause_end_time{};
/// Cycle timing
u64 cpu_ticks{};
s64 downcount{};
Common::Event event{};
Common::Event pause_event{};
std::function<void()> on_thread_init{};
std::jthread timer_thread;
mutable std::mutex basic_lock;
std::mutex advance_lock;
std::atomic<bool> paused{};
std::atomic<bool> paused_set{};
std::atomic<bool> wait_set{};
std::atomic<bool> has_started{};
bool is_multicore{};
};
/// Creates a core timing event with the given name and callback.
///
/// @param name The name of the core timing event to create.
/// @param callback The callback to execute for the event.
///
/// @returns An EventType instance representing the created event.
///
std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback);
} // namespace Core::Timing
+5 -6
View File
@@ -1,4 +1,4 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
@@ -13,11 +13,10 @@ namespace Kernel {
void KHardwareTimer::Initialize() {
// Create the timing callback to register with CoreTiming.
m_event_type = Core::Timing::CreateEvent("KHardwareTimer::Callback",
[this](s64, std::chrono::nanoseconds) {
this->DoTask();
return std::nullopt;
});
m_event_type = m_kernel.System().CreateTimingEvent("KHardwareTimer::Callback", [this](s64, std::chrono::nanoseconds) {
this->DoTask();
return std::nullopt;
});
}
void KHardwareTimer::Finalize() {
+1 -1
View File
@@ -255,7 +255,7 @@ struct KernelCore::Impl {
}
void InitializePreemption(KernelCore& kernel) {
preemption_event = Core::Timing::CreateEvent("PreemptionCallback", [this, &kernel](s64 time, std::chrono::nanoseconds) -> std::optional<std::chrono::nanoseconds> {
preemption_event = system.CreateTimingEvent("PreemptionCallback", [this, &kernel](s64 time, std::chrono::nanoseconds) -> std::optional<std::chrono::nanoseconds> {
{
KScopedSchedulerLock lock(kernel);
global_scheduler_context->PreemptThreads(kernel);
@@ -25,16 +25,11 @@ AlarmWorker::~AlarmWorker() {
void AlarmWorker::Initialize(std::shared_ptr<Service::PSC::Time::ServiceManager> time_m) {
m_time_m = std::move(time_m);
m_timer_event = m_ctx.CreateEvent("Glue:AlarmWorker:TimerEvent");
m_timer_timing_event = Core::Timing::CreateEvent(
"Glue:AlarmWorker::AlarmTimer",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_timer_event->Signal(m_system.Kernel());
return std::nullopt;
});
m_timer_timing_event = m_system.CreateTimingEvent("Glue:AlarmWorker::AlarmTimer", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_timer_event->Signal(m_system.Kernel());
return std::nullopt;
});
AttachToClosestAlarmEvent();
}
+10 -19
View File
@@ -22,28 +22,19 @@ namespace Service::Glue::Time {
TimeWorker::TimeWorker(Core::System& system, StandardSteadyClockResource& steady_clock_resource,
FileTimestampWorker& file_timestamp_worker)
: m_system{system}, m_ctx{m_system, "Glue:TimeWorker"}, m_event{m_ctx.CreateEvent(
"Glue:TimeWorker:Event")},
: m_system{system}, m_ctx{m_system, "Glue:TimeWorker"}, m_event{m_ctx.CreateEvent("Glue:TimeWorker:Event")},
m_steady_clock_resource{steady_clock_resource},
m_file_timestamp_worker{file_timestamp_worker}, m_timer_steady_clock{m_ctx.CreateEvent(
"Glue:TimeWorker:SteadyClockTimerEvent")},
m_file_timestamp_worker{file_timestamp_worker}, m_timer_steady_clock{m_ctx.CreateEvent("Glue:TimeWorker:SteadyClockTimerEvent")},
m_timer_file_system{m_ctx.CreateEvent("Glue:TimeWorker:FileTimeTimerEvent")},
m_alarm_worker{m_system, m_steady_clock_resource}, m_pm_state_change_handler{m_alarm_worker} {
m_timer_steady_clock_timing_event = Core::Timing::CreateEvent(
"Time::SteadyClockEvent",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_timer_steady_clock->Signal(m_system.Kernel());
return std::nullopt;
});
m_timer_file_system_timing_event = Core::Timing::CreateEvent(
"Time::SteadyClockEvent",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_timer_file_system->Signal(m_system.Kernel());
return std::nullopt;
});
m_timer_steady_clock_timing_event = m_system.CreateTimingEvent("Time::SteadyClockEvent", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_timer_steady_clock->Signal(m_system.Kernel());
return std::nullopt;
});
m_timer_file_system_timing_event = m_system.CreateTimingEvent("Time::SteadyClockEvent", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_timer_file_system->Signal(m_system.Kernel());
return std::nullopt;
});
}
TimeWorker::~TimeWorker() {
+6 -11
View File
@@ -51,17 +51,12 @@ Hidbus::Hidbus(Core::System& system_)
RegisterHandlers(functions);
// Register update callbacks
hidbus_update_event = Core::Timing::CreateEvent(
"Hidbus::UpdateCallback",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
const auto guard = LockService();
UpdateHidbus(ns_late);
return std::nullopt;
});
system_.CoreTiming().ScheduleLoopingEvent(hidbus_update_ns, hidbus_update_ns,
hidbus_update_event);
hidbus_update_event = system_.CreateTimingEvent("Hidbus::UpdateCallback", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
const auto guard = LockService();
UpdateHidbus(ns_late);
return std::nullopt;
});
system_.CoreTiming().ScheduleLoopingEvent(hidbus_update_ns, hidbus_update_ns, hidbus_update_event);
}
Hidbus::~Hidbus() {
+8 -16
View File
@@ -23,25 +23,17 @@ Conductor::Conductor(Core::System& system, Container& container, DisplayList& di
});
if (system.IsMulticore()) {
m_event = Core::Timing::CreateEvent(
"ScreenComposition",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_signal.Set();
return std::chrono::nanoseconds(this->GetNextTicks());
});
m_event = system.CreateTimingEvent("ScreenComposition", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
m_signal.Set();
return std::chrono::nanoseconds(this->GetNextTicks());
});
system.CoreTiming().ScheduleLoopingEvent(FrameNs, FrameNs, m_event);
m_thread = std::jthread([this](std::stop_token token) { this->VsyncThread(token); });
} else {
m_event = Core::Timing::CreateEvent(
"ScreenComposition",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
this->ProcessVsync();
return std::chrono::nanoseconds(this->GetNextTicks());
});
m_event = system.CreateTimingEvent("ScreenComposition", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
this->ProcessVsync();
return std::chrono::nanoseconds(this->GetNextTicks());
});
system.CoreTiming().ScheduleLoopingEvent(FrameNs, FrameNs, m_event);
}
}
+4 -6
View File
@@ -231,12 +231,10 @@ CheatEngine::~CheatEngine() {
}
void CheatEngine::Initialize() {
event = Core::Timing::CreateEvent(
"CheatEngine::FrameCallback::" + Common::HexToString(metadata.main_nso_build_id),
[this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
FrameCallback(ns_late);
return std::nullopt;
});
event = system.CreateTimingEvent("CheatEngine::FrameCallback::" + Common::HexToString(metadata.main_nso_build_id), [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
FrameCallback(ns_late);
return std::nullopt;
});
core_timing.ScheduleLoopingEvent(CHEAT_ENGINE_NS, CHEAT_ENGINE_NS, event);
metadata.process_id = system.ApplicationProcess()->GetProcessId();
+5 -7
View File
@@ -52,14 +52,12 @@ void MemoryWriteWidth(Core::Memory::Memory& memory, u32 width, VAddr addr, u64 v
} // Anonymous namespace
Freezer::Freezer(Core::Timing::CoreTiming& core_timing_, Core::Memory::Memory& memory_)
Freezer::Freezer(Core::System& system_, Core::Timing::CoreTiming& core_timing_, Core::Memory::Memory& memory_)
: core_timing{core_timing_}, memory{memory_} {
event = Core::Timing::CreateEvent("MemoryFreezer::FrameCallback",
[this](s64 time, std::chrono::nanoseconds ns_late)
-> std::optional<std::chrono::nanoseconds> {
FrameCallback(ns_late);
return std::nullopt;
});
event = system_.CreateTimingEvent("MemoryFreezer::FrameCallback", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
FrameCallback(ns_late);
return std::nullopt;
});
core_timing.ScheduleEvent(memory_freezer_ns, event);
}
+10 -5
View File
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -11,14 +14,16 @@
#include <vector>
#include "common/common_types.h"
namespace Core::Timing {
namespace Core {
class System;
namespace Timing {
class CoreTiming;
struct EventType;
} // namespace Core::Timing
namespace Core::Memory {
namespace Memory {
class Memory;
}
} //namespace Core::Memory
} //namespace Core
namespace Tools {
@@ -38,7 +43,7 @@ public:
u64 value;
};
explicit Freezer(Core::Timing::CoreTiming& core_timing_, Core::Memory::Memory& memory_);
explicit Freezer(Core::System& system_, Core::Timing::CoreTiming& core_timing_, Core::Memory::Memory& memory_);
~Freezer();
// Enables or disables the entire memory freezer.
+20 -34
View File
@@ -56,33 +56,22 @@ ResourceManager::ResourceManager(Core::System& system_,
applet_resource = std::make_shared<AppletResource>(system);
// Register update callbacks
npad_update_event = Core::Timing::CreateEvent("HID::UpdatePadCallback",
[this](s64 time, std::chrono::nanoseconds ns_late)
-> std::optional<std::chrono::nanoseconds> {
UpdateNpad(ns_late);
return std::nullopt;
});
default_update_event = Core::Timing::CreateEvent(
"HID::UpdateDefaultCallback",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
UpdateControllers(ns_late);
return std::nullopt;
});
mouse_keyboard_update_event = Core::Timing::CreateEvent(
"HID::UpdateMouseKeyboardCallback",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
UpdateMouseKeyboard(ns_late);
return std::nullopt;
});
motion_update_event = Core::Timing::CreateEvent(
"HID::UpdateMotionCallback",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
UpdateMotion(ns_late);
return std::nullopt;
});
npad_update_event = system.CreateTimingEvent("HID::UpdatePadCallback", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
UpdateNpad(ns_late);
return std::nullopt;
});
default_update_event = system.CreateTimingEvent("HID::UpdateDefaultCallback", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
UpdateControllers(ns_late);
return std::nullopt;
});
mouse_keyboard_update_event = system.CreateTimingEvent("HID::UpdateMouseKeyboardCallback", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
UpdateMouseKeyboard(ns_late);
return std::nullopt;
});
motion_update_event = system.CreateTimingEvent("HID::UpdateMotionCallback", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
UpdateMotion(ns_late);
return std::nullopt;
});
}
ResourceManager::~ResourceManager() {
@@ -267,13 +256,10 @@ void ResourceManager::InitializeTouchScreenSampler() {
touch_screen = std::make_shared<TouchScreen>(touch_resource);
gesture = std::make_shared<Gesture>(touch_resource);
touch_update_event = Core::Timing::CreateEvent(
"HID::TouchUpdateCallback",
[this](s64 time,
std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
touch_resource->OnTouchUpdate(time);
return std::nullopt;
});
touch_update_event = system.CreateTimingEvent("HID::TouchUpdateCallback", [this](s64 time, std::chrono::nanoseconds ns_late) -> std::optional<std::chrono::nanoseconds> {
touch_resource->OnTouchUpdate(time);
return std::nullopt;
});
touch_resource->SetTouchDriver(touch_driver);
touch_resource->SetAppletResource(applet_resource, &shared_mutex);
+1 -6
View File
@@ -266,12 +266,7 @@ void Init(QWidget* root) {
Common::GetMemInfo().TotalPhysicalMemory / f64{1_GiB});
LOG_INFO(Frontend, "Host Swap: {:.2f} GiB", Common::GetMemInfo().TotalSwapMemory / f64{1_GiB});
#ifdef _WIN32
LOG_INFO(Frontend, "Host Timer Resolution: {:.4f} ms",
std::chrono::duration_cast<std::chrono::duration<f64, std::milli>>(
Common::Windows::SetCurrentTimerResolutionToMaximum())
.count());
QtCommon::system->CoreTiming().SetTimerResolutionNs(
Common::Windows::GetCurrentTimerResolution());
LOG_INFO(Frontend, "Host Timer Resolution: {:.4f} ms", std::chrono::duration_cast<std::chrono::duration<f64, std::milli>>(Common::Windows::SetCurrentTimerResolutionToMaximum()).count());
#endif
// Remove cached contents generated during the previous session
+15 -10
View File
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: 2016 Dolphin Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -53,14 +56,15 @@ u64 TestTimerSpeed(Core::Timing::CoreTiming& core_timing) {
} // Anonymous namespace
TEST_CASE("CoreTiming[BasicOrder]", "[core]") {
Core::System system{};
ScopeInit guard;
auto& core_timing = guard.core_timing;
std::vector<std::shared_ptr<Core::Timing::EventType>> events{
Core::Timing::CreateEvent("callbackA", HostCallbackTemplate<0>),
Core::Timing::CreateEvent("callbackB", HostCallbackTemplate<1>),
Core::Timing::CreateEvent("callbackC", HostCallbackTemplate<2>),
Core::Timing::CreateEvent("callbackD", HostCallbackTemplate<3>),
Core::Timing::CreateEvent("callbackE", HostCallbackTemplate<4>),
system.CreateTimingEvent("callbackA", HostCallbackTemplate<0>),
system.CreateTimingEvent("callbackB", HostCallbackTemplate<1>),
system.CreateTimingEvent("callbackC", HostCallbackTemplate<2>),
system.CreateTimingEvent("callbackD", HostCallbackTemplate<3>),
system.CreateTimingEvent("callbackE", HostCallbackTemplate<4>),
};
expected_callback = 0;
@@ -93,14 +97,15 @@ TEST_CASE("CoreTiming[BasicOrder]", "[core]") {
}
TEST_CASE("CoreTiming[BasicOrderNoPausing]", "[core]") {
Core::System system{};
ScopeInit guard;
auto& core_timing = guard.core_timing;
std::vector<std::shared_ptr<Core::Timing::EventType>> events{
Core::Timing::CreateEvent("callbackA", HostCallbackTemplate<0>),
Core::Timing::CreateEvent("callbackB", HostCallbackTemplate<1>),
Core::Timing::CreateEvent("callbackC", HostCallbackTemplate<2>),
Core::Timing::CreateEvent("callbackD", HostCallbackTemplate<3>),
Core::Timing::CreateEvent("callbackE", HostCallbackTemplate<4>),
system.CreateTimingEvent("callbackA", HostCallbackTemplate<0>),
system.CreateTimingEvent("callbackB", HostCallbackTemplate<1>),
system.CreateTimingEvent("callbackC", HostCallbackTemplate<2>),
system.CreateTimingEvent("callbackD", HostCallbackTemplate<3>),
system.CreateTimingEvent("callbackE", HostCallbackTemplate<4>),
};
core_timing.SyncPause(true);
@@ -11,9 +11,8 @@
#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
#define END_PUSH_CONSTANTS };
#define UNIFORM(n)
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_IMAGE 2
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 1
#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
@@ -26,8 +25,7 @@
#define BEGIN_PUSH_CONSTANTS
#define END_PUSH_CONSTANTS
#define UNIFORM(n) layout (location = n) uniform
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 0
#endif
@@ -43,10 +41,6 @@ UNIFORM(6) uint block_height;
UNIFORM(7) uint block_height_mask;
END_PUSH_CONSTANTS
layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
uint swizzle_table[];
};
#if HAS_EXTENDED_TYPES
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
@@ -71,9 +65,19 @@ const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHI
const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
uint SwizzleTable(uint pos) {
const uint t[8] = uint[](
0x12100200, 0x13110301, 0x16140604, 0x17150705,
0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
);
const uint i = pos >> 4;
const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
return (h << 4) | (pos & 0xf);
}
uint SwizzleOffset(uvec2 pos) {
pos = pos & SWIZZLE_MASK;
return swizzle_table[pos.y * 64 + pos.x];
return SwizzleTable(pos.y * 64 + pos.x);
}
uvec4 ReadTexel(uint offset) {
@@ -11,9 +11,8 @@
#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
#define END_PUSH_CONSTANTS };
#define UNIFORM(n)
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_IMAGE 2
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 1
#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
@@ -26,8 +25,7 @@
#define BEGIN_PUSH_CONSTANTS
#define END_PUSH_CONSTANTS
#define UNIFORM(n) layout (location = n) uniform
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 0
#endif
@@ -45,10 +43,6 @@ UNIFORM(8) uint block_depth;
UNIFORM(9) uint block_depth_mask;
END_PUSH_CONSTANTS
layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
uint swizzle_table[];
};
#if HAS_EXTENDED_TYPES
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
@@ -73,9 +67,19 @@ const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHI
const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
uint SwizzleTable(uint pos) {
const uint t[8] = uint[](
0x12100200, 0x13110301, 0x16140604, 0x17150705,
0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
);
const uint i = pos >> 4;
const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
return (h << 4) | (pos & 0xf);
}
uint SwizzleOffset(uvec2 pos) {
pos = pos & SWIZZLE_MASK;
return swizzle_table[pos.y * 64 + pos.x];
return SwizzleTable(pos.y * 64 + pos.x);
}
uvec4 ReadTexel(uint offset) {
@@ -10,9 +10,8 @@
#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
#define END_PUSH_CONSTANTS };
#define UNIFORM(n)
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_BUFFER 2
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_BUFFER 1
#else
#extension GL_NV_gpu_shader5 : enable
#ifdef GL_NV_gpu_shader5
@@ -23,7 +22,6 @@
#define BEGIN_PUSH_CONSTANTS
#define END_PUSH_CONSTANTS
#define UNIFORM(n) layout(location = n) uniform
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_BUFFER 0
#endif
@@ -66,13 +64,9 @@ END_PUSH_CONSTANTS
#endif
// --- Buffers ---
layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
uint swizzle_table[];
};
#if HAS_EXTENDED_TYPES
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
#endif
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU32 { uint u32data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU64 { uvec2 u64data[]; };
@@ -96,10 +90,20 @@ const uint GOB_SIZE_Z_SHIFT = 0;
const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1u, GOB_SIZE_Y - 1u);
uint SwizzleTable(uint pos) {
const uint t[8] = uint[](
0x12100200, 0x13110301, 0x16140604, 0x17150705,
0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
);
const uint i = pos >> 4;
const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
return (h << 4) | (pos & 0xf);
}
// --- Helpers ---
uint SwizzleOffset(uvec2 pos) {
pos &= SWIZZLE_MASK;
return swizzle_table[pos.y * 64u + pos.x];
return SwizzleTable(pos.y * 64u + pos.x);
}
uvec4 ReadTexel(uint offset) {
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -56,10 +59,8 @@ UtilShaders::UtilShaders(ProgramManager& program_manager_)
copy_bc4_program(MakeProgram(OPENGL_COPY_BC4_COMP)),
convert_s8d24_program(MakeProgram(OPENGL_CONVERT_S8D24_COMP)),
convert_ms_to_nonms_program(MakeProgram(CONVERT_MSAA_TO_NON_MSAA_COMP)),
convert_nonms_to_ms_program(MakeProgram(CONVERT_NON_MSAA_TO_MSAA_COMP)) {
const auto swizzle_table = Tegra::Texture::MakeSwizzleTable();
swizzle_table_buffer.Create();
glNamedBufferStorage(swizzle_table_buffer.handle, sizeof(swizzle_table), &swizzle_table, 0);
convert_nonms_to_ms_program(MakeProgram(CONVERT_NON_MSAA_TO_MSAA_COMP))
{
}
UtilShaders::~UtilShaders() = default;
@@ -116,13 +117,11 @@ void UtilShaders::ASTCDecode(Image& image, const StagingBufferMap& map,
void UtilShaders::BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
std::span<const SwizzleParameters> swizzles) {
static constexpr Extent3D WORKGROUP_SIZE{32, 32, 1};
static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
static constexpr GLuint BINDING_INPUT_BUFFER = 1;
static constexpr GLuint BINDING_INPUT_BUFFER = 0;
static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
program_manager.BindComputeProgram(block_linear_unswizzle_2d_program.handle);
glFlushMappedNamedBufferRange(map.buffer, map.offset, image.guest_size_bytes);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
for (const SwizzleParameters& swizzle : swizzles) {
@@ -153,14 +152,11 @@ void UtilShaders::BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
void UtilShaders::BlockLinearUpload3D(Image& image, const StagingBufferMap& map,
std::span<const SwizzleParameters> swizzles) {
static constexpr Extent3D WORKGROUP_SIZE{16, 8, 8};
static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
static constexpr GLuint BINDING_INPUT_BUFFER = 1;
static constexpr GLuint BINDING_INPUT_BUFFER = 0;
static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
glFlushMappedNamedBufferRange(map.buffer, map.offset, image.guest_size_bytes);
program_manager.BindComputeProgram(block_linear_unswizzle_3d_program.handle);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
for (const SwizzleParameters& swizzle : swizzles) {
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -45,9 +48,6 @@ public:
private:
ProgramManager& program_manager;
OGLBuffer swizzle_table_buffer;
OGLProgram astc_decoder_program;
OGLProgram block_linear_unswizzle_2d_program;
OGLProgram block_linear_unswizzle_3d_program;
@@ -203,7 +203,11 @@ inline void PushImageDescriptors(TextureCache& texture_cache,
const VideoCommon::ImageViewId image_view_id{(views++)->id};
const VideoCommon::SamplerId sampler_id{*(samplers++)};
ImageView& image_view{texture_cache.GetImageView(image_view_id)};
const VkImageView vk_image_view{image_view.Handle(desc.type)};
VkImageView vk_image_view{image_view.Handle(desc.type)};
if (vk_image_view == VK_NULL_HANDLE) {
const VkImageView null_image_view{texture_cache.GetImageView(VideoCommon::NULL_IMAGE_VIEW_ID).Handle(desc.type)};
if (null_image_view != VK_NULL_HANDLE) vk_image_view = null_image_view;
}
const Sampler& sampler{texture_cache.GetSampler(sampler_id)};
const bool use_fallback_sampler{sampler.HasAddedAnisotropy() &&
!image_view.SupportsAnisotropy()};
@@ -653,71 +653,8 @@ void ASTCDecoderPass::Assemble(Image& image, const StagingBufferRef& map,
scheduler.Finish();
}
constexpr u32 BL3D_BINDING_SWIZZLE_TABLE = 0;
constexpr u32 BL3D_BINDING_INPUT_BUFFER = 1;
constexpr u32 BL3D_BINDING_OUTPUT_BUFFER = 2;
constexpr std::array<VkDescriptorSetLayoutBinding, 3> BL3D_DESCRIPTOR_SET_BINDINGS{{
{
.binding = BL3D_BINDING_SWIZZLE_TABLE,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // swizzle_table[]
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
{
.binding = BL3D_BINDING_INPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // block-linear input
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
{
.binding = BL3D_BINDING_OUTPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
}};
constexpr DescriptorBankInfo BL3D_BANK_INFO{
.uniform_buffers = 0,
.storage_buffers = 3,
.texture_buffers = 0,
.image_buffers = 0,
.textures = 0,
.images = 0,
.score = 3,
};
constexpr std::array<VkDescriptorUpdateTemplateEntry, 3>
BL3D_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY{{
{
.dstBinding = BL3D_BINDING_SWIZZLE_TABLE,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_SWIZZLE_TABLE * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
},
{
.dstBinding = BL3D_BINDING_INPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_INPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
},
{
.dstBinding = BL3D_BINDING_OUTPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_OUTPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
}
}};
constexpr u32 BL3D_BINDING_INPUT_BUFFER = 0;
constexpr u32 BL3D_BINDING_OUTPUT_BUFFER = 1;
struct alignas(16) BlockLinearUnswizzle3DPushConstants {
u32 blocks_dim[3]; // Offset 0
@@ -745,11 +682,50 @@ BlockLinearUnswizzle3DPass::BlockLinearUnswizzle3DPass(
DescriptorPool& descriptor_pool_,
StagingBufferPool& staging_buffer_pool_,
ComputePassDescriptorQueue& compute_pass_descriptor_queue_)
: ComputePass(
device_, scheduler_, descriptor_pool_,
BL3D_DESCRIPTOR_SET_BINDINGS,
BL3D_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY,
BL3D_BANK_INFO,
: ComputePass(device_, scheduler_, descriptor_pool_,
std::array<VkDescriptorSetLayoutBinding, 2>{{
{
.binding = BL3D_BINDING_INPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // block-linear input
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
{
.binding = BL3D_BINDING_OUTPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
}},
std::array<VkDescriptorUpdateTemplateEntry, 2>{{
{
.dstBinding = BL3D_BINDING_INPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_INPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
},
{
.dstBinding = BL3D_BINDING_OUTPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_OUTPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
}
}},
DescriptorBankInfo{
.uniform_buffers = 0,
.storage_buffers = 2,
.texture_buffers = 0,
.image_buffers = 0,
.textures = 0,
.images = 0,
.score = 2,
},
COMPUTE_PUSH_CONSTANT_RANGE<sizeof(BlockLinearUnswizzle3DPushConstants)>,
BLOCK_LINEAR_UNSWIZZLE_3D_BCN_COMP_SPV),
scheduler{scheduler_},
@@ -822,8 +798,6 @@ void BlockLinearUnswizzle3DPass::UnswizzleChunk(
pc.blocks_dim[2] = z_count; // Only process the count
compute_pass_descriptor_queue.Acquire(scheduler, 3);
compute_pass_descriptor_queue.AddBuffer(*image.runtime->swizzle_table_buffer, 0,
image.runtime->swizzle_table_size);
compute_pass_descriptor_queue.AddBuffer(swizzled.buffer,
sw.buffer_offset + swizzled.offset,
image.guest_size_bytes - sw.buffer_offset);
@@ -909,40 +909,6 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, Scheduler& sched
bl3d_unswizzle_pass.emplace(device, scheduler, descriptor_pool,
staging_buffer_pool, compute_pass_descriptor_queue);
}
// --- Create swizzle table buffer ---
{
auto table = Tegra::Texture::MakeSwizzleTable();
swizzle_table_size = static_cast<VkDeviceSize>(table.size() * sizeof(table[0]));
auto staging = staging_buffer_pool.Request(swizzle_table_size, MemoryUsage::Upload);
std::memcpy(staging.mapped_span.data(), table.data(), static_cast<size_t>(swizzle_table_size));
VkBufferCreateInfo ci{
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.size = swizzle_table_size,
.usage = VK_BUFFER_USAGE_STORAGE_BUFFER_BIT |
VK_BUFFER_USAGE_TRANSFER_DST_BIT |
VK_BUFFER_USAGE_TRANSFER_SRC_BIT,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
};
swizzle_table_buffer = memory_allocator.CreateBuffer(ci, MemoryUsage::DeviceLocal);
scheduler.RequestOutsideRenderPassOperationContext();
scheduler.Record([staging_buf = staging.buffer,
dst_buf = *swizzle_table_buffer,
size = swizzle_table_size,
src_off = staging.offset](vk::CommandBuffer cmdbuf) {
const VkBufferCopy region{
.srcOffset = src_off,
.dstOffset = 0,
.size = size,
};
cmdbuf.CopyBuffer(staging_buf, dst_buf, region);
});
}
}
void TextureCacheRuntime::Finish() {
@@ -2484,18 +2450,13 @@ void TextureCacheRuntime::AccelerateImageUpload(
if (!Settings::values.gpu_unswizzle_enabled.GetValue() || !bl3d_unswizzle_pass) {
if (IsPixelFormatBCn(image.info.format) && image.info.type == ImageType::e3D) {
ASSERT_MSG(false, "GPU unswizzle is disabled for BCn 3D texture");
ASSERT(false && "GPU unswizzle is disabled for BCn 3D texture");
}
ASSERT(false);
return;
}
if (bl3d_unswizzle_pass &&
IsPixelFormatBCn(image.info.format) &&
image.info.type == ImageType::e3D &&
image.info.resources.levels == 1 &&
image.info.resources.layers == 1) {
if (bl3d_unswizzle_pass && IsPixelFormatBCn(image.info.format) && image.info.type == ImageType::e3D && image.info.resources.levels == 1 && image.info.resources.layers == 1) {
return bl3d_unswizzle_pass->Unswizzle(image, map, swizzles, z_start, z_count);
}
@@ -1,4 +1,4 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
@@ -130,9 +130,6 @@ public:
std::optional<ASTCDecoderPass> astc_decoder_pass;
std::optional<BlockLinearUnswizzle3DPass> bl3d_unswizzle_pass;
vk::Buffer swizzle_table_buffer;
VkDeviceSize swizzle_table_size = 0;
std::optional<MSAACopyPass> msaa_copy_pass;
const Settings::ResolutionScalingInfo& resolution;
std::array<std::vector<VkFormat>, VideoCore::Surface::MaxPixelFormat> view_formats;
+2 -3
View File
@@ -255,8 +255,7 @@ std::optional<u64> GenericEnvironment::TryFindSize() {
static constexpr u64 SELF_BRANCH_A = 0xE2400FFFFF87000FULL;
static constexpr u64 SELF_BRANCH_B = 0xE2400FFFFF07000FULL;
static constexpr u64 MESA_EXIT_MASK = 0xFFF00000000F001FULL;
static constexpr u64 MESA_EXIT_VALUE = (0xE30ULL << 52) | (0x7ULL << 16) | 0xFULL;
static constexpr u64 EXIT_VALUE = 0xE30000000007000FULL;
code.resize(MAXIMUM_SIZE / INST_SIZE);
@@ -271,7 +270,7 @@ std::optional<u64> GenericEnvironment::TryFindSize() {
if (inst == SELF_BRANCH_A || inst == SELF_BRANCH_B) {
return offset + index;
}
if ((inst & MESA_EXIT_MASK) == MESA_EXIT_VALUE) {
if (!is_proprietary_driver && inst == EXIT_VALUE) {
return offset + index + INST_SIZE;
}
}
+3 -18
View File
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -23,24 +26,6 @@ constexpr u32 GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_
constexpr u32 SWIZZLE_X_BITS = 0b100101111;
constexpr u32 SWIZZLE_Y_BITS = 0b011010000;
using SwizzleTable = std::array<std::array<u32, GOB_SIZE_X>, GOB_SIZE_Y>;
/**
* This table represents the internal swizzle of a gob, in format 16 bytes x 2 sector packing.
* Calculates the offset of an (x, y) position within a swizzled texture.
* Taken from the Tegra X1 Technical Reference Manual. pages 1187-1188
*/
constexpr SwizzleTable MakeSwizzleTable() {
SwizzleTable table{};
for (u32 y = 0; y < table.size(); ++y) {
for (u32 x = 0; x < table[0].size(); ++x) {
table[y][x] = ((x % 64) / 32) * 256 + ((y % 8) / 2) * 64 + ((x % 32) / 16) * 32 +
(y % 2) * 16 + (x % 16);
}
}
return table;
}
/// Unswizzles a block linear texture into linear memory.
void UnswizzleTexture(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel,
u32 width, u32 height, u32 depth, u32 block_height, u32 block_depth,
-1
View File
@@ -371,7 +371,6 @@ int main(int argc, char** argv) {
#ifdef _WIN32
Common::Windows::SetCurrentTimerResolutionToMaximum();
system.CoreTiming().SetTimerResolutionNs(Common::Windows::GetCurrentTimerResolution());
#endif
system.SetContentProvider(std::make_unique<FileSys::ContentProviderUnion>());