video_core: Preserve multisampled textures when overlap

kernel: clone fpu status on CreateThread

CMakeLists.txt

@@ -477,8 +477,8 @@ if (APPLE)
     find_library(COCOA_LIBRARY Cocoa)
     set(PLATFORM_LIBRARIES ${COCOA_LIBRARY} ${IOKIT_LIBRARY} ${COREVIDEO_LIBRARY})
 elseif (WIN32)
-    # WSAPoll and SHGetKnownFolderPath (AppData/Roaming) didn't exist before WinNT 6.x (Vista)
-    add_definitions(-D_WIN32_WINNT=0x0600 -DWINVER=0x0600)
+    # Target Windows 10
+    add_definitions(-D_WIN32_WINNT=0x0A00 -DWINVER=0x0A00)
     set(PLATFORM_LIBRARIES winmm ws2_32 iphlpapi)
     if (MINGW)
         # PSAPI is the Process Status API

dist/yuzu.manifest

@@ -36,12 +36,6 @@ SPDX-License-Identifier: GPL-2.0-or-later
     <application>
       <!-- Windows 10 -->
       <supportedOS Id="{8e0f7a12-bfb3-4fe8-b9a5-48fd50a15a9a}"/>
-      <!-- Windows 8.1 -->
-      <supportedOS Id="{1f676c76-80e1-4239-95bb-83d0f6d0da78}"/>
-      <!-- Windows 8 -->
-      <supportedOS Id="{4a2f28e3-53b9-4441-ba9c-d69d4a4a6e38}"/>
-      <!-- Windows 7 -->
-      <supportedOS Id="{35138b9a-5d96-4fbd-8e2d-a2440225f93a}"/>
     </application>
   </compatibility>
   <trustInfo

src/common/CMakeLists.txt

@@ -91,6 +91,7 @@ add_library(common STATIC
     multi_level_page_table.h
     nvidia_flags.cpp
     nvidia_flags.h
+    overflow.h
     page_table.cpp
     page_table.h
     param_package.cpp
@@ -113,6 +114,8 @@ add_library(common STATIC
     socket_types.h
     spin_lock.cpp
     spin_lock.h
+    steady_clock.cpp
+    steady_clock.h
     stream.cpp
     stream.h
     string_util.cpp
@@ -142,6 +145,14 @@ add_library(common STATIC
     zstd_compression.h
 )
 
+if (WIN32)
+  target_sources(common PRIVATE
+    windows/timer_resolution.cpp
+    windows/timer_resolution.h
+  )
+  target_link_libraries(common PRIVATE ntdll)
+endif()
+
 if(ARCHITECTURE_x86_64)
     target_sources(common
         PRIVATE

src/common/bit_cast.h

@@ -3,19 +3,21 @@
 
 #pragma once
 
-#include <cstring>
-#include <type_traits>
+#include <version>
+
+#ifdef __cpp_lib_bit_cast
+#include <bit>
+#endif
 
 namespace Common {
 
 template <typename To, typename From>
-[[nodiscard]] std::enable_if_t<sizeof(To) == sizeof(From) && std::is_trivially_copyable_v<From> &&
-                                   std::is_trivially_copyable_v<To>,
-                               To>
-BitCast(const From& src) noexcept {
-    To dst;
-    std::memcpy(&dst, &src, sizeof(To));
-    return dst;
+constexpr inline To BitCast(const From& from) {
+#ifdef __cpp_lib_bit_cast
+    return std::bit_cast<To>(from);
+#else
+    return __builtin_bit_cast(To, from);
+#endif
 }
 
 } // namespace Common

src/common/input.h

@@ -46,7 +46,7 @@ enum class PollingMode {
     // Constant polling of buttons, analogs and motion data
     Active,
     // Only update on button change, digital analogs
-    Pasive,
+    Passive,
     // Enable near field communication polling
     NFC,
     // Enable infrared camera polling

src/common/overflow.h

@@ -0,0 +1,22 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <type_traits>
+#include "bit_cast.h"
+
+namespace Common {
+
+template <typename T>
+    requires(std::is_integral_v<T> && std::is_signed_v<T>)
+inline T WrappingAdd(T lhs, T rhs) {
+    using U = std::make_unsigned_t<T>;
+
+    U lhs_u = BitCast<U>(lhs);
+    U rhs_u = BitCast<U>(rhs);
+
+    return BitCast<T>(lhs_u + rhs_u);
+}
+
+} // namespace Common

src/common/settings.h

@@ -503,7 +503,7 @@ struct Values {
     Setting<bool> tas_loop{false, "tas_loop"};
 
     Setting<bool> mouse_panning{false, "mouse_panning"};
-    Setting<u8, true> mouse_panning_sensitivity{10, 1, 100, "mouse_panning_sensitivity"};
+    Setting<u8, true> mouse_panning_sensitivity{50, 1, 100, "mouse_panning_sensitivity"};
     Setting<bool> mouse_enabled{false, "mouse_enabled"};
 
     Setting<bool> emulate_analog_keyboard{false, "emulate_analog_keyboard"};

src/common/steady_clock.cpp

@@ -0,0 +1,56 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#if defined(_WIN32)
+#include <windows.h>
+#else
+#include <time.h>
+#endif
+
+#include "common/steady_clock.h"
+
+namespace Common {
+
+#ifdef _WIN32
+static s64 WindowsQueryPerformanceFrequency() {
+    LARGE_INTEGER frequency;
+    QueryPerformanceFrequency(&frequency);
+    return frequency.QuadPart;
+}
+
+static s64 WindowsQueryPerformanceCounter() {
+    LARGE_INTEGER counter;
+    QueryPerformanceCounter(&counter);
+    return counter.QuadPart;
+}
+#endif
+
+SteadyClock::time_point SteadyClock::Now() noexcept {
+#if defined(_WIN32)
+    static const auto freq = WindowsQueryPerformanceFrequency();
+    const auto counter = WindowsQueryPerformanceCounter();
+
+    // 10 MHz is a very common QPC frequency on modern PCs.
+    // Optimizing for this specific frequency can double the performance of
+    // this function by avoiding the expensive frequency conversion path.
+    static constexpr s64 TenMHz = 10'000'000;
+
+    if (freq == TenMHz) [[likely]] {
+        static_assert(period::den % TenMHz == 0);
+        static constexpr s64 Multiplier = period::den / TenMHz;
+        return time_point{duration{counter * Multiplier}};
+    }
+
+    const auto whole = (counter / freq) * period::den;
+    const auto part = (counter % freq) * period::den / freq;
+    return time_point{duration{whole + part}};
+#elif defined(__APPLE__)
+    return time_point{duration{clock_gettime_nsec_np(CLOCK_MONOTONIC_RAW)}};
+#else
+    timespec ts;
+    clock_gettime(CLOCK_MONOTONIC, &ts);
+    return time_point{std::chrono::seconds{ts.tv_sec} + std::chrono::nanoseconds{ts.tv_nsec}};
+#endif
+}
+
+}; // namespace Common

src/common/steady_clock.h

@@ -0,0 +1,23 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <chrono>
+
+#include "common/common_types.h"
+
+namespace Common {
+
+struct SteadyClock {
+    using rep = s64;
+    using period = std::nano;
+    using duration = std::chrono::nanoseconds;
+    using time_point = std::chrono::time_point<SteadyClock>;
+
+    static constexpr bool is_steady = true;
+
+    [[nodiscard]] static time_point Now() noexcept;
+};
+
+} // namespace Common

src/common/wall_clock.cpp

@@ -1,6 +1,7 @@
 // SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
+#include "common/steady_clock.h"
 #include "common/uint128.h"
 #include "common/wall_clock.h"
 
@@ -11,45 +12,32 @@
 
 namespace Common {
 
-using base_timer = std::chrono::steady_clock;
-using base_time_point = std::chrono::time_point<base_timer>;
-
 class StandardWallClock final : public WallClock {
 public:
     explicit StandardWallClock(u64 emulated_cpu_frequency_, u64 emulated_clock_frequency_)
-        : WallClock(emulated_cpu_frequency_, emulated_clock_frequency_, false) {
-        start_time = base_timer::now();
-    }
+        : WallClock{emulated_cpu_frequency_, emulated_clock_frequency_, false},
+          start_time{SteadyClock::Now()} {}
 
     std::chrono::nanoseconds GetTimeNS() override {
-        base_time_point current = base_timer::now();
-        auto elapsed = current - start_time;
-        return std::chrono::duration_cast<std::chrono::nanoseconds>(elapsed);
+        return SteadyClock::Now() - start_time;
     }
 
     std::chrono::microseconds GetTimeUS() override {
-        base_time_point current = base_timer::now();
-        auto elapsed = current - start_time;
-        return std::chrono::duration_cast<std::chrono::microseconds>(elapsed);
+        return std::chrono::duration_cast<std::chrono::microseconds>(GetTimeNS());
     }
 
     std::chrono::milliseconds GetTimeMS() override {
-        base_time_point current = base_timer::now();
-        auto elapsed = current - start_time;
-        return std::chrono::duration_cast<std::chrono::milliseconds>(elapsed);
+        return std::chrono::duration_cast<std::chrono::milliseconds>(GetTimeNS());
     }
 
     u64 GetClockCycles() override {
-        std::chrono::nanoseconds time_now = GetTimeNS();
-        const u128 temporary =
-            Common::Multiply64Into128(time_now.count(), emulated_clock_frequency);
-        return Common::Divide128On32(temporary, 1000000000).first;
+        const u128 temp = Common::Multiply64Into128(GetTimeNS().count(), emulated_clock_frequency);
+        return Common::Divide128On32(temp, NS_RATIO).first;
     }
 
     u64 GetCPUCycles() override {
-        std::chrono::nanoseconds time_now = GetTimeNS();
-        const u128 temporary = Common::Multiply64Into128(time_now.count(), emulated_cpu_frequency);
-        return Common::Divide128On32(temporary, 1000000000).first;
+        const u128 temp = Common::Multiply64Into128(GetTimeNS().count(), emulated_cpu_frequency);
+        return Common::Divide128On32(temp, NS_RATIO).first;
     }
 
     void Pause([[maybe_unused]] bool is_paused) override {
@@ -57,7 +45,7 @@ public:
     }
 
 private:
-    base_time_point start_time;
+    SteadyClock::time_point start_time;
 };
 
 #ifdef ARCHITECTURE_x86_64
@@ -93,4 +81,9 @@ std::unique_ptr<WallClock> CreateBestMatchingClock(u64 emulated_cpu_frequency,
 
 #endif
 
+std::unique_ptr<WallClock> CreateStandardWallClock(u64 emulated_cpu_frequency,
+                                                   u64 emulated_clock_frequency) {
+    return std::make_unique<StandardWallClock>(emulated_cpu_frequency, emulated_clock_frequency);
+}
+
 } // namespace Common

src/common/wall_clock.h

@@ -55,4 +55,7 @@ private:
 [[nodiscard]] std::unique_ptr<WallClock> CreateBestMatchingClock(u64 emulated_cpu_frequency,
                                                                  u64 emulated_clock_frequency);
 
+[[nodiscard]] std::unique_ptr<WallClock> CreateStandardWallClock(u64 emulated_cpu_frequency,
+                                                                 u64 emulated_clock_frequency);
+
 } // namespace Common

src/common/windows/timer_resolution.cpp

@@ -0,0 +1,109 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <windows.h>
+
+#include "common/windows/timer_resolution.h"
+
+extern "C" {
+// http://undocumented.ntinternals.net/index.html?page=UserMode%2FUndocumented%20Functions%2FTime%2FNtQueryTimerResolution.html
+NTSYSAPI LONG NTAPI NtQueryTimerResolution(PULONG MinimumResolution, PULONG MaximumResolution,
+                                           PULONG CurrentResolution);
+
+// http://undocumented.ntinternals.net/index.html?page=UserMode%2FUndocumented%20Functions%2FTime%2FNtSetTimerResolution.html
+NTSYSAPI LONG NTAPI NtSetTimerResolution(ULONG DesiredResolution, BOOLEAN SetResolution,
+                                         PULONG CurrentResolution);
+
+// http://undocumented.ntinternals.net/index.html?page=UserMode%2FUndocumented%20Functions%2FNT%20Objects%2FThread%2FNtDelayExecution.html
+NTSYSAPI LONG NTAPI NtDelayExecution(BOOLEAN Alertable, PLARGE_INTEGER DelayInterval);
+}
+
+// Defines for compatibility with older Windows 10 SDKs.
+
+#ifndef PROCESS_POWER_THROTTLING_EXECUTION_SPEED
+#define PROCESS_POWER_THROTTLING_EXECUTION_SPEED 0x1
+#endif
+#ifndef PROCESS_POWER_THROTTLING_IGNORE_TIMER_RESOLUTION
+#define PROCESS_POWER_THROTTLING_IGNORE_TIMER_RESOLUTION 0x4
+#endif
+
+namespace Common::Windows {
+
+namespace {
+
+using namespace std::chrono;
+
+constexpr nanoseconds ToNS(ULONG hundred_ns) {
+    return nanoseconds{hundred_ns * 100};
+}
+
+constexpr ULONG ToHundredNS(nanoseconds ns) {
+    return static_cast<ULONG>(ns.count()) / 100;
+}
+
+struct TimerResolution {
+    std::chrono::nanoseconds minimum;
+    std::chrono::nanoseconds maximum;
+    std::chrono::nanoseconds current;
+};
+
+TimerResolution GetTimerResolution() {
+    ULONG MinimumTimerResolution;
+    ULONG MaximumTimerResolution;
+    ULONG CurrentTimerResolution;
+    NtQueryTimerResolution(&MinimumTimerResolution, &MaximumTimerResolution,
+                           &CurrentTimerResolution);
+    return {
+        .minimum{ToNS(MinimumTimerResolution)},
+        .maximum{ToNS(MaximumTimerResolution)},
+        .current{ToNS(CurrentTimerResolution)},
+    };
+}
+
+void SetHighQoS() {
+    // https://learn.microsoft.com/en-us/windows/win32/procthread/quality-of-service
+    PROCESS_POWER_THROTTLING_STATE PowerThrottling{
+        .Version{PROCESS_POWER_THROTTLING_CURRENT_VERSION},
+        .ControlMask{PROCESS_POWER_THROTTLING_EXECUTION_SPEED |
+                     PROCESS_POWER_THROTTLING_IGNORE_TIMER_RESOLUTION},
+        .StateMask{},
+    };
+    SetProcessInformation(GetCurrentProcess(), ProcessPowerThrottling, &PowerThrottling,
+                          sizeof(PROCESS_POWER_THROTTLING_STATE));
+}
+
+} // Anonymous namespace
+
+nanoseconds GetMinimumTimerResolution() {
+    return GetTimerResolution().minimum;
+}
+
+nanoseconds GetMaximumTimerResolution() {
+    return GetTimerResolution().maximum;
+}
+
+nanoseconds GetCurrentTimerResolution() {
+    return GetTimerResolution().current;
+}
+
+nanoseconds SetCurrentTimerResolution(nanoseconds timer_resolution) {
+    // Set the timer resolution, and return the current timer resolution.
+    const auto DesiredTimerResolution = ToHundredNS(timer_resolution);
+    ULONG CurrentTimerResolution;
+    NtSetTimerResolution(DesiredTimerResolution, TRUE, &CurrentTimerResolution);
+    return ToNS(CurrentTimerResolution);
+}
+
+nanoseconds SetCurrentTimerResolutionToMaximum() {
+    SetHighQoS();
+    return SetCurrentTimerResolution(GetMaximumTimerResolution());
+}
+
+void SleepForOneTick() {
+    LARGE_INTEGER DelayInterval{
+        .QuadPart{-1},
+    };
+    NtDelayExecution(FALSE, &DelayInterval);
+}
+
+} // namespace Common::Windows

src/common/windows/timer_resolution.h

@@ -0,0 +1,38 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <chrono>
+
+namespace Common::Windows {
+
+/// Returns the minimum (least precise) supported timer resolution in nanoseconds.
+std::chrono::nanoseconds GetMinimumTimerResolution();
+
+/// Returns the maximum (most precise) supported timer resolution in nanoseconds.
+std::chrono::nanoseconds GetMaximumTimerResolution();
+
+/// Returns the current timer resolution in nanoseconds.
+std::chrono::nanoseconds GetCurrentTimerResolution();
+
+/**
+ * Sets the current timer resolution.
+ *
+ * @param timer_resolution Timer resolution in nanoseconds.
+ *
+ * @returns The current timer resolution.
+ */
+std::chrono::nanoseconds SetCurrentTimerResolution(std::chrono::nanoseconds timer_resolution);
+
+/**
+ * Sets the current timer resolution to the maximum supported timer resolution.
+ *
+ * @returns The current timer resolution.
+ */
+std::chrono::nanoseconds SetCurrentTimerResolutionToMaximum();
+
+/// Sleep for one tick of the current timer resolution.
+void SleepForOneTick();
+
+} // namespace Common::Windows

src/common/x64/native_clock.cpp

@@ -6,6 +6,7 @@
 #include <thread>
 
 #include "common/atomic_ops.h"
+#include "common/steady_clock.h"
 #include "common/uint128.h"
 #include "common/x64/native_clock.h"
 
@@ -39,6 +40,12 @@ static u64 FencedRDTSC() {
 }
 #endif
 
+template <u64 Nearest>
+static u64 RoundToNearest(u64 value) {
+    const auto mod = value % Nearest;
+    return mod >= (Nearest / 2) ? (value - mod + Nearest) : (value - mod);
+}
+
 u64 EstimateRDTSCFrequency() {
     // Discard the first result measuring the rdtsc.
     FencedRDTSC();
@@ -46,18 +53,18 @@ u64 EstimateRDTSCFrequency() {
     FencedRDTSC();
 
     // Get the current time.
-    const auto start_time = std::chrono::steady_clock::now();
+    const auto start_time = Common::SteadyClock::Now();
     const u64 tsc_start = FencedRDTSC();
-    // Wait for 200 milliseconds.
-    std::this_thread::sleep_for(std::chrono::milliseconds{200});
-    const auto end_time = std::chrono::steady_clock::now();
+    // Wait for 250 milliseconds.
+    std::this_thread::sleep_for(std::chrono::milliseconds{250});
+    const auto end_time = Common::SteadyClock::Now();
     const u64 tsc_end = FencedRDTSC();
     // Calculate differences.
     const u64 timer_diff = static_cast<u64>(
         std::chrono::duration_cast<std::chrono::nanoseconds>(end_time - start_time).count());
     const u64 tsc_diff = tsc_end - tsc_start;
     const u64 tsc_freq = MultiplyAndDivide64(tsc_diff, 1000000000ULL, timer_diff);
-    return tsc_freq;
+    return RoundToNearest<1000>(tsc_freq);
 }
 
 namespace X64 {

src/core/core_timing.cpp

@@ -6,6 +6,10 @@
 #include <string>
 #include <tuple>
 
+#ifdef _WIN32
+#include "common/windows/timer_resolution.h"
+#endif
+
 #include "common/microprofile.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
@@ -38,7 +42,8 @@ struct CoreTiming::Event {
 };
 
 CoreTiming::CoreTiming()
-    : clock{Common::CreateBestMatchingClock(Hardware::BASE_CLOCK_RATE, Hardware::CNTFREQ)} {}
+    : cpu_clock{Common::CreateBestMatchingClock(Hardware::BASE_CLOCK_RATE, Hardware::CNTFREQ)},
+      event_clock{Common::CreateStandardWallClock(Hardware::BASE_CLOCK_RATE, Hardware::CNTFREQ)} {}
 
 CoreTiming::~CoreTiming() {
     Reset();
@@ -185,15 +190,15 @@ void CoreTiming::ResetTicks() {
 }
 
 u64 CoreTiming::GetCPUTicks() const {
-    if (is_multicore) {
-        return clock->GetCPUCycles();
+    if (is_multicore) [[likely]] {
+        return cpu_clock->GetCPUCycles();
     }
     return ticks;
 }
 
 u64 CoreTiming::GetClockTicks() const {
-    if (is_multicore) {
-        return clock->GetClockCycles();
+    if (is_multicore) [[likely]] {
+        return cpu_clock->GetClockCycles();
     }
     return CpuCyclesToClockCycles(ticks);
 }
@@ -252,21 +257,20 @@ void CoreTiming::ThreadLoop() {
             const auto next_time = Advance();
             if (next_time) {
                 // There are more events left in the queue, wait until the next event.
-                const auto wait_time = *next_time - GetGlobalTimeNs().count();
+                auto wait_time = *next_time - GetGlobalTimeNs().count();
                 if (wait_time > 0) {
 #ifdef _WIN32
-                    // Assume a timer resolution of 1ms.
-                    static constexpr s64 TimerResolutionNS = 1000000;
+                    const auto timer_resolution_ns =
+                        Common::Windows::GetCurrentTimerResolution().count();
 
-                    // Sleep in discrete intervals of the timer resolution, and spin the rest.
-                    const auto sleep_time = wait_time - (wait_time % TimerResolutionNS);
-                    if (sleep_time > 0) {
-                        event.WaitFor(std::chrono::nanoseconds(sleep_time));
-                    }
+                    while (!paused && !event.IsSet() && wait_time > 0) {
+                        wait_time = *next_time - GetGlobalTimeNs().count();
 
-                    while (!paused && !event.IsSet() && GetGlobalTimeNs().count() < *next_time) {
-                        // Yield to reduce thread starvation.
-                        std::this_thread::yield();
+                        if (wait_time >= timer_resolution_ns) {
+                            Common::Windows::SleepForOneTick();
+                        } else {
+                            std::this_thread::yield();
+                        }
                     }
 
                     if (event.IsSet()) {
@@ -285,9 +289,9 @@ void CoreTiming::ThreadLoop() {
         }
 
         paused_set = true;
-        clock->Pause(true);
+        event_clock->Pause(true);
         pause_event.Wait();
-        clock->Pause(false);
+        event_clock->Pause(false);
     }
 }
 
@@ -303,16 +307,23 @@ void CoreTiming::Reset() {
     has_started = false;
 }
 
+std::chrono::nanoseconds CoreTiming::GetCPUTimeNs() const {
+    if (is_multicore) [[likely]] {
+        return cpu_clock->GetTimeNS();
+    }
+    return CyclesToNs(ticks);
+}
+
 std::chrono::nanoseconds CoreTiming::GetGlobalTimeNs() const {
-    if (is_multicore) {
-        return clock->GetTimeNS();
+    if (is_multicore) [[likely]] {
+        return event_clock->GetTimeNS();
     }
     return CyclesToNs(ticks);
 }
 
 std::chrono::microseconds CoreTiming::GetGlobalTimeUs() const {
-    if (is_multicore) {
-        return clock->GetTimeUS();
+    if (is_multicore) [[likely]] {
+        return event_clock->GetTimeUS();
     }
     return CyclesToUs(ticks);
 }

src/core/core_timing.h

@@ -122,6 +122,9 @@ public:
     /// Returns current time in emulated in Clock cycles
     u64 GetClockTicks() const;
 
+    /// Returns current time in nanoseconds.
+    std::chrono::nanoseconds GetCPUTimeNs() const;
+
     /// Returns current time in microseconds.
     std::chrono::microseconds GetGlobalTimeUs() const;
 
@@ -139,7 +142,8 @@ private:
 
     void Reset();
 
-    std::unique_ptr<Common::WallClock> clock;
+    std::unique_ptr<Common::WallClock> cpu_clock;
+    std::unique_ptr<Common::WallClock> event_clock;
 
     s64 global_timer = 0;
 

src/core/hardware_properties.h

@@ -13,11 +13,9 @@ namespace Core {
 
 namespace Hardware {
 
-// The below clock rate is based on Switch's clockspeed being widely known as 1.020GHz
-// The exact value used is of course unverified.
-constexpr u64 BASE_CLOCK_RATE = 1019215872; // Switch cpu frequency is 1020MHz un/docked
-constexpr u64 CNTFREQ = 19200000;           // Switch's hardware clock speed
-constexpr u32 NUM_CPU_CORES = 4;            // Number of CPU Cores
+constexpr u64 BASE_CLOCK_RATE = 1'020'000'000; // Default CPU Frequency = 1020 MHz
+constexpr u64 CNTFREQ = 19'200'000;            // CNTPCT_EL0 Frequency = 19.2 MHz
+constexpr u32 NUM_CPU_CORES = 4;               // Number of CPU Cores
 
 // Virtual to Physical core map.
 constexpr std::array<s32, Common::BitSize<u64>()> VirtualToPhysicalCoreMap{

src/core/hle/kernel/k_address_space_info.cpp

@@ -44,11 +44,11 @@ const KAddressSpaceInfo& GetAddressSpaceInfo(size_t width, KAddressSpaceInfo::Ty
 
 } // namespace
 
-uintptr_t KAddressSpaceInfo::GetAddressSpaceStart(size_t width, KAddressSpaceInfo::Type type) {
+std::size_t KAddressSpaceInfo::GetAddressSpaceStart(size_t width, KAddressSpaceInfo::Type type) {
     return GetAddressSpaceInfo(width, type).address;
 }
 
-size_t KAddressSpaceInfo::GetAddressSpaceSize(size_t width, KAddressSpaceInfo::Type type) {
+std::size_t KAddressSpaceInfo::GetAddressSpaceSize(size_t width, KAddressSpaceInfo::Type type) {
     return GetAddressSpaceInfo(width, type).size;
 }
 

src/core/hle/kernel/k_address_space_info.h

@@ -18,7 +18,7 @@ struct KAddressSpaceInfo final {
         Count,
     };
 
-    static u64 GetAddressSpaceStart(std::size_t width, Type type);
+    static std::size_t GetAddressSpaceStart(std::size_t width, Type type);
     static std::size_t GetAddressSpaceSize(std::size_t width, Type type);
 
     const std::size_t bit_width{};

src/core/hle/kernel/k_device_address_space.h

@@ -21,9 +21,9 @@ public:
     ~KDeviceAddressSpace();
 
     Result Initialize(u64 address, u64 size);
-    void Finalize();
+    void Finalize() override;
 
-    bool IsInitialized() const {
+    bool IsInitialized() const override {
         return m_is_initialized;
     }
     static void PostDestroy(uintptr_t arg) {}

src/core/hle/kernel/k_resource_limit.cpp

@@ -2,6 +2,7 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 
 #include "common/assert.h"
+#include "common/overflow.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/hle/kernel/k_resource_limit.h"
@@ -104,7 +105,7 @@ bool KResourceLimit::Reserve(LimitableResource which, s64 value, s64 timeout) {
         ASSERT(current_hints[index] <= current_values[index]);
 
         // If we would overflow, don't allow to succeed.
-        if (current_values[index] + value <= current_values[index]) {
+        if (Common::WrappingAdd(current_values[index], value) <= current_values[index]) {
             break;
         }
 

src/core/hle/kernel/k_thread.cpp

@@ -49,6 +49,7 @@ static void ResetThreadContext32(Core::ARM_Interface::ThreadContext32& context,
     context.cpu_registers[0] = arg;
     context.cpu_registers[15] = entry_point;
     context.cpu_registers[13] = stack_top;
+    context.fpscr = 0;
 }
 
 static void ResetThreadContext64(Core::ARM_Interface::ThreadContext64& context, VAddr stack_top,
@@ -58,8 +59,8 @@ static void ResetThreadContext64(Core::ARM_Interface::ThreadContext64& context,
     context.cpu_registers[18] = Kernel::KSystemControl::GenerateRandomU64() | 1;
     context.pc = entry_point;
     context.sp = stack_top;
-    // TODO(merry): Perform a hardware test to determine the below value.
     context.fpcr = 0;
+    context.fpsr = 0;
 }
 } // namespace
 
@@ -815,6 +816,27 @@ void KThread::Continue() {
     KScheduler::OnThreadStateChanged(kernel, this, old_state);
 }
 
+void KThread::CloneFpuStatus() {
+    // We shouldn't reach here when starting kernel threads.
+    ASSERT(this->GetOwnerProcess() != nullptr);
+    ASSERT(this->GetOwnerProcess() == GetCurrentProcessPointer(kernel));
+
+    if (this->GetOwnerProcess()->Is64BitProcess()) {
+        // Clone FPSR and FPCR.
+        ThreadContext64 cur_ctx{};
+        kernel.System().CurrentArmInterface().SaveContext(cur_ctx);
+
+        this->GetContext64().fpcr = cur_ctx.fpcr;
+        this->GetContext64().fpsr = cur_ctx.fpsr;
+    } else {
+        // Clone FPSCR.
+        ThreadContext32 cur_ctx{};
+        kernel.System().CurrentArmInterface().SaveContext(cur_ctx);
+
+        this->GetContext32().fpscr = cur_ctx.fpscr;
+    }
+}
+
 Result KThread::SetActivity(Svc::ThreadActivity activity) {
     // Lock ourselves.
     KScopedLightLock lk(activity_pause_lock);

src/core/hle/kernel/k_thread.h

@@ -254,6 +254,8 @@ public:
         thread_context_32.tpidr = static_cast<u32>(value);
     }
 
+    void CloneFpuStatus();
+
     [[nodiscard]] ThreadContext32& GetContext32() {
         return thread_context_32;
     }

src/core/hle/kernel/svc/svc_synchronization.cpp

@@ -48,19 +48,15 @@ Result ResetSignal(Core::System& system, Handle handle) {
     return ResultInvalidHandle;
 }
 
-/// Wait for the given handles to synchronize, timeout after the specified nanoseconds
-Result WaitSynchronization(Core::System& system, s32* index, VAddr handles_address, s32 num_handles,
-                           s64 nano_seconds) {
-    LOG_TRACE(Kernel_SVC, "called handles_address=0x{:X}, num_handles={}, nano_seconds={}",
-              handles_address, num_handles, nano_seconds);
-
+static Result WaitSynchronization(Core::System& system, int32_t* out_index, const Handle* handles,
+                                  int32_t num_handles, int64_t timeout_ns) {
     // Ensure number of handles is valid.
-    R_UNLESS(0 <= num_handles && num_handles <= ArgumentHandleCountMax, ResultOutOfRange);
+    R_UNLESS(0 <= num_handles && num_handles <= Svc::ArgumentHandleCountMax, ResultOutOfRange);
 
+    // Get the synchronization context.
     auto& kernel = system.Kernel();
+    auto& handle_table = GetCurrentProcess(kernel).GetHandleTable();
     std::vector<KSynchronizationObject*> objs(num_handles);
-    const auto& handle_table = GetCurrentProcess(kernel).GetHandleTable();
-    Handle* handles = system.Memory().GetPointer<Handle>(handles_address);
 
     // Copy user handles.
     if (num_handles > 0) {
@@ -68,21 +64,38 @@ Result WaitSynchronization(Core::System& system, s32* index, VAddr handles_addre
         R_UNLESS(handle_table.GetMultipleObjects<KSynchronizationObject>(objs.data(), handles,
                                                                          num_handles),
                  ResultInvalidHandle);
-        for (const auto& obj : objs) {
-            kernel.RegisterInUseObject(obj);
-        }
     }
 
     // Ensure handles are closed when we're done.
     SCOPE_EXIT({
-        for (s32 i = 0; i < num_handles; ++i) {
-            kernel.UnregisterInUseObject(objs[i]);
+        for (auto i = 0; i < num_handles; ++i) {
             objs[i]->Close();
         }
     });
 
-    return KSynchronizationObject::Wait(kernel, index, objs.data(), static_cast<s32>(objs.size()),
-                                        nano_seconds);
+    // Wait on the objects.
+    Result res = KSynchronizationObject::Wait(kernel, out_index, objs.data(),
+                                              static_cast<s32>(objs.size()), timeout_ns);
+
+    R_SUCCEED_IF(res == ResultSessionClosed);
+    R_RETURN(res);
+}
+
+/// Wait for the given handles to synchronize, timeout after the specified nanoseconds
+Result WaitSynchronization(Core::System& system, int32_t* out_index, VAddr user_handles,
+                           int32_t num_handles, int64_t timeout_ns) {
+    LOG_TRACE(Kernel_SVC, "called user_handles={:#x}, num_handles={}, timeout_ns={}", user_handles,
+              num_handles, timeout_ns);
+
+    // Ensure number of handles is valid.
+    R_UNLESS(0 <= num_handles && num_handles <= Svc::ArgumentHandleCountMax, ResultOutOfRange);
+
+    std::vector<Handle> handles(num_handles);
+    if (num_handles > 0) {
+        system.Memory().ReadBlock(user_handles, handles.data(), num_handles * sizeof(Handle));
+    }
+
+    R_RETURN(WaitSynchronization(system, out_index, handles.data(), num_handles, timeout_ns));
 }
 
 /// Resumes a thread waiting on WaitSynchronization

src/core/hle/kernel/svc/svc_thread.cpp

@@ -82,6 +82,9 @@ Result CreateThread(Core::System& system, Handle* out_handle, VAddr entry_point,
     // Commit the thread reservation.
     thread_reservation.Commit();
 
+    // Clone the current fpu status to the new thread.
+    thread->CloneFpuStatus();
+
     // Register the new thread.
     KThread::Register(kernel, thread);
 

src/input_common/drivers/joycon.cpp

@@ -307,8 +307,8 @@ Common::Input::DriverResult Joycons::SetPollingMode(const PadIdentifier& identif
     switch (polling_mode) {
     case Common::Input::PollingMode::Active:
         return static_cast<Common::Input::DriverResult>(handle->SetActiveMode());
-    case Common::Input::PollingMode::Pasive:
-        return static_cast<Common::Input::DriverResult>(handle->SetPasiveMode());
+    case Common::Input::PollingMode::Passive:
+        return static_cast<Common::Input::DriverResult>(handle->SetPassiveMode());
     case Common::Input::PollingMode::IR:
         return static_cast<Common::Input::DriverResult>(handle->SetIrMode());
     case Common::Input::PollingMode::NFC:

src/input_common/drivers/mouse.cpp

@@ -3,6 +3,7 @@
 
 #include <thread>
 #include <fmt/format.h>
+#include <math.h>
 
 #include "common/param_package.h"
 #include "common/settings.h"
@@ -11,8 +12,9 @@
 
 namespace InputCommon {
 constexpr int update_time = 10;
-constexpr float default_stick_sensitivity = 0.022f;
-constexpr float default_motion_sensitivity = 0.008f;
+constexpr float default_stick_sensitivity = 0.0044f;
+constexpr float default_motion_sensitivity = 0.0003f;
+constexpr float maximum_rotation_speed = 2.0f;
 constexpr int mouse_axis_x = 0;
 constexpr int mouse_axis_y = 1;
 constexpr int wheel_axis_x = 2;
@@ -99,11 +101,13 @@ void Mouse::UpdateMotionInput() {
     const float sensitivity =
         Settings::values.mouse_panning_sensitivity.GetValue() * default_motion_sensitivity;
 
-    // Slow movement by 7%
-    if (Settings::values.mouse_panning) {
-        last_motion_change *= 0.93f;
-    } else {
-        last_motion_change.z *= 0.93f;
+    const float rotation_velocity = std::sqrt(last_motion_change.x * last_motion_change.x +
+                                              last_motion_change.y * last_motion_change.y);
+
+    if (rotation_velocity > maximum_rotation_speed / sensitivity) {
+        const float multiplier = maximum_rotation_speed / rotation_velocity / sensitivity;
+        last_motion_change.x = last_motion_change.x * multiplier;
+        last_motion_change.y = last_motion_change.y * multiplier;
     }
 
     const BasicMotion motion_data{
@@ -116,6 +120,12 @@ void Mouse::UpdateMotionInput() {
         .delta_timestamp = update_time * 1000,
     };
 
+    if (Settings::values.mouse_panning) {
+        last_motion_change.x = 0;
+        last_motion_change.y = 0;
+    }
+    last_motion_change.z = 0;
+
     SetMotion(motion_identifier, 0, motion_data);
 }
 
@@ -125,7 +135,7 @@ void Mouse::Move(int x, int y, int center_x, int center_y) {
 
         auto mouse_change =
             (Common::MakeVec(x, y) - Common::MakeVec(center_x, center_y)).Cast<float>();
-        Common::Vec3<float> motion_change{-mouse_change.y, -mouse_change.x, last_motion_change.z};
+        last_motion_change += {-mouse_change.y, -mouse_change.x, last_motion_change.z};
 
         const auto move_distance = mouse_change.Length();
         if (move_distance == 0) {
@@ -141,7 +151,6 @@ void Mouse::Move(int x, int y, int center_x, int center_y) {
 
         // Average mouse movements
         last_mouse_change = (last_mouse_change * 0.91f) + (mouse_change * 0.09f);
-        last_motion_change = (last_motion_change * 0.69f) + (motion_change * 0.31f);
 
         const auto last_move_distance = last_mouse_change.Length();
 

src/input_common/drivers/virtual_amiibo.h

@@ -60,6 +60,6 @@ private:
     std::string file_path{};
     State state{State::Initialized};
     std::vector<u8> nfc_data;
-    Common::Input::PollingMode polling_mode{Common::Input::PollingMode::Pasive};
+    Common::Input::PollingMode polling_mode{Common::Input::PollingMode::Passive};
 };
 } // namespace InputCommon

src/input_common/helpers/joycon_driver.cpp

@@ -410,7 +410,7 @@ DriverResult JoyconDriver::SetIrsConfig(IrsMode mode_, IrsResolution format_) {
     return result;
 }
 
-DriverResult JoyconDriver::SetPasiveMode() {
+DriverResult JoyconDriver::SetPassiveMode() {
     std::scoped_lock lock{mutex};
     motion_enabled = false;
     hidbus_enabled = false;

src/input_common/helpers/joycon_driver.h

@@ -44,7 +44,7 @@ public:
     DriverResult SetVibration(const VibrationValue& vibration);
     DriverResult SetLedConfig(u8 led_pattern);
     DriverResult SetIrsConfig(IrsMode mode_, IrsResolution format_);
-    DriverResult SetPasiveMode();
+    DriverResult SetPassiveMode();
     DriverResult SetActiveMode();
     DriverResult SetIrMode();
     DriverResult SetNfcMode();

src/input_common/helpers/joycon_protocol/joycon_types.h

@@ -78,7 +78,7 @@ enum class PadButton : u32 {
     Capture = 0x200000,
 };
 
-enum class PasivePadButton : u32 {
+enum class PassivePadButton : u32 {
     Down_A = 0x0001,
     Right_X = 0x0002,
     Left_B = 0x0004,
@@ -95,7 +95,7 @@ enum class PasivePadButton : u32 {
     ZL_ZR = 0x8000,
 };
 
-enum class PasivePadStick : u8 {
+enum class PassivePadStick : u8 {
     Right = 0x00,
     RightDown = 0x01,
     Down = 0x02,

src/input_common/helpers/joycon_protocol/poller.cpp

@@ -48,13 +48,13 @@ void JoyconPoller::ReadPassiveMode(std::span<u8> buffer) {
 
     switch (device_type) {
     case ControllerType::Left:
-        UpdatePasiveLeftPadInput(data);
+        UpdatePassiveLeftPadInput(data);
         break;
     case ControllerType::Right:
-        UpdatePasiveRightPadInput(data);
+        UpdatePassiveRightPadInput(data);
         break;
     case ControllerType::Pro:
-        UpdatePasiveProPadInput(data);
+        UpdatePassiveProPadInput(data);
         break;
     default:
         break;
@@ -210,12 +210,12 @@ void JoyconPoller::UpdateActiveProPadInput(const InputReportActive& input,
     }
 }
 
-void JoyconPoller::UpdatePasiveLeftPadInput(const InputReportPassive& input) {
-    static constexpr std::array<PasivePadButton, 11> left_buttons{
-        PasivePadButton::Down_A,  PasivePadButton::Right_X, PasivePadButton::Left_B,
-        PasivePadButton::Up_Y,    PasivePadButton::SL,      PasivePadButton::SR,
-        PasivePadButton::L_R,     PasivePadButton::ZL_ZR,   PasivePadButton::Minus,
-        PasivePadButton::Capture, PasivePadButton::StickL,
+void JoyconPoller::UpdatePassiveLeftPadInput(const InputReportPassive& input) {
+    static constexpr std::array<PassivePadButton, 11> left_buttons{
+        PassivePadButton::Down_A,  PassivePadButton::Right_X, PassivePadButton::Left_B,
+        PassivePadButton::Up_Y,    PassivePadButton::SL,      PassivePadButton::SR,
+        PassivePadButton::L_R,     PassivePadButton::ZL_ZR,   PassivePadButton::Minus,
+        PassivePadButton::Capture, PassivePadButton::StickL,
     };
 
     for (auto left_button : left_buttons) {
@@ -225,17 +225,17 @@ void JoyconPoller::UpdatePasiveLeftPadInput(const InputReportPassive& input) {
     }
 
     const auto [left_axis_x, left_axis_y] =
-        GetPassiveAxisValue(static_cast<PasivePadStick>(input.stick_state));
+        GetPassiveAxisValue(static_cast<PassivePadStick>(input.stick_state));
     callbacks.on_stick_data(static_cast<int>(PadAxes::LeftStickX), left_axis_x);
     callbacks.on_stick_data(static_cast<int>(PadAxes::LeftStickY), left_axis_y);
 }
 
-void JoyconPoller::UpdatePasiveRightPadInput(const InputReportPassive& input) {
-    static constexpr std::array<PasivePadButton, 11> right_buttons{
-        PasivePadButton::Down_A, PasivePadButton::Right_X, PasivePadButton::Left_B,
-        PasivePadButton::Up_Y,   PasivePadButton::SL,      PasivePadButton::SR,
-        PasivePadButton::L_R,    PasivePadButton::ZL_ZR,   PasivePadButton::Plus,
-        PasivePadButton::Home,   PasivePadButton::StickR,
+void JoyconPoller::UpdatePassiveRightPadInput(const InputReportPassive& input) {
+    static constexpr std::array<PassivePadButton, 11> right_buttons{
+        PassivePadButton::Down_A, PassivePadButton::Right_X, PassivePadButton::Left_B,
+        PassivePadButton::Up_Y,   PassivePadButton::SL,      PassivePadButton::SR,
+        PassivePadButton::L_R,    PassivePadButton::ZL_ZR,   PassivePadButton::Plus,
+        PassivePadButton::Home,   PassivePadButton::StickR,
     };
 
     for (auto right_button : right_buttons) {
@@ -245,18 +245,18 @@ void JoyconPoller::UpdatePasiveRightPadInput(const InputReportPassive& input) {
     }
 
     const auto [right_axis_x, right_axis_y] =
-        GetPassiveAxisValue(static_cast<PasivePadStick>(input.stick_state));
+        GetPassiveAxisValue(static_cast<PassivePadStick>(input.stick_state));
     callbacks.on_stick_data(static_cast<int>(PadAxes::RightStickX), right_axis_x);
     callbacks.on_stick_data(static_cast<int>(PadAxes::RightStickY), right_axis_y);
 }
 
-void JoyconPoller::UpdatePasiveProPadInput(const InputReportPassive& input) {
-    static constexpr std::array<PasivePadButton, 14> pro_buttons{
-        PasivePadButton::Down_A, PasivePadButton::Right_X, PasivePadButton::Left_B,
-        PasivePadButton::Up_Y,   PasivePadButton::SL,      PasivePadButton::SR,
-        PasivePadButton::L_R,    PasivePadButton::ZL_ZR,   PasivePadButton::Minus,
-        PasivePadButton::Plus,   PasivePadButton::Capture, PasivePadButton::Home,
-        PasivePadButton::StickL, PasivePadButton::StickR,
+void JoyconPoller::UpdatePassiveProPadInput(const InputReportPassive& input) {
+    static constexpr std::array<PassivePadButton, 14> pro_buttons{
+        PassivePadButton::Down_A, PassivePadButton::Right_X, PassivePadButton::Left_B,
+        PassivePadButton::Up_Y,   PassivePadButton::SL,      PassivePadButton::SR,
+        PassivePadButton::L_R,    PassivePadButton::ZL_ZR,   PassivePadButton::Minus,
+        PassivePadButton::Plus,   PassivePadButton::Capture, PassivePadButton::Home,
+        PassivePadButton::StickL, PassivePadButton::StickR,
     };
 
     for (auto pro_button : pro_buttons) {
@@ -266,9 +266,9 @@ void JoyconPoller::UpdatePasiveProPadInput(const InputReportPassive& input) {
     }
 
     const auto [left_axis_x, left_axis_y] =
-        GetPassiveAxisValue(static_cast<PasivePadStick>(input.stick_state && 0xf));
+        GetPassiveAxisValue(static_cast<PassivePadStick>(input.stick_state & 0xf));
     const auto [right_axis_x, right_axis_y] =
-        GetPassiveAxisValue(static_cast<PasivePadStick>(input.stick_state >> 4));
+        GetPassiveAxisValue(static_cast<PassivePadStick>(input.stick_state >> 4));
     callbacks.on_stick_data(static_cast<int>(PadAxes::LeftStickX), left_axis_x);
     callbacks.on_stick_data(static_cast<int>(PadAxes::LeftStickY), left_axis_y);
     callbacks.on_stick_data(static_cast<int>(PadAxes::RightStickX), right_axis_x);
@@ -283,25 +283,25 @@ f32 JoyconPoller::GetAxisValue(u16 raw_value, Joycon::JoyStickAxisCalibration ca
     return value / calibration.min;
 }
 
-std::pair<f32, f32> JoyconPoller::GetPassiveAxisValue(PasivePadStick raw_value) const {
+std::pair<f32, f32> JoyconPoller::GetPassiveAxisValue(PassivePadStick raw_value) const {
     switch (raw_value) {
-    case PasivePadStick::Right:
+    case PassivePadStick::Right:
         return {1.0f, 0.0f};
-    case PasivePadStick::RightDown:
+    case PassivePadStick::RightDown:
         return {1.0f, -1.0f};
-    case PasivePadStick::Down:
+    case PassivePadStick::Down:
         return {0.0f, -1.0f};
-    case PasivePadStick::DownLeft:
+    case PassivePadStick::DownLeft:
         return {-1.0f, -1.0f};
-    case PasivePadStick::Left:
+    case PassivePadStick::Left:
         return {-1.0f, 0.0f};
-    case PasivePadStick::LeftUp:
+    case PassivePadStick::LeftUp:
         return {-1.0f, 1.0f};
-    case PasivePadStick::Up:
+    case PassivePadStick::Up:
         return {0.0f, 1.0f};
-    case PasivePadStick::UpRight:
+    case PassivePadStick::UpRight:
         return {1.0f, 1.0f};
-    case PasivePadStick::Neutral:
+    case PassivePadStick::Neutral:
     default:
         return {0.0f, 0.0f};
     }

src/input_common/helpers/joycon_protocol/poller.h

@@ -46,15 +46,15 @@ private:
                                    const MotionStatus& motion_status);
     void UpdateActiveProPadInput(const InputReportActive& input, const MotionStatus& motion_status);
 
-    void UpdatePasiveLeftPadInput(const InputReportPassive& buffer);
-    void UpdatePasiveRightPadInput(const InputReportPassive& buffer);
-    void UpdatePasiveProPadInput(const InputReportPassive& buffer);
+    void UpdatePassiveLeftPadInput(const InputReportPassive& buffer);
+    void UpdatePassiveRightPadInput(const InputReportPassive& buffer);
+    void UpdatePassiveProPadInput(const InputReportPassive& buffer);
 
     /// Returns a calibrated joystick axis from raw axis data
     f32 GetAxisValue(u16 raw_value, JoyStickAxisCalibration calibration) const;
 
     /// Returns a digital joystick axis from passive axis data
-    std::pair<f32, f32> GetPassiveAxisValue(PasivePadStick raw_value) const;
+    std::pair<f32, f32> GetPassiveAxisValue(PassivePadStick raw_value) const;
 
     /// Returns a calibrated accelerometer axis from raw motion data
     f32 GetAccelerometerValue(s16 raw, const MotionSensorCalibration& cal,

src/input_common/input_mapping.cpp

@@ -146,6 +146,7 @@ void MappingFactory::RegisterMotion(const MappingData& data) {
     if (data.engine == "mouse") {
         new_input.Set("motion", 0);
         new_input.Set("pad", 1);
+        new_input.Set("threshold", 0.001f);
         input_queue.Push(new_input);
         return;
     }

src/shader_recompiler/ir_opt/global_memory_to_storage_buffer_pass.cpp

@@ -35,6 +35,7 @@ struct Bias {
     u32 index;
     u32 offset_begin;
     u32 offset_end;
+    u32 alignment;
 };
 
 using boost::container::flat_set;
@@ -349,7 +350,8 @@ std::optional<StorageBufferAddr> Track(const IR::Value& value, const Bias* bias)
             .index = index.U32(),
             .offset = offset.U32(),
         };
-        if (!Common::IsAligned(storage_buffer.offset, 16)) {
+        const u32 alignment{bias ? bias->alignment : 8U};
+        if (!Common::IsAligned(storage_buffer.offset, alignment)) {
             // The SSBO pointer has to be aligned
             return std::nullopt;
         }
@@ -371,6 +373,7 @@ void CollectStorageBuffers(IR::Block& block, IR::Inst& inst, StorageInfo& info)
         .index = 0,
         .offset_begin = 0x110,
         .offset_end = 0x610,
+        .alignment = 16,
     };
     // Track the low address of the instruction
     const std::optional<LowAddrInfo> low_addr_info{TrackLowAddress(&inst)};
@@ -386,8 +389,11 @@ void CollectStorageBuffers(IR::Block& block, IR::Inst& inst, StorageInfo& info)
         storage_buffer = Track(low_addr, nullptr);
         if (!storage_buffer) {
             // If that also fails, use NVN fallbacks
+            LOG_WARNING(Shader, "Storage buffer failed to track, using global memory fallbacks");
             return;
         }
+        LOG_WARNING(Shader, "Storage buffer tracked without bias, index {} offset {}",
+                    storage_buffer->index, storage_buffer->offset);
     }
     // Collect storage buffer and the instruction
     if (IsGlobalMemoryWrite(inst)) {

src/video_core/buffer_cache/buffer_cache.h

@@ -383,7 +383,8 @@ private:
 
     void NotifyBufferDeletion();
 
-    [[nodiscard]] Binding StorageBufferBinding(GPUVAddr ssbo_addr, bool is_written = false) const;
+    [[nodiscard]] Binding StorageBufferBinding(GPUVAddr ssbo_addr, u32 cbuf_index,
+                                               bool is_written = false) const;
 
     [[nodiscard]] TextureBufferBinding GetTextureBufferBinding(GPUVAddr gpu_addr, u32 size,
                                                                PixelFormat format);
@@ -802,7 +803,7 @@ void BufferCache<P>::BindGraphicsStorageBuffer(size_t stage, size_t ssbo_index,
 
     const auto& cbufs = maxwell3d->state.shader_stages[stage];
     const GPUVAddr ssbo_addr = cbufs.const_buffers[cbuf_index].address + cbuf_offset;
-    storage_buffers[stage][ssbo_index] = StorageBufferBinding(ssbo_addr, is_written);
+    storage_buffers[stage][ssbo_index] = StorageBufferBinding(ssbo_addr, cbuf_index, is_written);
 }
 
 template <class P>
@@ -842,7 +843,7 @@ void BufferCache<P>::BindComputeStorageBuffer(size_t ssbo_index, u32 cbuf_index,
 
     const auto& cbufs = launch_desc.const_buffer_config;
     const GPUVAddr ssbo_addr = cbufs[cbuf_index].Address() + cbuf_offset;
-    compute_storage_buffers[ssbo_index] = StorageBufferBinding(ssbo_addr, is_written);
+    compute_storage_buffers[ssbo_index] = StorageBufferBinding(ssbo_addr, cbuf_index, is_written);
 }
 
 template <class P>
@@ -1988,11 +1989,26 @@ void BufferCache<P>::NotifyBufferDeletion() {
 
 template <class P>
 typename BufferCache<P>::Binding BufferCache<P>::StorageBufferBinding(GPUVAddr ssbo_addr,
+                                                                      u32 cbuf_index,
                                                                       bool is_written) const {
     const GPUVAddr gpu_addr = gpu_memory->Read<u64>(ssbo_addr);
-    const u32 size = gpu_memory->Read<u32>(ssbo_addr + 8);
+    const auto size = [&]() {
+        const bool is_nvn_cbuf = cbuf_index == 0;
+        // The NVN driver buffer (index 0) is known to pack the SSBO address followed by its size.
+        if (is_nvn_cbuf) {
+            return gpu_memory->Read<u32>(ssbo_addr + 8);
+        }
+        // Other titles (notably Doom Eternal) may use STG/LDG on buffer addresses in custom defined
+        // cbufs, which do not store the sizes adjacent to the addresses, so use the fully
+        // mapped buffer size for now.
+        const u32 memory_layout_size = static_cast<u32>(gpu_memory->GetMemoryLayoutSize(gpu_addr));
+        LOG_INFO(HW_GPU, "Binding storage buffer for cbuf index {}, MemoryLayoutSize 0x{:X}",
+                 cbuf_index, memory_layout_size);
+        return memory_layout_size;
+    }();
     const std::optional<VAddr> cpu_addr = gpu_memory->GpuToCpuAddress(gpu_addr);
     if (!cpu_addr || size == 0) {
+        LOG_WARNING(HW_GPU, "Failed to find storage buffer for cbuf index {}", cbuf_index);
         return NULL_BINDING;
     }
     const VAddr cpu_end = Common::AlignUp(*cpu_addr + size, Core::Memory::YUZU_PAGESIZE);

src/video_core/gpu.cpp

@@ -197,7 +197,7 @@ struct GPU::Impl {
         constexpr u64 gpu_ticks_num = 384;
         constexpr u64 gpu_ticks_den = 625;
 
-        u64 nanoseconds = system.CoreTiming().GetGlobalTimeNs().count();
+        u64 nanoseconds = system.CoreTiming().GetCPUTimeNs().count();
         if (Settings::values.use_fast_gpu_time.GetValue()) {
             nanoseconds /= 256;
         }

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -112,13 +112,17 @@ GLenum ImageTarget(Shader::TextureType type, int num_samples = 1) {
     return GL_NONE;
 }
 
-GLenum TextureMode(PixelFormat format, bool is_first) {
+GLenum TextureMode(PixelFormat format, std::array<SwizzleSource, 4> swizzle) {
+    bool any_r =
+        std::ranges::any_of(swizzle, [](SwizzleSource s) { return s == SwizzleSource::R; });
     switch (format) {
     case PixelFormat::D24_UNORM_S8_UINT:
     case PixelFormat::D32_FLOAT_S8_UINT:
-        return is_first ? GL_DEPTH_COMPONENT : GL_STENCIL_INDEX;
+        // R = depth, G = stencil
+        return any_r ? GL_DEPTH_COMPONENT : GL_STENCIL_INDEX;
     case PixelFormat::S8_UINT_D24_UNORM:
-        return is_first ? GL_STENCIL_INDEX : GL_DEPTH_COMPONENT;
+        // R = stencil, G = depth
+        return any_r ? GL_STENCIL_INDEX : GL_DEPTH_COMPONENT;
     default:
         ASSERT(false);
         return GL_DEPTH_COMPONENT;
@@ -208,8 +212,7 @@ void ApplySwizzle(GLuint handle, PixelFormat format, std::array<SwizzleSource, 4
     case PixelFormat::D32_FLOAT_S8_UINT:
     case PixelFormat::S8_UINT_D24_UNORM:
         UNIMPLEMENTED_IF(swizzle[0] != SwizzleSource::R && swizzle[0] != SwizzleSource::G);
-        glTextureParameteri(handle, GL_DEPTH_STENCIL_TEXTURE_MODE,
-                            TextureMode(format, swizzle[0] == SwizzleSource::R));
+        glTextureParameteri(handle, GL_DEPTH_STENCIL_TEXTURE_MODE, TextureMode(format, swizzle));
         std::ranges::transform(swizzle, swizzle.begin(), ConvertGreenRed);
         break;
     case PixelFormat::A5B5G5R1_UNORM: {

src/video_core/renderer_vulkan/vk_buffer_cache.cpp

@@ -238,7 +238,7 @@ private:
         return indices;
     }
 
-    void MakeAndUpdateIndices(u8* staging_data, size_t quad_size, u32 quad, u32 first) {
+    void MakeAndUpdateIndices(u8* staging_data, size_t quad_size, u32 quad, u32 first) override {
         switch (index_type) {
         case VK_INDEX_TYPE_UINT8_EXT:
             std::memcpy(staging_data, MakeIndices<u8>(quad, first).data(), quad_size);
@@ -278,7 +278,7 @@ private:
         return indices;
     }
 
-    void MakeAndUpdateIndices(u8* staging_data, size_t quad_size, u32 quad, u32 first) {
+    void MakeAndUpdateIndices(u8* staging_data, size_t quad_size, u32 quad, u32 first) override {
         switch (index_type) {
         case VK_INDEX_TYPE_UINT8_EXT:
             std::memcpy(staging_data, MakeIndices<u8>(quad, first).data(), quad_size);

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -1294,7 +1294,7 @@ void RasterizerVulkan::UpdateDepthBoundsTestEnable(Tegra::Engines::Maxwell3D::Re
         LOG_WARNING(Render_Vulkan, "Depth bounds is enabled but not supported");
         enabled = false;
     }
-    scheduler.Record([enable = regs.depth_bounds_enable](vk::CommandBuffer cmdbuf) {
+    scheduler.Record([enable = enabled](vk::CommandBuffer cmdbuf) {
         cmdbuf.SetDepthBoundsTestEnableEXT(enable);
     });
 }

src/video_core/renderer_vulkan/vk_texture_cache.cpp

@@ -189,13 +189,16 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
     if (info.IsRenderTarget()) {
         return ImageAspectMask(info.format);
     }
-    const bool is_first = info.Swizzle()[0] == SwizzleSource::R;
+    bool any_r =
+        std::ranges::any_of(info.Swizzle(), [](SwizzleSource s) { return s == SwizzleSource::R; });
     switch (info.format) {
     case PixelFormat::D24_UNORM_S8_UINT:
     case PixelFormat::D32_FLOAT_S8_UINT:
-        return is_first ? VK_IMAGE_ASPECT_DEPTH_BIT : VK_IMAGE_ASPECT_STENCIL_BIT;
+        // R = depth, G = stencil
+        return any_r ? VK_IMAGE_ASPECT_DEPTH_BIT : VK_IMAGE_ASPECT_STENCIL_BIT;
     case PixelFormat::S8_UINT_D24_UNORM:
-        return is_first ? VK_IMAGE_ASPECT_STENCIL_BIT : VK_IMAGE_ASPECT_DEPTH_BIT;
+        // R = stencil, G = depth
+        return any_r ? VK_IMAGE_ASPECT_STENCIL_BIT : VK_IMAGE_ASPECT_DEPTH_BIT;
     case PixelFormat::D16_UNORM:
     case PixelFormat::D32_FLOAT:
         return VK_IMAGE_ASPECT_DEPTH_BIT;
@@ -1769,7 +1772,7 @@ Sampler::Sampler(TextureCacheRuntime& runtime, const Tegra::Texture::TSCEntry& t
         .minLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.0f : tsc.MinLod(),
         .maxLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.25f : tsc.MaxLod(),
         .borderColor =
-            arbitrary_borders ? VK_BORDER_COLOR_INT_CUSTOM_EXT : ConvertBorderColor(color),
+            arbitrary_borders ? VK_BORDER_COLOR_FLOAT_CUSTOM_EXT : ConvertBorderColor(color),
         .unnormalizedCoordinates = VK_FALSE,
     });
 }

src/video_core/texture_cache/texture_cache.h

@@ -1244,6 +1244,7 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA
         auto copies = MakeShrinkImageCopies(new_info, overlap.info, base, up_scale, down_shift);
         if (overlap.info.num_samples != new_image.info.num_samples) {
             runtime.CopyImageMSAA(new_image, overlap, std::move(copies));
+            continue;
         } else {
             runtime.CopyImage(new_image, overlap, std::move(copies));
         }

src/yuzu/main.cpp

@@ -91,6 +91,9 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual
 #include "common/microprofile.h"
 #include "common/scm_rev.h"
 #include "common/scope_exit.h"
+#ifdef _WIN32
+#include "common/windows/timer_resolution.h"
+#endif
 #ifdef ARCHITECTURE_x86_64
 #include "common/x64/cpu_detect.h"
 #endif
@@ -377,6 +380,12 @@ GMainWindow::GMainWindow(std::unique_ptr<Config> config_, bool has_broken_vulkan
     LOG_INFO(Frontend, "Host RAM: {:.2f} GiB",
              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());
+#endif
     UpdateWindowTitle();
 
     show();

src/yuzu_cmd/yuzu.cpp

@@ -42,6 +42,8 @@
 #include <windows.h>
 
 #include <shellapi.h>
+
+#include "common/windows/timer_resolution.h"
 #endif
 
 #undef _UNICODE
@@ -314,6 +316,8 @@ int main(int argc, char** argv) {
 
 #ifdef _WIN32
     LocalFree(argv_w);
+
+    Common::Windows::SetCurrentTimerResolutionToMaximum();
 #endif
 
     MicroProfileOnThreadCreate("EmuThread");