Update README.md

Merge pull request #3448 from bunnei/fix-audio-interp-2
audio_core: interpolate: Improvements to fix audio crackling.
2020-03-01 18:03:32 +01:00 · 2020-02-28 16:07:10 -05:00 · 2020-02-28 01:22:00 -05:00 · 2020-02-28 01:13:38 -05:00 · 2020-02-27 19:51:55 -05:00 · 2020-02-27 19:50:17 -05:00
159 changed files with 5343 additions and 2193 deletions
@@ -151,15 +151,22 @@ if (ENABLE_SDL2)
        set(SDL2_INCLUDE_DIR "${SDL2_PREFIX}/include" CACHE PATH "Path to SDL2 headers")
        set(SDL2_LIBRARY "${SDL2_PREFIX}/lib/x64/SDL2.lib" CACHE PATH "Path to SDL2 library")
        set(SDL2_DLL_DIR "${SDL2_PREFIX}/lib/x64/" CACHE PATH "Path to SDL2.dll")
-    else()
-        find_package(SDL2 REQUIRED)
-    endif()

-    if (SDL2_FOUND)
-        # TODO(yuriks): Make FindSDL2.cmake export an IMPORTED library instead
        add_library(SDL2 INTERFACE)
        target_link_libraries(SDL2 INTERFACE "${SDL2_LIBRARY}")
        target_include_directories(SDL2 INTERFACE "${SDL2_INCLUDE_DIR}")
+    else()
+        find_package(SDL2 REQUIRED)
+
+        # Some installations don't set SDL2_LIBRARIES
+        if("${SDL2_LIBRARIES}" STREQUAL "")
+            message(WARNING "SDL2_LIBRARIES wasn't set, manually setting to SDL2::SDL2")
+            set(SDL2_LIBRARIES "SDL2::SDL2")
+        endif()
+
+        include_directories(${SDL2_INCLUDE_DIRS})
+        add_library(SDL2 INTERFACE)
+        target_link_libraries(SDL2 INTERFACE "${SDL2_LIBRARIES}")
    endif()
 else()
    set(SDL2_FOUND NO)
@@ -1,6 +1,6 @@
 yuzu emulator
 =============
-[![Travis CI Build Status](https://travis-ci.org/yuzu-emu/yuzu.svg?branch=master)](https://travis-ci.org/yuzu-emu/yuzu)
+[![Travis CI Build Status](https://travis-ci.com/yuzu-emu/yuzu.svg?branch=master)](https://travis-ci.com/yuzu-emu/yuzu)
 [![Azure Mainline CI Build Status](https://dev.azure.com/yuzu-emu/yuzu/_apis/build/status/yuzu%20mainline?branchName=master)](https://dev.azure.com/yuzu-emu/yuzu/)

 yuzu is an experimental open-source emulator for the Nintendo Switch from the creators of [Citra](https://citra-emu.org/).
@@ -21,7 +21,7 @@ For development discussion, please join us on [Discord](https://discord.gg/XQV6d

 Most of the development happens on GitHub. It's also where [our central repository](https://github.com/yuzu-emu/yuzu) is hosted.

-If you want to contribute please take a look at the [Contributor's Guide](CONTRIBUTING.md) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You should as well contact any of the developers on Discord in order to know about the current state of the emulator.
+If you want to contribute please take a look at the [Contributor's Guide](CONTRIBUTING.md) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You should also contact any of the developers on Discord in order to know about the current state of the emulator.

 ### Building

@@ -1,239 +0,0 @@
-
-# This module defines
-# SDL2_LIBRARY, the name of the library to link against
-# SDL2_FOUND, if false, do not try to link to SDL2
-# SDL2_INCLUDE_DIR, where to find SDL.h
-# SDL2_DLL_DIR, where to find SDL2.dll if it exists
-#
-# This module responds to the the flag:
-# SDL2_BUILDING_LIBRARY
-# If this is defined, then no SDL2main will be linked in because
-# only applications need main().
-# Otherwise, it is assumed you are building an application and this
-# module will attempt to locate and set the the proper link flags
-# as part of the returned SDL2_LIBRARY variable.
-#
-# Don't forget to include SDLmain.h and SDLmain.m your project for the
-# OS X framework based version. (Other versions link to -lSDL2main which
-# this module will try to find on your behalf.) Also for OS X, this
-# module will automatically add the -framework Cocoa on your behalf.
-#
-#
-# Additional Note: If you see an empty SDL2_LIBRARY_TEMP in your configuration
-# and no SDL2_LIBRARY, it means CMake did not find your SDL2 library
-# (SDL2.dll, libsdl2.so, SDL2.framework, etc).
-# Set SDL2_LIBRARY_TEMP to point to your SDL2 library, and configure again.
-# Similarly, if you see an empty SDL2MAIN_LIBRARY, you should set this value
-# as appropriate. These values are used to generate the final SDL2_LIBRARY
-# variable, but when these values are unset, SDL2_LIBRARY does not get created.
-#
-#
-# $SDL2DIR is an environment variable that would
-# correspond to the ./configure --prefix=$SDL2DIR
-# used in building SDL2.
-# l.e.galup  9-20-02
-#
-# Modified by Eric Wing.
-# Added code to assist with automated building by using environmental variables
-# and providing a more controlled/consistent search behavior.
-# Added new modifications to recognize OS X frameworks and
-# additional Unix paths (FreeBSD, etc).
-# Also corrected the header search path to follow "proper" SDL guidelines.
-# Added a search for SDL2main which is needed by some platforms.
-# Added a search for threads which is needed by some platforms.
-# Added needed compile switches for MinGW.
-#
-# On OSX, this will prefer the Framework version (if found) over others.
-# People will have to manually change the cache values of
-# SDL2_LIBRARY to override this selection or set the CMake environment
-# CMAKE_INCLUDE_PATH to modify the search paths.
-#
-# Note that the header path has changed from SDL2/SDL.h to just SDL.h
-# This needed to change because "proper" SDL convention
-# is #include "SDL.h", not <SDL2/SDL.h>. This is done for portability
-# reasons because not all systems place things in SDL2/ (see FreeBSD).
-
-#=============================================================================
-# Copyright 2003-2009 Kitware, Inc.
-#
-# Distributed under the OSI-approved BSD License (the "License").
-#
-# This software is distributed WITHOUT ANY WARRANTY; without even the
-# implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
-# See the License for more information.
-#=============================================================================
-# CMake - Cross Platform Makefile Generator
-# Copyright 2000-2016 Kitware, Inc.
-# Copyright 2000-2011 Insight Software Consortium
-# All rights reserved.
-# 
-# Redistribution and use in source and binary forms, with or without
-# modification, are permitted provided that the following conditions
-# are met:
-# 
-# * Redistributions of source code must retain the above copyright
-#   notice, this list of conditions and the following disclaimer.
-# 
-# * Redistributions in binary form must reproduce the above copyright
-#   notice, this list of conditions and the following disclaimer in the
-#   documentation and/or other materials provided with the distribution.
-# 
-# * Neither the names of Kitware, Inc., the Insight Software Consortium,
-#   nor the names of their contributors may be used to endorse or promote
-#   products derived from this software without specific prior written
-#   permission.
-# 
-# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-# HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-# 
-# ------------------------------------------------------------------------------
-# 
-# The above copyright and license notice applies to distributions of
-# CMake in source and binary form.  Some source files contain additional
-# notices of original copyright by their contributors; see each source
-# for details.  Third-party software packages supplied with CMake under
-# compatible licenses provide their own copyright notices documented in
-# corresponding subdirectories.
-# 
-# ------------------------------------------------------------------------------
-# 
-# CMake was initially developed by Kitware with the following sponsorship:
-# 
-#  * National Library of Medicine at the National Institutes of Health
-#    as part of the Insight Segmentation and Registration Toolkit (ITK).
-# 
-#  * US National Labs (Los Alamos, Livermore, Sandia) ASC Parallel
-#    Visualization Initiative.
-# 
-#  * National Alliance for Medical Image Computing (NAMIC) is funded by the
-#    National Institutes of Health through the NIH Roadmap for Medical Research,
-#    Grant U54 EB005149.
-# 
-#  * Kitware, Inc.
-#
-
-message("<FindSDL2.cmake>")
-
-SET(SDL2_SEARCH_PATHS
-    ~/Library/Frameworks
-    /Library/Frameworks
-    /usr/local
-    /usr
-    /sw # Fink
-    /opt/local # DarwinPorts
-    /opt/csw # Blastwave
-    /opt
-    ${SDL2_PATH}
-)
-
-if(CMAKE_SIZEOF_VOID_P EQUAL 8)
-    set(VC_LIB_PATH_SUFFIX lib/x64)
-else()
-    set(VC_LIB_PATH_SUFFIX lib/x86)
-endif()
-
-FIND_LIBRARY(SDL2_LIBRARY_TEMP
-    NAMES SDL2
-    HINTS
-    $ENV{SDL2DIR}
-    PATH_SUFFIXES lib64 lib ${VC_LIB_PATH_SUFFIX}
-    PATHS ${SDL2_SEARCH_PATHS}
-)
-
-IF(SDL2_LIBRARY_TEMP)
-    if(MSVC)
-        get_filename_component(SDL2_DLL_DIR_TEMP ${SDL2_LIBRARY_TEMP} DIRECTORY)
-        if(EXISTS ${SDL2_DLL_DIR_TEMP}/SDL2.dll)
-            set(SDL2_DLL_DIR ${SDL2_DLL_DIR_TEMP})
-            unset(SDL2_DLL_DIR_TEMP)
-        endif()
-    endif()
-
-    FIND_PATH(SDL2_INCLUDE_DIR SDL.h
-        HINTS
-        $ENV{SDL2DIR}
-        PATH_SUFFIXES include/SDL2 include
-        PATHS ${SDL2_SEARCH_PATHS}
-    )
-
-    IF(NOT SDL2_BUILDING_LIBRARY)
-        IF(NOT ${SDL2_INCLUDE_DIR} MATCHES ".framework")
-            # Non-OS X framework versions expect you to also dynamically link to
-            # SDL2main. This is mainly for Windows and OS X. Other (Unix) platforms
-            # seem to provide SDL2main for compatibility even though they don't
-            # necessarily need it.
-            FIND_LIBRARY(SDL2MAIN_LIBRARY
-                NAMES SDL2main
-                HINTS
-                $ENV{SDL2DIR}
-                PATH_SUFFIXES lib64 lib
-                PATHS ${SDL2_SEARCH_PATHS}
-            )
-        ENDIF(NOT ${SDL2_INCLUDE_DIR} MATCHES ".framework")
-    ENDIF(NOT SDL2_BUILDING_LIBRARY)
-
-    # SDL2 may require threads on your system.
-    # The Apple build may not need an explicit flag because one of the
-    # frameworks may already provide it.
-    # But for non-OSX systems, I will use the CMake Threads package.
-    IF(NOT APPLE)
-        FIND_PACKAGE(Threads)
-    ENDIF(NOT APPLE)
-
-    # MinGW needs an additional library, mwindows
-    # It's total link flags should look like -lmingw32 -lSDL2main -lSDL2 -lmwindows
-    # (Actually on second look, I think it only needs one of the m* libraries.)
-    IF(MINGW)
-        SET(MINGW32_LIBRARY mingw32 CACHE STRING "mwindows for MinGW")
-    ENDIF(MINGW)
-
-    # For SDL2main
-    IF(NOT SDL2_BUILDING_LIBRARY)
-        IF(SDL2MAIN_LIBRARY)
-            SET(SDL2_LIBRARY_TEMP ${SDL2MAIN_LIBRARY} ${SDL2_LIBRARY_TEMP})
-        ENDIF(SDL2MAIN_LIBRARY)
-    ENDIF(NOT SDL2_BUILDING_LIBRARY)
-
-    # For OS X, SDL2 uses Cocoa as a backend so it must link to Cocoa.
-    # CMake doesn't display the -framework Cocoa string in the UI even
-    # though it actually is there if I modify a pre-used variable.
-    # I think it has something to do with the CACHE STRING.
-    # So I use a temporary variable until the end so I can set the
-    # "real" variable in one-shot.
-    IF(APPLE)
-        SET(SDL2_LIBRARY_TEMP ${SDL2_LIBRARY_TEMP} "-framework Cocoa")
-    ENDIF(APPLE)
-
-    # For threads, as mentioned Apple doesn't need this.
-    # In fact, there seems to be a problem if I used the Threads package
-    # and try using this line, so I'm just skipping it entirely for OS X.
-    IF(NOT APPLE)
-        SET(SDL2_LIBRARY_TEMP ${SDL2_LIBRARY_TEMP} ${CMAKE_THREAD_LIBS_INIT})
-    ENDIF(NOT APPLE)
-
-    # For MinGW library
-    IF(MINGW)
-        SET(SDL2_LIBRARY_TEMP ${MINGW32_LIBRARY} ${SDL2_LIBRARY_TEMP})
-    ENDIF(MINGW)
-
-    # Set the final string here so the GUI reflects the final state.
-    SET(SDL2_LIBRARY ${SDL2_LIBRARY_TEMP} CACHE STRING "Where the SDL2 Library can be found")
-
-    # Unset the temp variable to INTERNAL so it is not seen in the CMake GUI
-    UNSET(SDL2_LIBRARY_TEMP)
-ENDIF(SDL2_LIBRARY_TEMP)
-
-message("</FindSDL2.cmake>")
-
-INCLUDE(FindPackageHandleStandardArgs)
-
-FIND_PACKAGE_HANDLE_STANDARD_ARGS(SDL2 REQUIRED_VARS SDL2_LIBRARY SDL2_INCLUDE_DIR)
@@ -1,4 +1,4 @@
-From https://github.com/yhirose/cpp-httplib/commit/d9479bc0b12e8a1e8bce2d34da4feeef488581f3
+From https://github.com/yhirose/cpp-httplib/tree/fce8e6fefdab4ad48bc5b25c98e5ebfda4f3cf53

 MIT License

@@ -5,21 +5,141 @@
 #define _USE_MATH_DEFINES

 #include <algorithm>
+#include <climits>
 #include <cmath>
 #include <vector>
+
 #include "audio_core/algorithm/interpolate.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"

 namespace AudioCore {

-/// The Lanczos kernel
-static double Lanczos(std::size_t a, double x) {
-    if (x == 0.0)
-        return 1.0;
-    const double px = M_PI * x;
-    return a * std::sin(px) * std::sin(px / a) / (px * px);
-}
+constexpr std::array<s16, 512> curve_lut0{
+    6600,  19426, 6722,  3,     6479,  19424, 6845,  9,     6359,  19419, 6968,  15,    6239,
+    19412, 7093,  22,    6121,  19403, 7219,  28,    6004,  19391, 7345,  34,    5888,  19377,
+    7472,  41,    5773,  19361, 7600,  48,    5659,  19342, 7728,  55,    5546,  19321, 7857,
+    62,    5434,  19298, 7987,  69,    5323,  19273, 8118,  77,    5213,  19245, 8249,  84,
+    5104,  19215, 8381,  92,    4997,  19183, 8513,  101,   4890,  19148, 8646,  109,   4785,
+    19112, 8780,  118,   4681,  19073, 8914,  127,   4579,  19031, 9048,  137,   4477,  18988,
+    9183,  147,   4377,  18942, 9318,  157,   4277,  18895, 9454,  168,   4179,  18845, 9590,
+    179,   4083,  18793, 9726,  190,   3987,  18738, 9863,  202,   3893,  18682, 10000, 215,
+    3800,  18624, 10137, 228,   3709,  18563, 10274, 241,   3618,  18500, 10411, 255,   3529,
+    18436, 10549, 270,   3441,  18369, 10687, 285,   3355,  18300, 10824, 300,   3269,  18230,
+    10962, 317,   3186,  18157, 11100, 334,   3103,  18082, 11238, 351,   3022,  18006, 11375,
+    369,   2942,  17927, 11513, 388,   2863,  17847, 11650, 408,   2785,  17765, 11788, 428,
+    2709,  17681, 11925, 449,   2635,  17595, 12062, 471,   2561,  17507, 12198, 494,   2489,
+    17418, 12334, 517,   2418,  17327, 12470, 541,   2348,  17234, 12606, 566,   2280,  17140,
+    12741, 592,   2213,  17044, 12876, 619,   2147,  16946, 13010, 647,   2083,  16846, 13144,
+    675,   2020,  16745, 13277, 704,   1958,  16643, 13409, 735,   1897,  16539, 13541, 766,
+    1838,  16434, 13673, 798,   1780,  16327, 13803, 832,   1723,  16218, 13933, 866,   1667,
+    16109, 14062, 901,   1613,  15998, 14191, 937,   1560,  15885, 14318, 975,   1508,  15772,
+    14445, 1013,  1457,  15657, 14571, 1052,  1407,  15540, 14695, 1093,  1359,  15423, 14819,
+    1134,  1312,  15304, 14942, 1177,  1266,  15185, 15064, 1221,  1221,  15064, 15185, 1266,
+    1177,  14942, 15304, 1312,  1134,  14819, 15423, 1359,  1093,  14695, 15540, 1407,  1052,
+    14571, 15657, 1457,  1013,  14445, 15772, 1508,  975,   14318, 15885, 1560,  937,   14191,
+    15998, 1613,  901,   14062, 16109, 1667,  866,   13933, 16218, 1723,  832,   13803, 16327,
+    1780,  798,   13673, 16434, 1838,  766,   13541, 16539, 1897,  735,   13409, 16643, 1958,
+    704,   13277, 16745, 2020,  675,   13144, 16846, 2083,  647,   13010, 16946, 2147,  619,
+    12876, 17044, 2213,  592,   12741, 17140, 2280,  566,   12606, 17234, 2348,  541,   12470,
+    17327, 2418,  517,   12334, 17418, 2489,  494,   12198, 17507, 2561,  471,   12062, 17595,
+    2635,  449,   11925, 17681, 2709,  428,   11788, 17765, 2785,  408,   11650, 17847, 2863,
+    388,   11513, 17927, 2942,  369,   11375, 18006, 3022,  351,   11238, 18082, 3103,  334,
+    11100, 18157, 3186,  317,   10962, 18230, 3269,  300,   10824, 18300, 3355,  285,   10687,
+    18369, 3441,  270,   10549, 18436, 3529,  255,   10411, 18500, 3618,  241,   10274, 18563,
+    3709,  228,   10137, 18624, 3800,  215,   10000, 18682, 3893,  202,   9863,  18738, 3987,
+    190,   9726,  18793, 4083,  179,   9590,  18845, 4179,  168,   9454,  18895, 4277,  157,
+    9318,  18942, 4377,  147,   9183,  18988, 4477,  137,   9048,  19031, 4579,  127,   8914,
+    19073, 4681,  118,   8780,  19112, 4785,  109,   8646,  19148, 4890,  101,   8513,  19183,
+    4997,  92,    8381,  19215, 5104,  84,    8249,  19245, 5213,  77,    8118,  19273, 5323,
+    69,    7987,  19298, 5434,  62,    7857,  19321, 5546,  55,    7728,  19342, 5659,  48,
+    7600,  19361, 5773,  41,    7472,  19377, 5888,  34,    7345,  19391, 6004,  28,    7219,
+    19403, 6121,  22,    7093,  19412, 6239,  15,    6968,  19419, 6359,  9,     6845,  19424,
+    6479,  3,     6722,  19426, 6600};
+
+constexpr std::array<s16, 512> curve_lut1{
+    -68,   32639, 69,    -5,    -200,  32630, 212,   -15,   -328,  32613, 359,   -26,   -450,
+    32586, 512,   -36,   -568,  32551, 669,   -47,   -680,  32507, 832,   -58,   -788,  32454,
+    1000,  -69,   -891,  32393, 1174,  -80,   -990,  32323, 1352,  -92,   -1084, 32244, 1536,
+    -103,  -1173, 32157, 1724,  -115,  -1258, 32061, 1919,  -128,  -1338, 31956, 2118,  -140,
+    -1414, 31844, 2322,  -153,  -1486, 31723, 2532,  -167,  -1554, 31593, 2747,  -180,  -1617,
+    31456, 2967,  -194,  -1676, 31310, 3192,  -209,  -1732, 31157, 3422,  -224,  -1783, 30995,
+    3657,  -240,  -1830, 30826, 3897,  -256,  -1874, 30649, 4143,  -272,  -1914, 30464, 4393,
+    -289,  -1951, 30272, 4648,  -307,  -1984, 30072, 4908,  -325,  -2014, 29866, 5172,  -343,
+    -2040, 29652, 5442,  -362,  -2063, 29431, 5716,  -382,  -2083, 29203, 5994,  -403,  -2100,
+    28968, 6277,  -424,  -2114, 28727, 6565,  -445,  -2125, 28480, 6857,  -468,  -2133, 28226,
+    7153,  -490,  -2139, 27966, 7453,  -514,  -2142, 27700, 7758,  -538,  -2142, 27428, 8066,
+    -563,  -2141, 27151, 8378,  -588,  -2136, 26867, 8694,  -614,  -2130, 26579, 9013,  -641,
+    -2121, 26285, 9336,  -668,  -2111, 25987, 9663,  -696,  -2098, 25683, 9993,  -724,  -2084,
+    25375, 10326, -753,  -2067, 25063, 10662, -783,  -2049, 24746, 11000, -813,  -2030, 24425,
+    11342, -844,  -2009, 24100, 11686, -875,  -1986, 23771, 12033, -907,  -1962, 23438, 12382,
+    -939,  -1937, 23103, 12733, -972,  -1911, 22764, 13086, -1005, -1883, 22422, 13441, -1039,
+    -1855, 22077, 13798, -1072, -1825, 21729, 14156, -1107, -1795, 21380, 14516, -1141, -1764,
+    21027, 14877, -1176, -1732, 20673, 15239, -1211, -1700, 20317, 15602, -1246, -1667, 19959,
+    15965, -1282, -1633, 19600, 16329, -1317, -1599, 19239, 16694, -1353, -1564, 18878, 17058,
+    -1388, -1530, 18515, 17423, -1424, -1495, 18151, 17787, -1459, -1459, 17787, 18151, -1495,
+    -1424, 17423, 18515, -1530, -1388, 17058, 18878, -1564, -1353, 16694, 19239, -1599, -1317,
+    16329, 19600, -1633, -1282, 15965, 19959, -1667, -1246, 15602, 20317, -1700, -1211, 15239,
+    20673, -1732, -1176, 14877, 21027, -1764, -1141, 14516, 21380, -1795, -1107, 14156, 21729,
+    -1825, -1072, 13798, 22077, -1855, -1039, 13441, 22422, -1883, -1005, 13086, 22764, -1911,
+    -972,  12733, 23103, -1937, -939,  12382, 23438, -1962, -907,  12033, 23771, -1986, -875,
+    11686, 24100, -2009, -844,  11342, 24425, -2030, -813,  11000, 24746, -2049, -783,  10662,
+    25063, -2067, -753,  10326, 25375, -2084, -724,  9993,  25683, -2098, -696,  9663,  25987,
+    -2111, -668,  9336,  26285, -2121, -641,  9013,  26579, -2130, -614,  8694,  26867, -2136,
+    -588,  8378,  27151, -2141, -563,  8066,  27428, -2142, -538,  7758,  27700, -2142, -514,
+    7453,  27966, -2139, -490,  7153,  28226, -2133, -468,  6857,  28480, -2125, -445,  6565,
+    28727, -2114, -424,  6277,  28968, -2100, -403,  5994,  29203, -2083, -382,  5716,  29431,
+    -2063, -362,  5442,  29652, -2040, -343,  5172,  29866, -2014, -325,  4908,  30072, -1984,
+    -307,  4648,  30272, -1951, -289,  4393,  30464, -1914, -272,  4143,  30649, -1874, -256,
+    3897,  30826, -1830, -240,  3657,  30995, -1783, -224,  3422,  31157, -1732, -209,  3192,
+    31310, -1676, -194,  2967,  31456, -1617, -180,  2747,  31593, -1554, -167,  2532,  31723,
+    -1486, -153,  2322,  31844, -1414, -140,  2118,  31956, -1338, -128,  1919,  32061, -1258,
+    -115,  1724,  32157, -1173, -103,  1536,  32244, -1084, -92,   1352,  32323, -990,  -80,
+    1174,  32393, -891,  -69,   1000,  32454, -788,  -58,   832,   32507, -680,  -47,   669,
+    32551, -568,  -36,   512,   32586, -450,  -26,   359,   32613, -328,  -15,   212,   32630,
+    -200,  -5,    69,    32639, -68};
+
+constexpr std::array<s16, 512> curve_lut2{
+    3195,  26287, 3329,  -32,   3064,  26281, 3467,  -34,   2936,  26270, 3608,  -38,   2811,
+    26253, 3751,  -42,   2688,  26230, 3897,  -46,   2568,  26202, 4046,  -50,   2451,  26169,
+    4199,  -54,   2338,  26130, 4354,  -58,   2227,  26085, 4512,  -63,   2120,  26035, 4673,
+    -67,   2015,  25980, 4837,  -72,   1912,  25919, 5004,  -76,   1813,  25852, 5174,  -81,
+    1716,  25780, 5347,  -87,   1622,  25704, 5522,  -92,   1531,  25621, 5701,  -98,   1442,
+    25533, 5882,  -103,  1357,  25440, 6066,  -109,  1274,  25342, 6253,  -115,  1193,  25239,
+    6442,  -121,  1115,  25131, 6635,  -127,  1040,  25018, 6830,  -133,  967,   24899, 7027,
+    -140,  897,   24776, 7227,  -146,  829,   24648, 7430,  -153,  764,   24516, 7635,  -159,
+    701,   24379, 7842,  -166,  641,   24237, 8052,  -174,  583,   24091, 8264,  -181,  526,
+    23940, 8478,  -187,  472,   23785, 8695,  -194,  420,   23626, 8914,  -202,  371,   23462,
+    9135,  -209,  324,   23295, 9358,  -215,  279,   23123, 9583,  -222,  236,   22948, 9809,
+    -230,  194,   22769, 10038, -237,  154,   22586, 10269, -243,  117,   22399, 10501, -250,
+    81,    22208, 10735, -258,  47,    22015, 10970, -265,  15,    21818, 11206, -271,  -16,
+    21618, 11444, -277,  -44,   21415, 11684, -283,  -71,   21208, 11924, -290,  -97,   20999,
+    12166, -296,  -121,  20786, 12409, -302,  -143,  20571, 12653, -306,  -163,  20354, 12898,
+    -311,  -183,  20134, 13143, -316,  -201,  19911, 13389, -321,  -218,  19686, 13635, -325,
+    -234,  19459, 13882, -328,  -248,  19230, 14130, -332,  -261,  18998, 14377, -335,  -273,
+    18765, 14625, -337,  -284,  18531, 14873, -339,  -294,  18295, 15121, -341,  -302,  18057,
+    15369, -341,  -310,  17817, 15617, -341,  -317,  17577, 15864, -340,  -323,  17335, 16111,
+    -340,  -328,  17092, 16357, -338,  -332,  16848, 16603, -336,  -336,  16603, 16848, -332,
+    -338,  16357, 17092, -328,  -340,  16111, 17335, -323,  -340,  15864, 17577, -317,  -341,
+    15617, 17817, -310,  -341,  15369, 18057, -302,  -341,  15121, 18295, -294,  -339,  14873,
+    18531, -284,  -337,  14625, 18765, -273,  -335,  14377, 18998, -261,  -332,  14130, 19230,
+    -248,  -328,  13882, 19459, -234,  -325,  13635, 19686, -218,  -321,  13389, 19911, -201,
+    -316,  13143, 20134, -183,  -311,  12898, 20354, -163,  -306,  12653, 20571, -143,  -302,
+    12409, 20786, -121,  -296,  12166, 20999, -97,   -290,  11924, 21208, -71,   -283,  11684,
+    21415, -44,   -277,  11444, 21618, -16,   -271,  11206, 21818, 15,    -265,  10970, 22015,
+    47,    -258,  10735, 22208, 81,    -250,  10501, 22399, 117,   -243,  10269, 22586, 154,
+    -237,  10038, 22769, 194,   -230,  9809,  22948, 236,   -222,  9583,  23123, 279,   -215,
+    9358,  23295, 324,   -209,  9135,  23462, 371,   -202,  8914,  23626, 420,   -194,  8695,
+    23785, 472,   -187,  8478,  23940, 526,   -181,  8264,  24091, 583,   -174,  8052,  24237,
+    641,   -166,  7842,  24379, 701,   -159,  7635,  24516, 764,   -153,  7430,  24648, 829,
+    -146,  7227,  24776, 897,   -140,  7027,  24899, 967,   -133,  6830,  25018, 1040,  -127,
+    6635,  25131, 1115,  -121,  6442,  25239, 1193,  -115,  6253,  25342, 1274,  -109,  6066,
+    25440, 1357,  -103,  5882,  25533, 1442,  -98,   5701,  25621, 1531,  -92,   5522,  25704,
+    1622,  -87,   5347,  25780, 1716,  -81,   5174,  25852, 1813,  -76,   5004,  25919, 1912,
+    -72,   4837,  25980, 2015,  -67,   4673,  26035, 2120,  -63,   4512,  26085, 2227,  -58,
+    4354,  26130, 2338,  -54,   4199,  26169, 2451,  -50,   4046,  26202, 2568,  -46,   3897,
+    26230, 2688,  -42,   3751,  26253, 2811,  -38,   3608,  26270, 2936,  -34,   3467,  26281,
+    3064,  -32,   3329,  26287, 3195};

 std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input, double ratio) {
    if (input.size() < 2)
@@ -27,43 +147,51 @@ std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input,

    if (ratio <= 0) {
        LOG_CRITICAL(Audio, "Nonsensical interpolation ratio {}", ratio);
-        ratio = 1.0;
+        return input;
    }

-    if (ratio != state.current_ratio) {
-        const double cutoff_frequency = std::min(0.5 / ratio, 0.5 * ratio);
-        state.nyquist = CascadingFilter::LowPass(std::clamp(cutoff_frequency, 0.0, 0.4), 3);
-        state.current_ratio = ratio;
-    }
-    state.nyquist.Process(input);
+    const s32 step{static_cast<s32>(ratio * 0x8000)};
+    const std::array<s16, 512>& lut = [step] {
+        if (step > 0xaaaa) {
+            return curve_lut0;
+        }
+        if (step <= 0x8000) {
+            return curve_lut1;
+        }
+        return curve_lut2;
+    }();

-    constexpr std::size_t taps = InterpolationState::lanczos_taps;
-    const std::size_t num_frames = input.size() / 2;
+    const std::size_t num_frames{input.size() / 2};

    std::vector<s16> output;
-    output.reserve(static_cast<std::size_t>(input.size() / ratio + 4));
+    output.reserve(static_cast<std::size_t>(input.size() / ratio + InterpolationState::taps));

-    double& pos = state.position;
-    auto& h = state.history;
-    for (std::size_t i = 0; i < num_frames; ++i) {
-        std::rotate(h.begin(), h.end() - 1, h.end());
-        h[0][0] = input[i * 2 + 0];
-        h[0][1] = input[i * 2 + 1];
+    for (std::size_t frame{}; frame < num_frames; ++frame) {
+        const std::size_t lut_index{(state.fraction >> 8) * InterpolationState::taps};

-        while (pos <= 1.0) {
-            double l = 0.0;
-            double r = 0.0;
-            for (std::size_t j = 0; j < h.size(); j++) {
-                const double lanczos_calc = Lanczos(taps, pos + j - taps + 1);
-                l += lanczos_calc * h[j][0];
-                r += lanczos_calc * h[j][1];
-            }
-            output.emplace_back(static_cast<s16>(std::clamp(l, -32768.0, 32767.0)));
-            output.emplace_back(static_cast<s16>(std::clamp(r, -32768.0, 32767.0)));
+        std::rotate(state.history.begin(), state.history.end() - 1, state.history.end());
+        state.history[0][0] = input[frame * 2 + 0];
+        state.history[0][1] = input[frame * 2 + 1];

-            pos += ratio;
+        while (state.position <= 1.0) {
+            const s32 left{state.history[0][0] * lut[lut_index + 0] +
+                           state.history[1][0] * lut[lut_index + 1] +
+                           state.history[2][0] * lut[lut_index + 2] +
+                           state.history[3][0] * lut[lut_index + 3]};
+            const s32 right{state.history[0][1] * lut[lut_index + 0] +
+                            state.history[1][1] * lut[lut_index + 1] +
+                            state.history[2][1] * lut[lut_index + 2] +
+                            state.history[3][1] * lut[lut_index + 3]};
+            const s32 new_offset{state.fraction + step};
+
+            state.fraction = new_offset & 0x7fff;
+
+            output.emplace_back(static_cast<s16>(std::clamp(left >> 15, SHRT_MIN, SHRT_MAX)));
+            output.emplace_back(static_cast<s16>(std::clamp(right >> 15, SHRT_MIN, SHRT_MAX)));
+
+            state.position += ratio;
        }
-        pos -= 1.0;
+        state.position -= 1.0;
    }

    return output;
@@ -6,19 +6,17 @@

 #include <array>
 #include <vector>
-#include "audio_core/algorithm/filter.h"
+
 #include "common/common_types.h"

 namespace AudioCore {

 struct InterpolationState {
-    static constexpr std::size_t lanczos_taps = 4;
-    static constexpr std::size_t history_size = lanczos_taps * 2 - 1;
-
-    double current_ratio = 0.0;
-    CascadingFilter nyquist;
-    std::array<std::array<s16, 2>, history_size> history = {};
-    double position = 0;
+    static constexpr std::size_t taps{4};
+    static constexpr std::size_t history_size{taps * 2 - 1};
+    std::array<std::array<s16, 2>, history_size> history{};
+    double position{};
+    s32 fraction{};
 };

 /// Interpolates input signal to produce output signal.
@@ -131,8 +131,8 @@ add_library(core STATIC
    frontend/framebuffer_layout.cpp
    frontend/framebuffer_layout.h
    frontend/input.h
-    frontend/scope_acquire_window_context.cpp
-    frontend/scope_acquire_window_context.h
+    frontend/scope_acquire_context.cpp
+    frontend/scope_acquire_context.h
    gdbstub/gdbstub.cpp
    gdbstub/gdbstub.h
    hardware_interrupt_manager.cpp
@@ -181,14 +181,18 @@ add_library(core STATIC
    hle/kernel/svc.cpp
    hle/kernel/svc.h
    hle/kernel/svc_wrap.h
+    hle/kernel/synchronization_object.cpp
+    hle/kernel/synchronization_object.h
+    hle/kernel/synchronization.cpp
+    hle/kernel/synchronization.h
    hle/kernel/thread.cpp
    hle/kernel/thread.h
+    hle/kernel/time_manager.cpp
+    hle/kernel/time_manager.h
    hle/kernel/transfer_memory.cpp
    hle/kernel/transfer_memory.h
    hle/kernel/vm_manager.cpp
    hle/kernel/vm_manager.h
-    hle/kernel/wait_object.cpp
-    hle/kernel/wait_object.h
    hle/kernel/writable_event.cpp
    hle/kernel/writable_event.h
    hle/lock.cpp
@@ -14,6 +14,7 @@
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/svc.h"
@@ -153,7 +154,7 @@ std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit(Common::PageTable& pag
    config.tpidr_el0 = &cb->tpidr_el0;
    config.dczid_el0 = 4;
    config.ctr_el0 = 0x8444c004;
-    config.cntfrq_el0 = Timing::CNTFREQ;
+    config.cntfrq_el0 = Hardware::CNTFREQ;

    // Unpredictable instructions
    config.define_unpredictable_behaviour = true;
@@ -24,6 +24,7 @@
 #include "core/file_sys/sdmc_factory.h"
 #include "core/file_sys/vfs_concat.h"
 #include "core/file_sys/vfs_real.h"
+#include "core/frontend/scope_acquire_context.h"
 #include "core/gdbstub/gdbstub.h"
 #include "core/hardware_interrupt_manager.h"
 #include "core/hle/kernel/client_port.h"
@@ -184,6 +185,8 @@ struct System::Impl {

    ResultStatus Load(System& system, Frontend::EmuWindow& emu_window,
                      const std::string& filepath) {
+        Core::Frontend::ScopeAcquireContext acquire_context{emu_window};
+
        app_loader = Loader::GetLoader(GetGameFileFromPath(virtual_filesystem, filepath));
        if (!app_loader) {
            LOG_CRITICAL(Core, "Failed to obtain loader for {}!", filepath);
@@ -707,4 +710,12 @@ const Service::SM::ServiceManager& System::ServiceManager() const {
    return *impl->service_manager;
 }

+void System::RegisterCoreThread(std::size_t id) {
+    impl->kernel.RegisterCoreThread(id);
+}
+
+void System::RegisterHostThread() {
+    impl->kernel.RegisterHostThread();
+}
+
 } // namespace Core
@@ -360,6 +360,12 @@ public:

    const CurrentBuildProcessID& GetCurrentProcessBuildID() const;

+    /// Register a host thread as an emulated CPU Core.
+    void RegisterCoreThread(std::size_t id);
+
+    /// Register a host thread as an auxiliary thread.
+    void RegisterHostThread();
+
 private:
    System();

@@ -12,6 +12,7 @@
 #include "common/assert.h"
 #include "common/thread.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"

 namespace Core::Timing {

@@ -215,7 +216,7 @@ void CoreTiming::Idle() {
 }

 std::chrono::microseconds CoreTiming::GetGlobalTimeUs() const {
-    return std::chrono::microseconds{GetTicks() * 1000000 / BASE_CLOCK_RATE};
+    return std::chrono::microseconds{GetTicks() * 1000000 / Hardware::BASE_CLOCK_RATE};
 }

 s64 CoreTiming::GetDowncount() const {
@@ -11,7 +11,7 @@

 namespace Core::Timing {

-constexpr u64 MAX_VALUE_TO_MULTIPLY = std::numeric_limits<s64>::max() / BASE_CLOCK_RATE;
+constexpr u64 MAX_VALUE_TO_MULTIPLY = std::numeric_limits<s64>::max() / Hardware::BASE_CLOCK_RATE;

 s64 msToCycles(std::chrono::milliseconds ms) {
    if (static_cast<u64>(ms.count() / 1000) > MAX_VALUE_TO_MULTIPLY) {
@@ -20,9 +20,9 @@ s64 msToCycles(std::chrono::milliseconds ms) {
    }
    if (static_cast<u64>(ms.count()) > MAX_VALUE_TO_MULTIPLY) {
        LOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return BASE_CLOCK_RATE * (ms.count() / 1000);
+        return Hardware::BASE_CLOCK_RATE * (ms.count() / 1000);
    }
-    return (BASE_CLOCK_RATE * ms.count()) / 1000;
+    return (Hardware::BASE_CLOCK_RATE * ms.count()) / 1000;
 }

 s64 usToCycles(std::chrono::microseconds us) {
@@ -32,9 +32,9 @@ s64 usToCycles(std::chrono::microseconds us) {
    }
    if (static_cast<u64>(us.count()) > MAX_VALUE_TO_MULTIPLY) {
        LOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return BASE_CLOCK_RATE * (us.count() / 1000000);
+        return Hardware::BASE_CLOCK_RATE * (us.count() / 1000000);
    }
-    return (BASE_CLOCK_RATE * us.count()) / 1000000;
+    return (Hardware::BASE_CLOCK_RATE * us.count()) / 1000000;
 }

 s64 nsToCycles(std::chrono::nanoseconds ns) {
@@ -44,14 +44,14 @@ s64 nsToCycles(std::chrono::nanoseconds ns) {
    }
    if (static_cast<u64>(ns.count()) > MAX_VALUE_TO_MULTIPLY) {
        LOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return BASE_CLOCK_RATE * (ns.count() / 1000000000);
+        return Hardware::BASE_CLOCK_RATE * (ns.count() / 1000000000);
    }
-    return (BASE_CLOCK_RATE * ns.count()) / 1000000000;
+    return (Hardware::BASE_CLOCK_RATE * ns.count()) / 1000000000;
 }

 u64 CpuCyclesToClockCycles(u64 ticks) {
-    const u128 temporal = Common::Multiply64Into128(ticks, CNTFREQ);
-    return Common::Divide128On32(temporal, static_cast<u32>(BASE_CLOCK_RATE)).first;
+    const u128 temporal = Common::Multiply64Into128(ticks, Hardware::CNTFREQ);
+    return Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first;
 }

 } // namespace Core::Timing
@@ -6,28 +6,24 @@

 #include <chrono>
 #include "common/common_types.h"
+#include "core/hardware_properties.h"

 namespace Core::Timing {

-// 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 clock speed is 1020MHz un/docked
-constexpr u64 CNTFREQ = 19200000;           // Value from fusee.
-
 s64 msToCycles(std::chrono::milliseconds ms);
 s64 usToCycles(std::chrono::microseconds us);
 s64 nsToCycles(std::chrono::nanoseconds ns);

 inline std::chrono::milliseconds CyclesToMs(s64 cycles) {
-    return std::chrono::milliseconds(cycles * 1000 / BASE_CLOCK_RATE);
+    return std::chrono::milliseconds(cycles * 1000 / Hardware::BASE_CLOCK_RATE);
 }

 inline std::chrono::nanoseconds CyclesToNs(s64 cycles) {
-    return std::chrono::nanoseconds(cycles * 1000000000 / BASE_CLOCK_RATE);
+    return std::chrono::nanoseconds(cycles * 1000000000 / Hardware::BASE_CLOCK_RATE);
 }

 inline std::chrono::microseconds CyclesToUs(s64 cycles) {
-    return std::chrono::microseconds(cycles * 1000000 / BASE_CLOCK_RATE);
+    return std::chrono::microseconds(cycles * 1000000 / Hardware::BASE_CLOCK_RATE);
 }

 u64 CpuCyclesToClockCycles(u64 ticks);
@@ -6,6 +6,7 @@

 #include <array>
 #include <memory>
+#include "core/hardware_properties.h"

 namespace Core {

@@ -39,9 +40,7 @@ public:
    void RunLoop(bool tight_loop);

 private:
-    static constexpr std::size_t NUM_CPU_CORES = 4;
-
-    std::array<std::unique_ptr<CoreManager>, NUM_CPU_CORES> core_managers;
+    std::array<std::unique_ptr<CoreManager>, Hardware::NUM_CPU_CORES> core_managers;
    std::size_t active_core{}; ///< Active core, only used in single thread mode

    System& system;
@@ -26,9 +26,6 @@ public:

    /// Releases (dunno if this is the "right" word) the context from the caller thread
    virtual void DoneCurrent() = 0;
-
-    /// Swap buffers to display the next frame
-    virtual void SwapBuffers() = 0;
 };

 /**
@@ -27,11 +27,9 @@ FramebufferLayout DefaultFrameLayout(u32 width, u32 height) {
    // so just calculate them both even if the other isn't showing.
    FramebufferLayout res{width, height};

-    const float emulation_aspect_ratio =
-        Settings::values.aspect_ratio
-            ? static_cast<float>(height) / width
-            : static_cast<float>(ScreenUndocked::Height) / ScreenUndocked::Width;
-    const auto window_aspect_ratio = static_cast<float>(height) / width;
+    const float window_aspect_ratio = static_cast<float>(height) / width;
+    const float emulation_aspect_ratio = EmulationAspectRatio(
+        static_cast<AspectRatio>(Settings::values.aspect_ratio), window_aspect_ratio);

    const Common::Rectangle<u32> screen_window_area{0, 0, width, height};
    Common::Rectangle<u32> screen = MaxRectangle(screen_window_area, emulation_aspect_ratio);
@@ -60,4 +58,19 @@ FramebufferLayout FrameLayoutFromResolutionScale(u32 res_scale) {
    return DefaultFrameLayout(width, height);
 }

+float EmulationAspectRatio(AspectRatio aspect, float window_aspect_ratio) {
+    switch (aspect) {
+    case AspectRatio::Default:
+        return static_cast<float>(ScreenUndocked::Height) / ScreenUndocked::Width;
+    case AspectRatio::R4_3:
+        return 3.0f / 4.0f;
+    case AspectRatio::R21_9:
+        return 9.0f / 21.0f;
+    case AspectRatio::StretchToWindow:
+        return window_aspect_ratio;
+    default:
+        return static_cast<float>(ScreenUndocked::Height) / ScreenUndocked::Width;
+    }
+}
+
 } // namespace Layout
@@ -18,10 +18,18 @@ enum ScreenDocked : u32 {
    HeightDocked = 1080,
 };

+enum class AspectRatio {
+    Default,
+    R4_3,
+    R21_9,
+    StretchToWindow,
+};
+
 /// Describes the layout of the window framebuffer
 struct FramebufferLayout {
    u32 width{ScreenUndocked::Width};
    u32 height{ScreenUndocked::Height};
+    bool is_srgb{};

    Common::Rectangle<u32> screen;

@@ -48,4 +56,12 @@ FramebufferLayout DefaultFrameLayout(u32 width, u32 height);
 */
 FramebufferLayout FrameLayoutFromResolutionScale(u32 res_scale);

+/**
+ * Convenience method to determine emulation aspect ratio
+ * @param aspect Represents the index of aspect ratio stored in Settings::values.aspect_ratio
+ * @param window_aspect_ratio Current window aspect ratio
+ * @return Emulation render window aspect ratio
+ */
+float EmulationAspectRatio(AspectRatio aspect, float window_aspect_ratio);
+
 } // namespace Layout
@@ -0,0 +1,18 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/frontend/emu_window.h"
+#include "core/frontend/scope_acquire_context.h"
+
+namespace Core::Frontend {
+
+ScopeAcquireContext::ScopeAcquireContext(Core::Frontend::GraphicsContext& context)
+    : context{context} {
+    context.MakeCurrent();
+}
+ScopeAcquireContext::~ScopeAcquireContext() {
+    context.DoneCurrent();
+}
+
+} // namespace Core::Frontend
@@ -8,16 +8,16 @@

 namespace Core::Frontend {

-class EmuWindow;
+class GraphicsContext;

 /// Helper class to acquire/release window context within a given scope
-class ScopeAcquireWindowContext : NonCopyable {
+class ScopeAcquireContext : NonCopyable {
 public:
-    explicit ScopeAcquireWindowContext(Core::Frontend::EmuWindow& window);
-    ~ScopeAcquireWindowContext();
+    explicit ScopeAcquireContext(Core::Frontend::GraphicsContext& context);
+    ~ScopeAcquireContext();

 private:
-    Core::Frontend::EmuWindow& emu_window;
+    Core::Frontend::GraphicsContext& context;
 };

 } // namespace Core::Frontend
@@ -1,18 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "core/frontend/emu_window.h"
-#include "core/frontend/scope_acquire_window_context.h"
-
-namespace Core::Frontend {
-
-ScopeAcquireWindowContext::ScopeAcquireWindowContext(Core::Frontend::EmuWindow& emu_window_)
-    : emu_window{emu_window_} {
-    emu_window.MakeCurrent();
-}
-ScopeAcquireWindowContext::~ScopeAcquireWindowContext() {
-    emu_window.DoneCurrent();
-}
-
-} // namespace Core::Frontend
@@ -0,0 +1,47 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <tuple>
+
+#include "common/common_types.h"
+
+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
+
+} // namespace Hardware
+
+constexpr u32 INVALID_HOST_THREAD_ID = 0xFFFFFFFF;
+
+struct EmuThreadHandle {
+    u32 host_handle;
+    u32 guest_handle;
+
+    u64 GetRaw() const {
+        return (static_cast<u64>(host_handle) << 32) | guest_handle;
+    }
+
+    bool operator==(const EmuThreadHandle& rhs) const {
+        return std::tie(host_handle, guest_handle) == std::tie(rhs.host_handle, rhs.guest_handle);
+    }
+
+    bool operator!=(const EmuThreadHandle& rhs) const {
+        return !operator==(rhs);
+    }
+
+    static constexpr EmuThreadHandle InvalidHandle() {
+        constexpr u32 invalid_handle = 0xFFFFFFFF;
+        return {invalid_handle, invalid_handle};
+    }
+};
+
+} // namespace Core
@@ -201,42 +201,39 @@ void AddressArbiter::HandleWakeupThread(std::shared_ptr<Thread> thread) {
 void AddressArbiter::InsertThread(std::shared_ptr<Thread> thread) {
    const VAddr arb_addr = thread->GetArbiterWaitAddress();
    std::list<std::shared_ptr<Thread>>& thread_list = arb_threads[arb_addr];
-    auto it = thread_list.begin();
-    while (it != thread_list.end()) {
-        const std::shared_ptr<Thread>& current_thread = *it;
-        if (current_thread->GetPriority() >= thread->GetPriority()) {
-            thread_list.insert(it, thread);
-            return;
-        }
-        ++it;
+
+    const auto iter =
+        std::find_if(thread_list.cbegin(), thread_list.cend(), [&thread](const auto& entry) {
+            return entry->GetPriority() >= thread->GetPriority();
+        });
+
+    if (iter == thread_list.cend()) {
+        thread_list.push_back(std::move(thread));
+    } else {
+        thread_list.insert(iter, std::move(thread));
    }
-    thread_list.push_back(std::move(thread));
 }

 void AddressArbiter::RemoveThread(std::shared_ptr<Thread> thread) {
    const VAddr arb_addr = thread->GetArbiterWaitAddress();
    std::list<std::shared_ptr<Thread>>& thread_list = arb_threads[arb_addr];
-    auto it = thread_list.begin();
-    while (it != thread_list.end()) {
-        const std::shared_ptr<Thread>& current_thread = *it;
-        if (current_thread.get() == thread.get()) {
-            thread_list.erase(it);
-            return;
-        }
-        ++it;
-    }
-    UNREACHABLE();
+
+    const auto iter = std::find_if(thread_list.cbegin(), thread_list.cend(),
+                                   [&thread](const auto& entry) { return thread == entry; });
+
+    ASSERT(iter != thread_list.cend());
+
+    thread_list.erase(iter);
 }

-std::vector<std::shared_ptr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(VAddr address) {
-    std::vector<std::shared_ptr<Thread>> result;
-    std::list<std::shared_ptr<Thread>>& thread_list = arb_threads[address];
-    auto it = thread_list.begin();
-    while (it != thread_list.end()) {
-        std::shared_ptr<Thread> current_thread = *it;
-        result.push_back(std::move(current_thread));
-        ++it;
+std::vector<std::shared_ptr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(
+    VAddr address) const {
+    const auto iter = arb_threads.find(address);
+    if (iter == arb_threads.cend()) {
+        return {};
    }
-    return result;
+
+    const std::list<std::shared_ptr<Thread>>& thread_list = iter->second;
+    return {thread_list.cbegin(), thread_list.cend()};
 }
 } // namespace Kernel
@@ -86,7 +86,7 @@ private:
    void RemoveThread(std::shared_ptr<Thread> thread);

    // Gets the threads waiting on an address.
-    std::vector<std::shared_ptr<Thread>> GetThreadsWaitingOnAddress(VAddr address);
+    std::vector<std::shared_ptr<Thread>> GetThreadsWaitingOnAddress(VAddr address) const;

    /// List of threads waiting for a address arbiter
    std::unordered_map<VAddr, std::list<std::shared_ptr<Thread>>> arb_threads;
@@ -12,7 +12,7 @@

 namespace Kernel {

-ClientSession::ClientSession(KernelCore& kernel) : WaitObject{kernel} {}
+ClientSession::ClientSession(KernelCore& kernel) : SynchronizationObject{kernel} {}

 ClientSession::~ClientSession() {
    // This destructor will be called automatically when the last ClientSession handle is closed by
@@ -31,6 +31,11 @@ void ClientSession::Acquire(Thread* thread) {
    UNIMPLEMENTED();
 }

+bool ClientSession::IsSignaled() const {
+    UNIMPLEMENTED();
+    return true;
+}
+
 ResultVal<std::shared_ptr<ClientSession>> ClientSession::Create(KernelCore& kernel,
                                                                std::shared_ptr<Session> parent,
                                                                std::string name) {
@@ -7,7 +7,7 @@
 #include <memory>
 #include <string>

-#include "core/hle/kernel/wait_object.h"
+#include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/result.h"

 union ResultCode;
@@ -22,7 +22,7 @@ class KernelCore;
 class Session;
 class Thread;

-class ClientSession final : public WaitObject {
+class ClientSession final : public SynchronizationObject {
 public:
    explicit ClientSession(KernelCore& kernel);
    ~ClientSession() override;
@@ -48,6 +48,8 @@ public:

    void Acquire(Thread* thread) override;

+    bool IsSignaled() const override;
+
 private:
    static ResultVal<std::shared_ptr<ClientSession>> Create(KernelCore& kernel,
                                                            std::shared_ptr<Session> parent,
@@ -47,15 +47,15 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
    const std::string& reason, u64 timeout, WakeupCallback&& callback,
    std::shared_ptr<WritableEvent> writable_event) {
    // Put the client thread to sleep until the wait event is signaled or the timeout expires.
-    thread->SetWakeupCallback([context = *this, callback](ThreadWakeupReason reason,
-                                                          std::shared_ptr<Thread> thread,
-                                                          std::shared_ptr<WaitObject> object,
-                                                          std::size_t index) mutable -> bool {
-        ASSERT(thread->GetStatus() == ThreadStatus::WaitHLEEvent);
-        callback(thread, context, reason);
-        context.WriteToOutgoingCommandBuffer(*thread);
-        return true;
-    });
+    thread->SetWakeupCallback(
+        [context = *this, callback](ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
+                                    std::shared_ptr<SynchronizationObject> object,
+                                    std::size_t index) mutable -> bool {
+            ASSERT(thread->GetStatus() == ThreadStatus::WaitHLEEvent);
+            callback(thread, context, reason);
+            context.WriteToOutgoingCommandBuffer(*thread);
+            return true;
+        });

    auto& kernel = Core::System::GetInstance().Kernel();
    if (!writable_event) {
@@ -67,7 +67,7 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
    const auto readable_event{writable_event->GetReadableEvent()};
    writable_event->Clear();
    thread->SetStatus(ThreadStatus::WaitHLEEvent);
-    thread->SetWaitObjects({readable_event});
+    thread->SetSynchronizationObjects({readable_event});
    readable_event->AddWaitingThread(thread);

    if (timeout > 0) {
@@ -3,9 +3,12 @@
 // Refer to the license.txt file included.

 #include <atomic>
+#include <bitset>
 #include <functional>
 #include <memory>
 #include <mutex>
+#include <thread>
+#include <unordered_map>
 #include <utility>

 #include "common/assert.h"
@@ -15,6 +18,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
@@ -23,7 +27,9 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/kernel/scheduler.h"
+#include "core/hle/kernel/synchronization.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
 #include "core/hle/lock.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
@@ -43,7 +49,7 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
    std::lock_guard lock{HLE::g_hle_lock};

    std::shared_ptr<Thread> thread =
-        system.Kernel().RetrieveThreadFromWakeupCallbackHandleTable(proper_handle);
+        system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
    if (thread == nullptr) {
        LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle);
        return;
@@ -54,10 +60,10 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
    if (thread->GetStatus() == ThreadStatus::WaitSynch ||
        thread->GetStatus() == ThreadStatus::WaitHLEEvent) {
        // Remove the thread from each of its waiting objects' waitlists
-        for (const auto& object : thread->GetWaitObjects()) {
+        for (const auto& object : thread->GetSynchronizationObjects()) {
            object->RemoveWaitingThread(thread);
        }
-        thread->ClearWaitObjects();
+        thread->ClearSynchronizationObjects();

        // Invoke the wakeup callback before clearing the wait objects
        if (thread->HasWakeupCallback()) {
@@ -96,7 +102,8 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
 }

 struct KernelCore::Impl {
-    explicit Impl(Core::System& system) : system{system}, global_scheduler{system} {}
+    explicit Impl(Core::System& system, KernelCore& kernel)
+        : system{system}, global_scheduler{kernel}, synchronization{system}, time_manager{system} {}

    void Initialize(KernelCore& kernel) {
        Shutdown();
@@ -118,7 +125,7 @@ struct KernelCore::Impl {

        system_resource_limit = nullptr;

-        thread_wakeup_callback_handle_table.Clear();
+        global_handle_table.Clear();
        thread_wakeup_event_type = nullptr;
        preemption_event = nullptr;

@@ -136,8 +143,8 @@ struct KernelCore::Impl {

    void InitializePhysicalCores() {
        exclusive_monitor =
-            Core::MakeExclusiveMonitor(system.Memory(), global_scheduler.CpuCoresCount());
-        for (std::size_t i = 0; i < global_scheduler.CpuCoresCount(); i++) {
+            Core::MakeExclusiveMonitor(system.Memory(), Core::Hardware::NUM_CPU_CORES);
+        for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
            cores.emplace_back(system, i, *exclusive_monitor);
        }
    }
@@ -182,6 +189,50 @@ struct KernelCore::Impl {
        system.Memory().SetCurrentPageTable(*process);
    }

+    void RegisterCoreThread(std::size_t core_id) {
+        std::unique_lock lock{register_thread_mutex};
+        const std::thread::id this_id = std::this_thread::get_id();
+        const auto it = host_thread_ids.find(this_id);
+        ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
+        ASSERT(it == host_thread_ids.end());
+        ASSERT(!registered_core_threads[core_id]);
+        host_thread_ids[this_id] = static_cast<u32>(core_id);
+        registered_core_threads.set(core_id);
+    }
+
+    void RegisterHostThread() {
+        std::unique_lock lock{register_thread_mutex};
+        const std::thread::id this_id = std::this_thread::get_id();
+        const auto it = host_thread_ids.find(this_id);
+        ASSERT(it == host_thread_ids.end());
+        host_thread_ids[this_id] = registered_thread_ids++;
+    }
+
+    u32 GetCurrentHostThreadID() const {
+        const std::thread::id this_id = std::this_thread::get_id();
+        const auto it = host_thread_ids.find(this_id);
+        if (it == host_thread_ids.end()) {
+            return Core::INVALID_HOST_THREAD_ID;
+        }
+        return it->second;
+    }
+
+    Core::EmuThreadHandle GetCurrentEmuThreadID() const {
+        Core::EmuThreadHandle result = Core::EmuThreadHandle::InvalidHandle();
+        result.host_handle = GetCurrentHostThreadID();
+        if (result.host_handle >= Core::Hardware::NUM_CPU_CORES) {
+            return result;
+        }
+        const Kernel::Scheduler& sched = cores[result.host_handle].Scheduler();
+        const Kernel::Thread* current = sched.GetCurrentThread();
+        if (current != nullptr) {
+            result.guest_handle = current->GetGlobalHandle();
+        } else {
+            result.guest_handle = InvalidHandle;
+        }
+        return result;
+    }
+
    std::atomic<u32> next_object_id{0};
    std::atomic<u64> next_kernel_process_id{Process::InitialKIPIDMin};
    std::atomic<u64> next_user_process_id{Process::ProcessIDMin};
@@ -191,15 +242,17 @@ struct KernelCore::Impl {
    std::vector<std::shared_ptr<Process>> process_list;
    Process* current_process = nullptr;
    Kernel::GlobalScheduler global_scheduler;
+    Kernel::Synchronization synchronization;
+    Kernel::TimeManager time_manager;

    std::shared_ptr<ResourceLimit> system_resource_limit;

    std::shared_ptr<Core::Timing::EventType> thread_wakeup_event_type;
    std::shared_ptr<Core::Timing::EventType> preemption_event;

-    // TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future,
-    // allowing us to simply use a pool index or similar.
-    Kernel::HandleTable thread_wakeup_callback_handle_table;
+    // This is the kernel's handle table or supervisor handle table which
+    // stores all the objects in place.
+    Kernel::HandleTable global_handle_table;

    /// Map of named ports managed by the kernel, which can be retrieved using
    /// the ConnectToPort SVC.
@@ -208,11 +261,17 @@ struct KernelCore::Impl {
    std::unique_ptr<Core::ExclusiveMonitor> exclusive_monitor;
    std::vector<Kernel::PhysicalCore> cores;

+    // 0-3 IDs represent core threads, >3 represent others
+    std::unordered_map<std::thread::id, u32> host_thread_ids;
+    u32 registered_thread_ids{Core::Hardware::NUM_CPU_CORES};
+    std::bitset<Core::Hardware::NUM_CPU_CORES> registered_core_threads;
+    std::mutex register_thread_mutex;
+
    // System context
    Core::System& system;
 };

-KernelCore::KernelCore(Core::System& system) : impl{std::make_unique<Impl>(system)} {}
+KernelCore::KernelCore(Core::System& system) : impl{std::make_unique<Impl>(system, *this)} {}
 KernelCore::~KernelCore() {
    Shutdown();
 }
@@ -229,9 +288,8 @@ std::shared_ptr<ResourceLimit> KernelCore::GetSystemResourceLimit() const {
    return impl->system_resource_limit;
 }

-std::shared_ptr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable(
-    Handle handle) const {
-    return impl->thread_wakeup_callback_handle_table.Get<Thread>(handle);
+std::shared_ptr<Thread> KernelCore::RetrieveThreadFromGlobalHandleTable(Handle handle) const {
+    return impl->global_handle_table.Get<Thread>(handle);
 }

 void KernelCore::AppendNewProcess(std::shared_ptr<Process> process) {
@@ -262,6 +320,14 @@ const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const {
    return impl->global_scheduler;
 }

+Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) {
+    return impl->cores[id].Scheduler();
+}
+
+const Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) const {
+    return impl->cores[id].Scheduler();
+}
+
 Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) {
    return impl->cores[id];
 }
@@ -270,6 +336,22 @@ const Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) const {
    return impl->cores[id];
 }

+Kernel::Synchronization& KernelCore::Synchronization() {
+    return impl->synchronization;
+}
+
+const Kernel::Synchronization& KernelCore::Synchronization() const {
+    return impl->synchronization;
+}
+
+Kernel::TimeManager& KernelCore::TimeManager() {
+    return impl->time_manager;
+}
+
+const Kernel::TimeManager& KernelCore::TimeManager() const {
+    return impl->time_manager;
+}
+
 Core::ExclusiveMonitor& KernelCore::GetExclusiveMonitor() {
    return *impl->exclusive_monitor;
 }
@@ -327,12 +409,28 @@ const std::shared_ptr<Core::Timing::EventType>& KernelCore::ThreadWakeupCallback
    return impl->thread_wakeup_event_type;
 }

-Kernel::HandleTable& KernelCore::ThreadWakeupCallbackHandleTable() {
-    return impl->thread_wakeup_callback_handle_table;
+Kernel::HandleTable& KernelCore::GlobalHandleTable() {
+    return impl->global_handle_table;
 }

-const Kernel::HandleTable& KernelCore::ThreadWakeupCallbackHandleTable() const {
-    return impl->thread_wakeup_callback_handle_table;
+const Kernel::HandleTable& KernelCore::GlobalHandleTable() const {
+    return impl->global_handle_table;
+}
+
+void KernelCore::RegisterCoreThread(std::size_t core_id) {
+    impl->RegisterCoreThread(core_id);
+}
+
+void KernelCore::RegisterHostThread() {
+    impl->RegisterHostThread();
+}
+
+u32 KernelCore::GetCurrentHostThreadID() const {
+    return impl->GetCurrentHostThreadID();
+}
+
+Core::EmuThreadHandle KernelCore::GetCurrentEmuThreadID() const {
+    return impl->GetCurrentEmuThreadID();
 }

 } // namespace Kernel
@@ -11,6 +11,7 @@
 #include "core/hle/kernel/object.h"

 namespace Core {
+struct EmuThreadHandle;
 class ExclusiveMonitor;
 class System;
 } // namespace Core
@@ -29,7 +30,10 @@ class HandleTable;
 class PhysicalCore;
 class Process;
 class ResourceLimit;
+class Scheduler;
+class Synchronization;
 class Thread;
+class TimeManager;

 /// Represents a single instance of the kernel.
 class KernelCore {
@@ -63,7 +67,7 @@ public:
    std::shared_ptr<ResourceLimit> GetSystemResourceLimit() const;

    /// Retrieves a shared pointer to a Thread instance within the thread wakeup handle table.
-    std::shared_ptr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const;
+    std::shared_ptr<Thread> RetrieveThreadFromGlobalHandleTable(Handle handle) const;

    /// Adds the given shared pointer to an internal list of active processes.
    void AppendNewProcess(std::shared_ptr<Process> process);
@@ -86,12 +90,30 @@ public:
    /// Gets the sole instance of the global scheduler
    const Kernel::GlobalScheduler& GlobalScheduler() const;

+    /// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
+    Kernel::Scheduler& Scheduler(std::size_t id);
+
+    /// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
+    const Kernel::Scheduler& Scheduler(std::size_t id) const;
+
    /// Gets the an instance of the respective physical CPU core.
    Kernel::PhysicalCore& PhysicalCore(std::size_t id);

    /// Gets the an instance of the respective physical CPU core.
    const Kernel::PhysicalCore& PhysicalCore(std::size_t id) const;

+    /// Gets the an instance of the Synchronization Interface.
+    Kernel::Synchronization& Synchronization();
+
+    /// Gets the an instance of the Synchronization Interface.
+    const Kernel::Synchronization& Synchronization() const;
+
+    /// Gets the an instance of the TimeManager Interface.
+    Kernel::TimeManager& TimeManager();
+
+    /// Gets the an instance of the TimeManager Interface.
+    const Kernel::TimeManager& TimeManager() const;
+
    /// Stops execution of 'id' core, in order to reschedule a new thread.
    void PrepareReschedule(std::size_t id);

@@ -113,6 +135,18 @@ public:
    /// Determines whether or not the given port is a valid named port.
    bool IsValidNamedPort(NamedPortTable::const_iterator port) const;

+    /// Gets the current host_thread/guest_thread handle.
+    Core::EmuThreadHandle GetCurrentEmuThreadID() const;
+
+    /// Gets the current host_thread handle.
+    u32 GetCurrentHostThreadID() const;
+
+    /// Register the current thread as a CPU Core Thread.
+    void RegisterCoreThread(std::size_t core_id);
+
+    /// Register the current thread as a non CPU core thread.
+    void RegisterHostThread();
+
 private:
    friend class Object;
    friend class Process;
@@ -133,11 +167,11 @@ private:
    /// Retrieves the event type used for thread wakeup callbacks.
    const std::shared_ptr<Core::Timing::EventType>& ThreadWakeupCallbackEventType() const;

-    /// Provides a reference to the thread wakeup callback handle table.
-    Kernel::HandleTable& ThreadWakeupCallbackHandleTable();
+    /// Provides a reference to the global handle table.
+    Kernel::HandleTable& GlobalHandleTable();

-    /// Provides a const reference to the thread wakeup callback handle table.
-    const Kernel::HandleTable& ThreadWakeupCallbackHandleTable() const;
+    /// Provides a const reference to the global handle table.
+    const Kernel::HandleTable& GlobalHandleTable() const;

    struct Impl;
    std::unique_ptr<Impl> impl;
@@ -337,7 +337,7 @@ void Process::LoadModule(CodeSet module_, VAddr base_addr) {
 }

 Process::Process(Core::System& system)
-    : WaitObject{system.Kernel()}, vm_manager{system},
+    : SynchronizationObject{system.Kernel()}, vm_manager{system},
      address_arbiter{system}, mutex{system}, system{system} {}

 Process::~Process() = default;
@@ -357,7 +357,7 @@ void Process::ChangeStatus(ProcessStatus new_status) {

    status = new_status;
    is_signaled = true;
-    WakeupAllWaitingThreads();
+    Signal();
 }

 void Process::AllocateMainThreadStack(u64 stack_size) {
@@ -15,8 +15,8 @@
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/mutex.h"
 #include "core/hle/kernel/process_capability.h"
+#include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/kernel/vm_manager.h"
-#include "core/hle/kernel/wait_object.h"
 #include "core/hle/result.h"

 namespace Core {
@@ -60,7 +60,7 @@ enum class ProcessStatus {
    DebugBreak,
 };

-class Process final : public WaitObject {
+class Process final : public SynchronizationObject {
 public:
    explicit Process(Core::System& system);
    ~Process() override;
@@ -359,10 +359,6 @@ private:
    /// specified by metadata provided to the process during loading.
    bool is_64bit_process = true;

-    /// Whether or not this process is signaled. This occurs
-    /// upon the process changing to a different state.
-    bool is_signaled = false;
-
    /// Total running time for the process in ticks.
    u64 total_process_running_time_ticks = 0;

@@ -11,30 +11,30 @@

 namespace Kernel {

-ReadableEvent::ReadableEvent(KernelCore& kernel) : WaitObject{kernel} {}
+ReadableEvent::ReadableEvent(KernelCore& kernel) : SynchronizationObject{kernel} {}
 ReadableEvent::~ReadableEvent() = default;

 bool ReadableEvent::ShouldWait(const Thread* thread) const {
-    return !signaled;
+    return !is_signaled;
 }

 void ReadableEvent::Acquire(Thread* thread) {
-    ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
+    ASSERT_MSG(IsSignaled(), "object unavailable!");
 }

 void ReadableEvent::Signal() {
-    if (!signaled) {
-        signaled = true;
-        WakeupAllWaitingThreads();
+    if (!is_signaled) {
+        is_signaled = true;
+        SynchronizationObject::Signal();
    };
 }

 void ReadableEvent::Clear() {
-    signaled = false;
+    is_signaled = false;
 }

 ResultCode ReadableEvent::Reset() {
-    if (!signaled) {
+    if (!is_signaled) {
        return ERR_INVALID_STATE;
    }

@@ -5,7 +5,7 @@
 #pragma once

 #include "core/hle/kernel/object.h"
-#include "core/hle/kernel/wait_object.h"
+#include "core/hle/kernel/synchronization_object.h"

 union ResultCode;

@@ -14,7 +14,7 @@ namespace Kernel {
 class KernelCore;
 class WritableEvent;

-class ReadableEvent final : public WaitObject {
+class ReadableEvent final : public SynchronizationObject {
    friend class WritableEvent;

 public:
@@ -46,13 +46,11 @@ public:
    ///      then ERR_INVALID_STATE will be returned.
    ResultCode Reset();

+    void Signal() override;
+
 private:
    explicit ReadableEvent(KernelCore& kernel);

-    void Signal();
-
-    bool signaled{};
-
    std::string name; ///< Name of event (optional)
 };

@@ -18,10 +18,11 @@
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
+#include "core/hle/kernel/time_manager.h"

 namespace Kernel {

-GlobalScheduler::GlobalScheduler(Core::System& system) : system{system} {}
+GlobalScheduler::GlobalScheduler(KernelCore& kernel) : kernel{kernel} {}

 GlobalScheduler::~GlobalScheduler() = default;

@@ -35,7 +36,7 @@ void GlobalScheduler::RemoveThread(std::shared_ptr<Thread> thread) {
 }

 void GlobalScheduler::UnloadThread(std::size_t core) {
-    Scheduler& sched = system.Scheduler(core);
+    Scheduler& sched = kernel.Scheduler(core);
    sched.UnloadThread();
 }

@@ -50,7 +51,7 @@ void GlobalScheduler::SelectThread(std::size_t core) {
        sched.is_context_switch_pending = sched.selected_thread != sched.current_thread;
        std::atomic_thread_fence(std::memory_order_seq_cst);
    };
-    Scheduler& sched = system.Scheduler(core);
+    Scheduler& sched = kernel.Scheduler(core);
    Thread* current_thread = nullptr;
    // Step 1: Get top thread in schedule queue.
    current_thread = scheduled_queue[core].empty() ? nullptr : scheduled_queue[core].front();
@@ -124,8 +125,8 @@ bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) {
               "Thread yielding without being in front");
    scheduled_queue[core_id].yield(priority);

-    std::array<Thread*, NUM_CPU_CORES> current_threads;
-    for (u32 i = 0; i < NUM_CPU_CORES; i++) {
+    std::array<Thread*, Core::Hardware::NUM_CPU_CORES> current_threads;
+    for (std::size_t i = 0; i < current_threads.size(); i++) {
        current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front();
    }

@@ -177,8 +178,8 @@ bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread
    // function...
    if (scheduled_queue[core_id].empty()) {
        // Here, "current_threads" is calculated after the ""yield"", unlike yield -1
-        std::array<Thread*, NUM_CPU_CORES> current_threads;
-        for (u32 i = 0; i < NUM_CPU_CORES; i++) {
+        std::array<Thread*, Core::Hardware::NUM_CPU_CORES> current_threads;
+        for (std::size_t i = 0; i < current_threads.size(); i++) {
            current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front();
        }
        for (auto& thread : suggested_queue[core_id]) {
@@ -208,7 +209,7 @@ bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread
 }

 void GlobalScheduler::PreemptThreads() {
-    for (std::size_t core_id = 0; core_id < NUM_CPU_CORES; core_id++) {
+    for (std::size_t core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
        const u32 priority = preemption_priorities[core_id];

        if (scheduled_queue[core_id].size(priority) > 0) {
@@ -349,13 +350,39 @@ bool GlobalScheduler::AskForReselectionOrMarkRedundant(Thread* current_thread,
 }

 void GlobalScheduler::Shutdown() {
-    for (std::size_t core = 0; core < NUM_CPU_CORES; core++) {
+    for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
        scheduled_queue[core].clear();
        suggested_queue[core].clear();
    }
    thread_list.clear();
 }

+void GlobalScheduler::Lock() {
+    Core::EmuThreadHandle current_thread = kernel.GetCurrentEmuThreadID();
+    if (current_thread == current_owner) {
+        ++scope_lock;
+    } else {
+        inner_lock.lock();
+        current_owner = current_thread;
+        ASSERT(current_owner != Core::EmuThreadHandle::InvalidHandle());
+        scope_lock = 1;
+    }
+}
+
+void GlobalScheduler::Unlock() {
+    if (--scope_lock != 0) {
+        ASSERT(scope_lock > 0);
+        return;
+    }
+    for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
+        SelectThread(i);
+    }
+    current_owner = Core::EmuThreadHandle::InvalidHandle();
+    scope_lock = 1;
+    inner_lock.unlock();
+    // TODO(Blinkhawk): Setup the interrupts and change context on current core.
+}
+
 Scheduler::Scheduler(Core::System& system, Core::ARM_Interface& cpu_core, std::size_t core_id)
    : system(system), cpu_core(cpu_core), core_id(core_id) {}

@@ -485,4 +512,27 @@ void Scheduler::Shutdown() {
    selected_thread = nullptr;
 }

+SchedulerLock::SchedulerLock(KernelCore& kernel) : kernel{kernel} {
+    kernel.GlobalScheduler().Lock();
+}
+
+SchedulerLock::~SchedulerLock() {
+    kernel.GlobalScheduler().Unlock();
+}
+
+SchedulerLockAndSleep::SchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle,
+                                             Thread* time_task, s64 nanoseconds)
+    : SchedulerLock{kernel}, event_handle{event_handle}, time_task{time_task}, nanoseconds{
+                                                                                   nanoseconds} {
+    event_handle = InvalidHandle;
+}
+
+SchedulerLockAndSleep::~SchedulerLockAndSleep() {
+    if (sleep_cancelled) {
+        return;
+    }
+    auto& time_manager = kernel.TimeManager();
+    time_manager.ScheduleTimeEvent(event_handle, time_task, nanoseconds);
+}
+
 } // namespace Kernel
@@ -6,10 +6,12 @@

 #include <atomic>
 #include <memory>
+#include <mutex>
 #include <vector>

 #include "common/common_types.h"
 #include "common/multi_level_queue.h"
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/thread.h"

 namespace Core {
@@ -19,13 +21,13 @@ class System;

 namespace Kernel {

+class KernelCore;
 class Process;
+class SchedulerLock;

 class GlobalScheduler final {
 public:
-    static constexpr u32 NUM_CPU_CORES = 4;
-
-    explicit GlobalScheduler(Core::System& system);
+    explicit GlobalScheduler(KernelCore& kernel);
    ~GlobalScheduler();

    /// Adds a new thread to the scheduler
@@ -125,7 +127,7 @@ public:
    void PreemptThreads();

    u32 CpuCoresCount() const {
-        return NUM_CPU_CORES;
+        return Core::Hardware::NUM_CPU_CORES;
    }

    void SetReselectionPending() {
@@ -139,6 +141,14 @@ public:
    void Shutdown();

 private:
+    friend class SchedulerLock;
+
+    /// Lock the scheduler to the current thread.
+    void Lock();
+
+    /// Unlocks the scheduler, reselects threads, interrupts cores for rescheduling
+    /// and reschedules current core if needed.
+    void Unlock();
    /**
     * Transfers a thread into an specific core. If the destination_core is -1
     * it will be unscheduled from its source code and added into its suggested
@@ -149,17 +159,24 @@ private:
    bool AskForReselectionOrMarkRedundant(Thread* current_thread, const Thread* winner);

    static constexpr u32 min_regular_priority = 2;
-    std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, NUM_CPU_CORES> scheduled_queue;
-    std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, NUM_CPU_CORES> suggested_queue;
+    std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, Core::Hardware::NUM_CPU_CORES>
+        scheduled_queue;
+    std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, Core::Hardware::NUM_CPU_CORES>
+        suggested_queue;
    std::atomic<bool> is_reselection_pending{false};

    // The priority levels at which the global scheduler preempts threads every 10 ms. They are
    // ordered from Core 0 to Core 3.
-    std::array<u32, NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62};
+    std::array<u32, Core::Hardware::NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62};
+
+    /// Scheduler lock mechanisms.
+    std::mutex inner_lock{}; // TODO(Blinkhawk): Replace for a SpinLock
+    std::atomic<s64> scope_lock{};
+    Core::EmuThreadHandle current_owner{Core::EmuThreadHandle::InvalidHandle()};

    /// Lists all thread ids that aren't deleted/etc.
    std::vector<std::shared_ptr<Thread>> thread_list;
-    Core::System& system;
+    KernelCore& kernel;
 };

 class Scheduler final {
@@ -226,4 +243,30 @@ private:
    bool is_context_switch_pending = false;
 };

+class SchedulerLock {
+public:
+    explicit SchedulerLock(KernelCore& kernel);
+    ~SchedulerLock();
+
+protected:
+    KernelCore& kernel;
+};
+
+class SchedulerLockAndSleep : public SchedulerLock {
+public:
+    explicit SchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle, Thread* time_task,
+                                   s64 nanoseconds);
+    ~SchedulerLockAndSleep();
+
+    void CancelSleep() {
+        sleep_cancelled = true;
+    }
+
+private:
+    Handle& event_handle;
+    Thread* time_task;
+    s64 nanoseconds;
+    bool sleep_cancelled{};
+};
+
 } // namespace Kernel
@@ -13,7 +13,7 @@

 namespace Kernel {

-ServerPort::ServerPort(KernelCore& kernel) : WaitObject{kernel} {}
+ServerPort::ServerPort(KernelCore& kernel) : SynchronizationObject{kernel} {}
 ServerPort::~ServerPort() = default;

 ResultVal<std::shared_ptr<ServerSession>> ServerPort::Accept() {
@@ -39,6 +39,10 @@ void ServerPort::Acquire(Thread* thread) {
    ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
 }

+bool ServerPort::IsSignaled() const {
+    return !pending_sessions.empty();
+}
+
 ServerPort::PortPair ServerPort::CreatePortPair(KernelCore& kernel, u32 max_sessions,
                                                std::string name) {
    std::shared_ptr<ServerPort> server_port = std::make_shared<ServerPort>(kernel);
@@ -10,7 +10,7 @@
 #include <vector>
 #include "common/common_types.h"
 #include "core/hle/kernel/object.h"
-#include "core/hle/kernel/wait_object.h"
+#include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/result.h"

 namespace Kernel {
@@ -20,7 +20,7 @@ class KernelCore;
 class ServerSession;
 class SessionRequestHandler;

-class ServerPort final : public WaitObject {
+class ServerPort final : public SynchronizationObject {
 public:
    explicit ServerPort(KernelCore& kernel);
    ~ServerPort() override;
@@ -82,6 +82,8 @@ public:
    bool ShouldWait(const Thread* thread) const override;
    void Acquire(Thread* thread) override;

+    bool IsSignaled() const override;
+
 private:
    /// ServerSessions waiting to be accepted by the port
    std::vector<std::shared_ptr<ServerSession>> pending_sessions;
@@ -24,7 +24,7 @@

 namespace Kernel {

-ServerSession::ServerSession(KernelCore& kernel) : WaitObject{kernel} {}
+ServerSession::ServerSession(KernelCore& kernel) : SynchronizationObject{kernel} {}
 ServerSession::~ServerSession() = default;

 ResultVal<std::shared_ptr<ServerSession>> ServerSession::Create(KernelCore& kernel,
@@ -50,6 +50,16 @@ bool ServerSession::ShouldWait(const Thread* thread) const {
    return pending_requesting_threads.empty() || currently_handling != nullptr;
 }

+bool ServerSession::IsSignaled() const {
+    // Closed sessions should never wait, an error will be returned from svcReplyAndReceive.
+    if (!parent->Client()) {
+        return true;
+    }
+
+    // Wait if we have no pending requests, or if we're currently handling a request.
+    return !pending_requesting_threads.empty() && currently_handling == nullptr;
+}
+
 void ServerSession::Acquire(Thread* thread) {
    ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
    // We are now handling a request, pop it from the stack.
@@ -10,7 +10,7 @@
 #include <vector>

 #include "common/threadsafe_queue.h"
-#include "core/hle/kernel/wait_object.h"
+#include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/result.h"

 namespace Memory {
@@ -41,7 +41,7 @@ class Thread;
 * After the server replies to the request, the response is marshalled back to the caller's
 * TLS buffer and control is transferred back to it.
 */
-class ServerSession final : public WaitObject {
+class ServerSession final : public SynchronizationObject {
 public:
    explicit ServerSession(KernelCore& kernel);
    ~ServerSession() override;
@@ -73,6 +73,8 @@ public:
        return parent.get();
    }

+    bool IsSignaled() const override;
+
    /**
     * Sets the HLE handler for the session. This handler will be called to service IPC requests
     * instead of the regular IPC machinery. (The regular IPC machinery is currently not
@@ -9,7 +9,7 @@

 namespace Kernel {

-Session::Session(KernelCore& kernel) : WaitObject{kernel} {}
+Session::Session(KernelCore& kernel) : SynchronizationObject{kernel} {}
 Session::~Session() = default;

 Session::SessionPair Session::Create(KernelCore& kernel, std::string name) {
@@ -29,6 +29,11 @@ bool Session::ShouldWait(const Thread* thread) const {
    return {};
 }

+bool Session::IsSignaled() const {
+    UNIMPLEMENTED();
+    return true;
+}
+
 void Session::Acquire(Thread* thread) {
    UNIMPLEMENTED();
 }
@@ -8,7 +8,7 @@
 #include <string>
 #include <utility>

-#include "core/hle/kernel/wait_object.h"
+#include "core/hle/kernel/synchronization_object.h"

 namespace Kernel {

@@ -19,7 +19,7 @@ class ServerSession;
 * Parent structure to link the client and server endpoints of a session with their associated
 * client port.
 */
-class Session final : public WaitObject {
+class Session final : public SynchronizationObject {
 public:
    explicit Session(KernelCore& kernel);
    ~Session() override;
@@ -39,6 +39,8 @@ public:

    bool ShouldWait(const Thread* thread) const override;

+    bool IsSignaled() const override;
+
    void Acquire(Thread* thread) override;

    std::shared_ptr<ClientSession> Client() {
@@ -32,6 +32,7 @@
 #include "core/hle/kernel/shared_memory.h"
 #include "core/hle/kernel/svc.h"
 #include "core/hle/kernel/svc_wrap.h"
+#include "core/hle/kernel/synchronization.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/kernel/transfer_memory.h"
 #include "core/hle/kernel/writable_event.h"
@@ -433,22 +434,6 @@ static ResultCode GetProcessId(Core::System& system, u64* process_id, Handle han
    return ERR_INVALID_HANDLE;
 }

-/// Default thread wakeup callback for WaitSynchronization
-static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                                        std::shared_ptr<WaitObject> object, std::size_t index) {
-    ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch);
-
-    if (reason == ThreadWakeupReason::Timeout) {
-        thread->SetWaitSynchronizationResult(RESULT_TIMEOUT);
-        return true;
-    }
-
-    ASSERT(reason == ThreadWakeupReason::Signal);
-    thread->SetWaitSynchronizationResult(RESULT_SUCCESS);
-    thread->SetWaitSynchronizationOutput(static_cast<u32>(index));
-    return true;
-};
-
 /// Wait for the given handles to synchronize, timeout after the specified nanoseconds
 static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr handles_address,
                                      u64 handle_count, s64 nano_seconds) {
@@ -472,14 +457,14 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr
    }

    auto* const thread = system.CurrentScheduler().GetCurrentThread();
-
-    using ObjectPtr = Thread::ThreadWaitObjects::value_type;
-    Thread::ThreadWaitObjects objects(handle_count);
-    const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
+    auto& kernel = system.Kernel();
+    using ObjectPtr = Thread::ThreadSynchronizationObjects::value_type;
+    Thread::ThreadSynchronizationObjects objects(handle_count);
+    const auto& handle_table = kernel.CurrentProcess()->GetHandleTable();

    for (u64 i = 0; i < handle_count; ++i) {
        const Handle handle = memory.Read32(handles_address + i * sizeof(Handle));
-        const auto object = handle_table.Get<WaitObject>(handle);
+        const auto object = handle_table.Get<SynchronizationObject>(handle);

        if (object == nullptr) {
            LOG_ERROR(Kernel_SVC, "Object is a nullptr");
@@ -488,47 +473,10 @@ static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr

        objects[i] = object;
    }
-
-    // Find the first object that is acquirable in the provided list of objects
-    auto itr = std::find_if(objects.begin(), objects.end(), [thread](const ObjectPtr& object) {
-        return !object->ShouldWait(thread);
-    });
-
-    if (itr != objects.end()) {
-        // We found a ready object, acquire it and set the result value
-        WaitObject* object = itr->get();
-        object->Acquire(thread);
-        *index = static_cast<s32>(std::distance(objects.begin(), itr));
-        return RESULT_SUCCESS;
-    }
-
-    // No objects were ready to be acquired, prepare to suspend the thread.
-
-    // If a timeout value of 0 was provided, just return the Timeout error code instead of
-    // suspending the thread.
-    if (nano_seconds == 0) {
-        return RESULT_TIMEOUT;
-    }
-
-    if (thread->IsSyncCancelled()) {
-        thread->SetSyncCancelled(false);
-        return ERR_SYNCHRONIZATION_CANCELED;
-    }
-
-    for (auto& object : objects) {
-        object->AddWaitingThread(SharedFrom(thread));
-    }
-
-    thread->SetWaitObjects(std::move(objects));
-    thread->SetStatus(ThreadStatus::WaitSynch);
-
-    // Create an event to wake the thread up after the specified nanosecond delay has passed
-    thread->WakeAfterDelay(nano_seconds);
-    thread->SetWakeupCallback(DefaultThreadWakeupCallback);
-
-    system.PrepareReschedule(thread->GetProcessorID());
-
-    return RESULT_TIMEOUT;
+    auto& synchronization = kernel.Synchronization();
+    const auto [result, handle_result] = synchronization.WaitFor(objects, nano_seconds);
+    *index = handle_result;
+    return result;
 }

 /// Resumes a thread waiting on WaitSynchronization
@@ -0,0 +1,87 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/core.h"
+#include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/scheduler.h"
+#include "core/hle/kernel/synchronization.h"
+#include "core/hle/kernel/synchronization_object.h"
+#include "core/hle/kernel/thread.h"
+
+namespace Kernel {
+
+/// Default thread wakeup callback for WaitSynchronization
+static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
+                                        std::shared_ptr<SynchronizationObject> object,
+                                        std::size_t index) {
+    ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch);
+
+    if (reason == ThreadWakeupReason::Timeout) {
+        thread->SetWaitSynchronizationResult(RESULT_TIMEOUT);
+        return true;
+    }
+
+    ASSERT(reason == ThreadWakeupReason::Signal);
+    thread->SetWaitSynchronizationResult(RESULT_SUCCESS);
+    thread->SetWaitSynchronizationOutput(static_cast<u32>(index));
+    return true;
+}
+
+Synchronization::Synchronization(Core::System& system) : system{system} {}
+
+void Synchronization::SignalObject(SynchronizationObject& obj) const {
+    if (obj.IsSignaled()) {
+        obj.WakeupAllWaitingThreads();
+    }
+}
+
+std::pair<ResultCode, Handle> Synchronization::WaitFor(
+    std::vector<std::shared_ptr<SynchronizationObject>>& sync_objects, s64 nano_seconds) {
+    auto* const thread = system.CurrentScheduler().GetCurrentThread();
+    // Find the first object that is acquirable in the provided list of objects
+    const auto itr = std::find_if(sync_objects.begin(), sync_objects.end(),
+                                  [thread](const std::shared_ptr<SynchronizationObject>& object) {
+                                      return object->IsSignaled();
+                                  });
+
+    if (itr != sync_objects.end()) {
+        // We found a ready object, acquire it and set the result value
+        SynchronizationObject* object = itr->get();
+        object->Acquire(thread);
+        const u32 index = static_cast<s32>(std::distance(sync_objects.begin(), itr));
+        return {RESULT_SUCCESS, index};
+    }
+
+    // No objects were ready to be acquired, prepare to suspend the thread.
+
+    // If a timeout value of 0 was provided, just return the Timeout error code instead of
+    // suspending the thread.
+    if (nano_seconds == 0) {
+        return {RESULT_TIMEOUT, InvalidHandle};
+    }
+
+    if (thread->IsSyncCancelled()) {
+        thread->SetSyncCancelled(false);
+        return {ERR_SYNCHRONIZATION_CANCELED, InvalidHandle};
+    }
+
+    for (auto& object : sync_objects) {
+        object->AddWaitingThread(SharedFrom(thread));
+    }
+
+    thread->SetSynchronizationObjects(std::move(sync_objects));
+    thread->SetStatus(ThreadStatus::WaitSynch);
+
+    // Create an event to wake the thread up after the specified nanosecond delay has passed
+    thread->WakeAfterDelay(nano_seconds);
+    thread->SetWakeupCallback(DefaultThreadWakeupCallback);
+
+    system.PrepareReschedule(thread->GetProcessorID());
+
+    return {RESULT_TIMEOUT, InvalidHandle};
+}
+
+} // namespace Kernel
@@ -0,0 +1,44 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <utility>
+#include <vector>
+
+#include "core/hle/kernel/object.h"
+#include "core/hle/result.h"
+
+namespace Core {
+class System;
+} // namespace Core
+
+namespace Kernel {
+
+class SynchronizationObject;
+
+/**
+ * The 'Synchronization' class is an interface for handling synchronization methods
+ * used by Synchronization objects and synchronization SVCs. This centralizes processing of
+ * such
+ */
+class Synchronization {
+public:
+    explicit Synchronization(Core::System& system);
+
+    /// Signals a synchronization object, waking up all its waiting threads
+    void SignalObject(SynchronizationObject& obj) const;
+
+    /// Tries to see if waiting for any of the sync_objects is necessary, if not
+    /// it returns Success and the handle index of the signaled sync object. In
+    /// case not, the current thread will be locked and wait for nano_seconds or
+    /// for a synchronization object to signal.
+    std::pair<ResultCode, Handle> WaitFor(
+        std::vector<std::shared_ptr<SynchronizationObject>>& sync_objects, s64 nano_seconds);
+
+private:
+    Core::System& system;
+};
+} // namespace Kernel
@@ -10,20 +10,26 @@
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/kernel/synchronization.h"
+#include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/kernel/thread.h"

 namespace Kernel {

-WaitObject::WaitObject(KernelCore& kernel) : Object{kernel} {}
-WaitObject::~WaitObject() = default;
+SynchronizationObject::SynchronizationObject(KernelCore& kernel) : Object{kernel} {}
+SynchronizationObject::~SynchronizationObject() = default;

-void WaitObject::AddWaitingThread(std::shared_ptr<Thread> thread) {
+void SynchronizationObject::Signal() {
+    kernel.Synchronization().SignalObject(*this);
+}
+
+void SynchronizationObject::AddWaitingThread(std::shared_ptr<Thread> thread) {
    auto itr = std::find(waiting_threads.begin(), waiting_threads.end(), thread);
    if (itr == waiting_threads.end())
        waiting_threads.push_back(std::move(thread));
 }

-void WaitObject::RemoveWaitingThread(std::shared_ptr<Thread> thread) {
+void SynchronizationObject::RemoveWaitingThread(std::shared_ptr<Thread> thread) {
    auto itr = std::find(waiting_threads.begin(), waiting_threads.end(), thread);
    // If a thread passed multiple handles to the same object,
    // the kernel might attempt to remove the thread from the object's
@@ -32,7 +38,7 @@ void WaitObject::RemoveWaitingThread(std::shared_ptr<Thread> thread) {
        waiting_threads.erase(itr);
 }

-std::shared_ptr<Thread> WaitObject::GetHighestPriorityReadyThread() const {
+std::shared_ptr<Thread> SynchronizationObject::GetHighestPriorityReadyThread() const {
    Thread* candidate = nullptr;
    u32 candidate_priority = THREADPRIO_LOWEST + 1;

@@ -57,7 +63,7 @@ std::shared_ptr<Thread> WaitObject::GetHighestPriorityReadyThread() const {
    return SharedFrom(candidate);
 }

-void WaitObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) {
+void SynchronizationObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) {
    ASSERT(!ShouldWait(thread.get()));

    if (!thread) {
@@ -65,7 +71,7 @@ void WaitObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) {
    }

    if (thread->IsSleepingOnWait()) {
-        for (const auto& object : thread->GetWaitObjects()) {
+        for (const auto& object : thread->GetSynchronizationObjects()) {
            ASSERT(!object->ShouldWait(thread.get()));
            object->Acquire(thread.get());
        }
@@ -73,9 +79,9 @@ void WaitObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) {
        Acquire(thread.get());
    }

-    const std::size_t index = thread->GetWaitObjectIndex(SharedFrom(this));
+    const std::size_t index = thread->GetSynchronizationObjectIndex(SharedFrom(this));

-    thread->ClearWaitObjects();
+    thread->ClearSynchronizationObjects();

    thread->CancelWakeupTimer();

@@ -90,13 +96,13 @@ void WaitObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) {
    }
 }

-void WaitObject::WakeupAllWaitingThreads() {
+void SynchronizationObject::WakeupAllWaitingThreads() {
    while (auto thread = GetHighestPriorityReadyThread()) {
        WakeupWaitingThread(thread);
    }
 }

-const std::vector<std::shared_ptr<Thread>>& WaitObject::GetWaitingThreads() const {
+const std::vector<std::shared_ptr<Thread>>& SynchronizationObject::GetWaitingThreads() const {
    return waiting_threads;
 }

@@ -15,10 +15,10 @@ class KernelCore;
 class Thread;

 /// Class that represents a Kernel object that a thread can be waiting on
-class WaitObject : public Object {
+class SynchronizationObject : public Object {
 public:
-    explicit WaitObject(KernelCore& kernel);
-    ~WaitObject() override;
+    explicit SynchronizationObject(KernelCore& kernel);
+    ~SynchronizationObject() override;

    /**
     * Check if the specified thread should wait until the object is available
@@ -30,6 +30,13 @@ public:
    /// Acquire/lock the object for the specified thread if it is available
    virtual void Acquire(Thread* thread) = 0;

+    /// Signal this object
+    virtual void Signal();
+
+    virtual bool IsSignaled() const {
+        return is_signaled;
+    }
+
    /**
     * Add a thread to wait on this object
     * @param thread Pointer to thread to add
@@ -60,16 +67,20 @@ public:
    /// Get a const reference to the waiting threads list for debug use
    const std::vector<std::shared_ptr<Thread>>& GetWaitingThreads() const;

+protected:
+    bool is_signaled{}; // Tells if this sync object is signalled;
+
 private:
    /// Threads waiting for this object to become available
    std::vector<std::shared_ptr<Thread>> waiting_threads;
 };

-// Specialization of DynamicObjectCast for WaitObjects
+// Specialization of DynamicObjectCast for SynchronizationObjects
 template <>
-inline std::shared_ptr<WaitObject> DynamicObjectCast<WaitObject>(std::shared_ptr<Object> object) {
+inline std::shared_ptr<SynchronizationObject> DynamicObjectCast<SynchronizationObject>(
+    std::shared_ptr<Object> object) {
    if (object != nullptr && object->IsWaitable()) {
-        return std::static_pointer_cast<WaitObject>(object);
+        return std::static_pointer_cast<SynchronizationObject>(object);
    }
    return nullptr;
 }
@@ -15,6 +15,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/kernel.h"
@@ -31,21 +32,25 @@ bool Thread::ShouldWait(const Thread* thread) const {
    return status != ThreadStatus::Dead;
 }

+bool Thread::IsSignaled() const {
+    return status == ThreadStatus::Dead;
+}
+
 void Thread::Acquire(Thread* thread) {
    ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
 }

-Thread::Thread(KernelCore& kernel) : WaitObject{kernel} {}
+Thread::Thread(KernelCore& kernel) : SynchronizationObject{kernel} {}
 Thread::~Thread() = default;

 void Thread::Stop() {
    // Cancel any outstanding wakeup events for this thread
    Core::System::GetInstance().CoreTiming().UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(),
-                                                             callback_handle);
-    kernel.ThreadWakeupCallbackHandleTable().Close(callback_handle);
-    callback_handle = 0;
+                                                             global_handle);
+    kernel.GlobalHandleTable().Close(global_handle);
+    global_handle = 0;
    SetStatus(ThreadStatus::Dead);
-    WakeupAllWaitingThreads();
+    Signal();

    // Clean up any dangling references in objects that this thread was waiting for
    for (auto& wait_object : wait_objects) {
@@ -68,12 +73,12 @@ void Thread::WakeAfterDelay(s64 nanoseconds) {
    // thread-safe version of ScheduleEvent.
    const s64 cycles = Core::Timing::nsToCycles(std::chrono::nanoseconds{nanoseconds});
    Core::System::GetInstance().CoreTiming().ScheduleEvent(
-        cycles, kernel.ThreadWakeupCallbackEventType(), callback_handle);
+        cycles, kernel.ThreadWakeupCallbackEventType(), global_handle);
 }

 void Thread::CancelWakeupTimer() {
    Core::System::GetInstance().CoreTiming().UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(),
-                                                             callback_handle);
+                                                             global_handle);
 }

 void Thread::ResumeFromWait() {
@@ -185,7 +190,7 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::strin
    thread->condvar_wait_address = 0;
    thread->wait_handle = 0;
    thread->name = std::move(name);
-    thread->callback_handle = kernel.ThreadWakeupCallbackHandleTable().Create(thread).Unwrap();
+    thread->global_handle = kernel.GlobalHandleTable().Create(thread).Unwrap();
    thread->owner_process = &owner_process;
    auto& scheduler = kernel.GlobalScheduler();
    scheduler.AddThread(thread);
@@ -215,7 +220,7 @@ void Thread::SetWaitSynchronizationOutput(s32 output) {
    context.cpu_registers[1] = output;
 }

-s32 Thread::GetWaitObjectIndex(std::shared_ptr<WaitObject> object) const {
+s32 Thread::GetSynchronizationObjectIndex(std::shared_ptr<SynchronizationObject> object) const {
    ASSERT_MSG(!wait_objects.empty(), "Thread is not waiting for anything");
    const auto match = std::find(wait_objects.rbegin(), wait_objects.rend(), object);
    return static_cast<s32>(std::distance(match, wait_objects.rend()) - 1);
@@ -336,14 +341,16 @@ void Thread::ChangeCore(u32 core, u64 mask) {
    SetCoreAndAffinityMask(core, mask);
 }

-bool Thread::AllWaitObjectsReady() const {
-    return std::none_of(
-        wait_objects.begin(), wait_objects.end(),
-        [this](const std::shared_ptr<WaitObject>& object) { return object->ShouldWait(this); });
+bool Thread::AllSynchronizationObjectsReady() const {
+    return std::none_of(wait_objects.begin(), wait_objects.end(),
+                        [this](const std::shared_ptr<SynchronizationObject>& object) {
+                            return object->ShouldWait(this);
+                        });
 }

 bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                                  std::shared_ptr<WaitObject> object, std::size_t index) {
+                                  std::shared_ptr<SynchronizationObject> object,
+                                  std::size_t index) {
    ASSERT(wakeup_callback);
    return wakeup_callback(reason, std::move(thread), std::move(object), index);
 }
@@ -425,7 +432,7 @@ ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
            const s32 old_core = processor_id;
            if (processor_id >= 0 && ((affinity_mask >> processor_id) & 1) == 0) {
                if (static_cast<s32>(ideal_core) < 0) {
-                    processor_id = HighestSetCore(affinity_mask, GlobalScheduler::NUM_CPU_CORES);
+                    processor_id = HighestSetCore(affinity_mask, Core::Hardware::NUM_CPU_CORES);
                } else {
                    processor_id = ideal_core;
                }
@@ -449,7 +456,7 @@ void Thread::AdjustSchedulingOnStatus(u32 old_flags) {
            scheduler.Unschedule(current_priority, static_cast<u32>(processor_id), this);
        }

-        for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+        for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
            if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
                scheduler.Unsuggest(current_priority, core, this);
            }
@@ -460,7 +467,7 @@ void Thread::AdjustSchedulingOnStatus(u32 old_flags) {
            scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this);
        }

-        for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+        for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
            if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
                scheduler.Suggest(current_priority, core, this);
            }
@@ -474,12 +481,12 @@ void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
    if (GetSchedulingStatus() != ThreadSchedStatus::Runnable) {
        return;
    }
-    auto& scheduler = Core::System::GetInstance().GlobalScheduler();
+    auto& scheduler = kernel.GlobalScheduler();
    if (processor_id >= 0) {
        scheduler.Unschedule(old_priority, static_cast<u32>(processor_id), this);
    }

-    for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
        if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
            scheduler.Unsuggest(old_priority, core, this);
        }
@@ -496,7 +503,7 @@ void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
        }
    }

-    for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
        if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
            scheduler.Suggest(current_priority, core, this);
        }
@@ -506,13 +513,13 @@ void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
 }

 void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) {
-    auto& scheduler = Core::System::GetInstance().GlobalScheduler();
+    auto& scheduler = kernel.GlobalScheduler();
    if (GetSchedulingStatus() != ThreadSchedStatus::Runnable ||
        current_priority >= THREADPRIO_COUNT) {
        return;
    }

-    for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
        if (((old_affinity_mask >> core) & 1) != 0) {
            if (core == static_cast<u32>(old_core)) {
                scheduler.Unschedule(current_priority, core, this);
@@ -522,7 +529,7 @@ void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) {
        }
    }

-    for (u32 core = 0; core < GlobalScheduler::NUM_CPU_CORES; core++) {
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
        if (((affinity_mask >> core) & 1) != 0) {
            if (core == static_cast<u32>(processor_id)) {
                scheduler.Schedule(current_priority, core, this);
@@ -11,7 +11,7 @@
 #include "common/common_types.h"
 #include "core/arm/arm_interface.h"
 #include "core/hle/kernel/object.h"
-#include "core/hle/kernel/wait_object.h"
+#include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/result.h"

 namespace Kernel {
@@ -95,7 +95,7 @@ enum class ThreadSchedMasks : u32 {
    ForcePauseMask = 0x0070,
 };

-class Thread final : public WaitObject {
+class Thread final : public SynchronizationObject {
 public:
    explicit Thread(KernelCore& kernel);
    ~Thread() override;
@@ -104,11 +104,11 @@ public:

    using ThreadContext = Core::ARM_Interface::ThreadContext;

-    using ThreadWaitObjects = std::vector<std::shared_ptr<WaitObject>>;
+    using ThreadSynchronizationObjects = std::vector<std::shared_ptr<SynchronizationObject>>;

    using WakeupCallback =
        std::function<bool(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                           std::shared_ptr<WaitObject> object, std::size_t index)>;
+                           std::shared_ptr<SynchronizationObject> object, std::size_t index)>;

    /**
     * Creates and returns a new thread. The new thread is immediately scheduled
@@ -146,6 +146,7 @@ public:

    bool ShouldWait(const Thread* thread) const override;
    void Acquire(Thread* thread) override;
+    bool IsSignaled() const override;

    /**
     * Gets the thread's current priority
@@ -233,7 +234,7 @@ public:
     *
     * @param object Object to query the index of.
     */
-    s32 GetWaitObjectIndex(std::shared_ptr<WaitObject> object) const;
+    s32 GetSynchronizationObjectIndex(std::shared_ptr<SynchronizationObject> object) const;

    /**
     * Stops a thread, invalidating it from further use
@@ -314,15 +315,15 @@ public:
        return owner_process;
    }

-    const ThreadWaitObjects& GetWaitObjects() const {
+    const ThreadSynchronizationObjects& GetSynchronizationObjects() const {
        return wait_objects;
    }

-    void SetWaitObjects(ThreadWaitObjects objects) {
+    void SetSynchronizationObjects(ThreadSynchronizationObjects objects) {
        wait_objects = std::move(objects);
    }

-    void ClearWaitObjects() {
+    void ClearSynchronizationObjects() {
        for (const auto& waiting_object : wait_objects) {
            waiting_object->RemoveWaitingThread(SharedFrom(this));
        }
@@ -330,7 +331,7 @@ public:
    }

    /// Determines whether all the objects this thread is waiting on are ready.
-    bool AllWaitObjectsReady() const;
+    bool AllSynchronizationObjectsReady() const;

    const MutexWaitingThreads& GetMutexWaitingThreads() const {
        return wait_mutex_threads;
@@ -395,7 +396,7 @@ public:
     *      will cause an assertion to trigger.
     */
    bool InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                              std::shared_ptr<WaitObject> object, std::size_t index);
+                              std::shared_ptr<SynchronizationObject> object, std::size_t index);

    u32 GetIdealCore() const {
        return ideal_core;
@@ -452,6 +453,10 @@ public:
        is_sync_cancelled = value;
    }

+    Handle GetGlobalHandle() const {
+        return global_handle;
+    }
+
 private:
    void SetSchedulingStatus(ThreadSchedStatus new_status);
    void SetCurrentPriority(u32 new_priority);
@@ -494,7 +499,7 @@ private:

    /// Objects that the thread is waiting on, in the same order as they were
    /// passed to WaitSynchronization.
-    ThreadWaitObjects wait_objects;
+    ThreadSynchronizationObjects wait_objects;

    /// List of threads that are waiting for a mutex that is held by this thread.
    MutexWaitingThreads wait_mutex_threads;
@@ -513,7 +518,7 @@ private:
    VAddr arb_wait_address{0};

    /// Handle used as userdata to reference this object when inserting into the CoreTiming queue.
-    Handle callback_handle = 0;
+    Handle global_handle = 0;

    /// Callback that will be invoked when the thread is resumed from a waiting state. If the thread
    /// was waiting via WaitSynchronization then the object will be the last object that became
@@ -0,0 +1,44 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
+#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
+
+namespace Kernel {
+
+TimeManager::TimeManager(Core::System& system) : system{system} {
+    time_manager_event_type = Core::Timing::CreateEvent(
+        "Kernel::TimeManagerCallback", [this](u64 thread_handle, [[maybe_unused]] s64 cycles_late) {
+            Handle proper_handle = static_cast<Handle>(thread_handle);
+            std::shared_ptr<Thread> thread =
+                this->system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
+            thread->ResumeFromWait();
+        });
+}
+
+void TimeManager::ScheduleTimeEvent(Handle& event_handle, Thread* timetask, s64 nanoseconds) {
+    if (nanoseconds > 0) {
+        ASSERT(timetask);
+        event_handle = timetask->GetGlobalHandle();
+        const s64 cycles = Core::Timing::nsToCycles(std::chrono::nanoseconds{nanoseconds});
+        system.CoreTiming().ScheduleEvent(cycles, time_manager_event_type, event_handle);
+    } else {
+        event_handle = InvalidHandle;
+    }
+}
+
+void TimeManager::UnscheduleTimeEvent(Handle event_handle) {
+    if (event_handle == InvalidHandle) {
+        return;
+    }
+    system.CoreTiming().UnscheduleEvent(time_manager_event_type, event_handle);
+}
+
+} // namespace Kernel
@@ -0,0 +1,43 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+
+#include "core/hle/kernel/object.h"
+
+namespace Core {
+class System;
+} // namespace Core
+
+namespace Core::Timing {
+struct EventType;
+} // namespace Core::Timing
+
+namespace Kernel {
+
+class Thread;
+
+/**
+ * The `TimeManager` takes care of scheduling time events on threads and executes their TimeUp
+ * method when the event is triggered.
+ */
+class TimeManager {
+public:
+    explicit TimeManager(Core::System& system);
+
+    /// Schedule a time event on `timetask` thread that will expire in 'nanoseconds'
+    /// returns a non-invalid handle in `event_handle` if correctly scheduled
+    void ScheduleTimeEvent(Handle& event_handle, Thread* timetask, s64 nanoseconds);
+
+    /// Unschedule an existing time event
+    void UnscheduleTimeEvent(Handle event_handle);
+
+private:
+    Core::System& system;
+    std::shared_ptr<Core::Timing::EventType> time_manager_event_type;
+};
+
+} // namespace Kernel
@@ -22,7 +22,6 @@ EventPair WritableEvent::CreateEventPair(KernelCore& kernel, std::string name) {
    writable_event->name = name + ":Writable";
    writable_event->readable = readable_event;
    readable_event->name = name + ":Readable";
-    readable_event->signaled = false;

    return {std::move(readable_event), std::move(writable_event)};
 }
@@ -40,7 +39,7 @@ void WritableEvent::Clear() {
 }

 bool WritableEvent::IsSignaled() const {
-    return readable->signaled;
+    return readable->IsSignaled();
 }

 } // namespace Kernel
@@ -607,7 +607,7 @@ ICommonStateGetter::ICommonStateGetter(Core::System& system,
        {40, nullptr, "GetCradleFwVersion"},
        {50, nullptr, "IsVrModeEnabled"},
        {51, nullptr, "SetVrModeEnabled"},
-        {52, nullptr, "SwitchLcdBacklight"},
+        {52, &ICommonStateGetter::SetLcdBacklighOffEnabled, "SetLcdBacklighOffEnabled"},
        {53, nullptr, "BeginVrModeEx"},
        {54, nullptr, "EndVrModeEx"},
        {55, nullptr, "IsInControllerFirmwareUpdateSection"},
@@ -636,7 +636,6 @@ void ICommonStateGetter::GetBootMode(Kernel::HLERequestContext& ctx) {

    IPC::ResponseBuilder rb{ctx, 3};
    rb.Push(RESULT_SUCCESS);
-
    rb.Push<u8>(static_cast<u8>(Service::PM::SystemBootMode::Normal)); // Normal boot mode
 }

@@ -660,6 +659,7 @@ void ICommonStateGetter::ReceiveMessage(Kernel::HLERequestContext& ctx) {
        rb.PushEnum<AppletMessageQueue::AppletMessage>(message);
        return;
    }
+
    rb.Push(RESULT_SUCCESS);
    rb.PushEnum<AppletMessageQueue::AppletMessage>(message);
 }
@@ -672,6 +672,17 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
    rb.Push(static_cast<u8>(FocusState::InFocus));
 }

+void ICommonStateGetter::SetLcdBacklighOffEnabled(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const auto is_lcd_backlight_off_enabled = rp.Pop<bool>();
+
+    LOG_WARNING(Service_AM, "(STUBBED) called. is_lcd_backlight_off_enabled={}",
+                is_lcd_backlight_off_enabled);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
 void ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent(Kernel::HLERequestContext& ctx) {
    LOG_DEBUG(Service_AM, "called");

@@ -847,7 +858,7 @@ private:
        LOG_DEBUG(Service_AM, "called");

        IPC::RequestParser rp{ctx};
-        applet->GetBroker().PushNormalDataFromGame(*rp.PopIpcInterface<IStorage>());
+        applet->GetBroker().PushNormalDataFromGame(rp.PopIpcInterface<IStorage>());

        IPC::ResponseBuilder rb{ctx, 2};
        rb.Push(RESULT_SUCCESS);
@@ -867,14 +878,14 @@ private:

        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
        rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<IStorage>(std::move(*storage));
+        rb.PushIpcInterface<IStorage>(std::move(storage));
    }

    void PushInteractiveInData(Kernel::HLERequestContext& ctx) {
        LOG_DEBUG(Service_AM, "called");

        IPC::RequestParser rp{ctx};
-        applet->GetBroker().PushInteractiveDataFromGame(*rp.PopIpcInterface<IStorage>());
+        applet->GetBroker().PushInteractiveDataFromGame(rp.PopIpcInterface<IStorage>());

        ASSERT(applet->IsInitialized());
        applet->ExecuteInteractive();
@@ -898,7 +909,7 @@ private:

        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
        rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<IStorage>(std::move(*storage));
+        rb.PushIpcInterface<IStorage>(std::move(storage));
    }

    void GetPopOutDataEvent(Kernel::HLERequestContext& ctx) {
@@ -182,6 +182,7 @@ private:
    void GetOperationMode(Kernel::HLERequestContext& ctx);
    void GetPerformanceMode(Kernel::HLERequestContext& ctx);
    void GetBootMode(Kernel::HLERequestContext& ctx);
+    void SetLcdBacklighOffEnabled(Kernel::HLERequestContext& ctx);
    void GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx);
    void SetCpuBoostMode(Kernel::HLERequestContext& ctx);

@@ -50,7 +50,7 @@ AppletDataBroker::RawChannelData AppletDataBroker::PeekDataToAppletForDebug() co
    return {std::move(out_normal), std::move(out_interactive)};
 }

-std::unique_ptr<IStorage> AppletDataBroker::PopNormalDataToGame() {
+std::shared_ptr<IStorage> AppletDataBroker::PopNormalDataToGame() {
    if (out_channel.empty())
        return nullptr;

@@ -60,7 +60,7 @@ std::unique_ptr<IStorage> AppletDataBroker::PopNormalDataToGame() {
    return out;
 }

-std::unique_ptr<IStorage> AppletDataBroker::PopNormalDataToApplet() {
+std::shared_ptr<IStorage> AppletDataBroker::PopNormalDataToApplet() {
    if (in_channel.empty())
        return nullptr;

@@ -69,7 +69,7 @@ std::unique_ptr<IStorage> AppletDataBroker::PopNormalDataToApplet() {
    return out;
 }

-std::unique_ptr<IStorage> AppletDataBroker::PopInteractiveDataToGame() {
+std::shared_ptr<IStorage> AppletDataBroker::PopInteractiveDataToGame() {
    if (out_interactive_channel.empty())
        return nullptr;

@@ -79,7 +79,7 @@ std::unique_ptr<IStorage> AppletDataBroker::PopInteractiveDataToGame() {
    return out;
 }

-std::unique_ptr<IStorage> AppletDataBroker::PopInteractiveDataToApplet() {
+std::shared_ptr<IStorage> AppletDataBroker::PopInteractiveDataToApplet() {
    if (in_interactive_channel.empty())
        return nullptr;

@@ -88,21 +88,21 @@ std::unique_ptr<IStorage> AppletDataBroker::PopInteractiveDataToApplet() {
    return out;
 }

-void AppletDataBroker::PushNormalDataFromGame(IStorage storage) {
-    in_channel.push_back(std::make_unique<IStorage>(storage));
+void AppletDataBroker::PushNormalDataFromGame(std::shared_ptr<IStorage>&& storage) {
+    in_channel.emplace_back(std::move(storage));
 }

-void AppletDataBroker::PushNormalDataFromApplet(IStorage storage) {
-    out_channel.push_back(std::make_unique<IStorage>(storage));
+void AppletDataBroker::PushNormalDataFromApplet(std::shared_ptr<IStorage>&& storage) {
+    out_channel.emplace_back(std::move(storage));
    pop_out_data_event.writable->Signal();
 }

-void AppletDataBroker::PushInteractiveDataFromGame(IStorage storage) {
-    in_interactive_channel.push_back(std::make_unique<IStorage>(storage));
+void AppletDataBroker::PushInteractiveDataFromGame(std::shared_ptr<IStorage>&& storage) {
+    in_interactive_channel.emplace_back(std::move(storage));
 }

-void AppletDataBroker::PushInteractiveDataFromApplet(IStorage storage) {
-    out_interactive_channel.push_back(std::make_unique<IStorage>(storage));
+void AppletDataBroker::PushInteractiveDataFromApplet(std::shared_ptr<IStorage>&& storage) {
+    out_interactive_channel.emplace_back(std::move(storage));
    pop_interactive_out_data_event.writable->Signal();
 }

@@ -72,17 +72,17 @@ public:
    // Retrieves but does not pop the data sent to applet.
    RawChannelData PeekDataToAppletForDebug() const;

-    std::unique_ptr<IStorage> PopNormalDataToGame();
-    std::unique_ptr<IStorage> PopNormalDataToApplet();
+    std::shared_ptr<IStorage> PopNormalDataToGame();
+    std::shared_ptr<IStorage> PopNormalDataToApplet();

-    std::unique_ptr<IStorage> PopInteractiveDataToGame();
-    std::unique_ptr<IStorage> PopInteractiveDataToApplet();
+    std::shared_ptr<IStorage> PopInteractiveDataToGame();
+    std::shared_ptr<IStorage> PopInteractiveDataToApplet();

-    void PushNormalDataFromGame(IStorage storage);
-    void PushNormalDataFromApplet(IStorage storage);
+    void PushNormalDataFromGame(std::shared_ptr<IStorage>&& storage);
+    void PushNormalDataFromApplet(std::shared_ptr<IStorage>&& storage);

-    void PushInteractiveDataFromGame(IStorage storage);
-    void PushInteractiveDataFromApplet(IStorage storage);
+    void PushInteractiveDataFromGame(std::shared_ptr<IStorage>&& storage);
+    void PushInteractiveDataFromApplet(std::shared_ptr<IStorage>&& storage);

    void SignalStateChanged() const;

@@ -94,16 +94,16 @@ private:
    // Queues are named from applet's perspective

    // PopNormalDataToApplet and PushNormalDataFromGame
-    std::deque<std::unique_ptr<IStorage>> in_channel;
+    std::deque<std::shared_ptr<IStorage>> in_channel;

    // PopNormalDataToGame and PushNormalDataFromApplet
-    std::deque<std::unique_ptr<IStorage>> out_channel;
+    std::deque<std::shared_ptr<IStorage>> out_channel;

    // PopInteractiveDataToApplet and PushInteractiveDataFromGame
-    std::deque<std::unique_ptr<IStorage>> in_interactive_channel;
+    std::deque<std::shared_ptr<IStorage>> in_interactive_channel;

    // PopInteractiveDataToGame and PushInteractiveDataFromApplet
-    std::deque<std::unique_ptr<IStorage>> out_interactive_channel;
+    std::deque<std::shared_ptr<IStorage>> out_interactive_channel;

    Kernel::EventPair state_changed_event;

@@ -186,7 +186,7 @@ void Error::Execute() {

 void Error::DisplayCompleted() {
    complete = true;
-    broker.PushNormalDataFromApplet(IStorage{std::vector<u8>{}});
+    broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::vector<u8>{}));
    broker.SignalStateChanged();
 }

@@ -20,7 +20,7 @@ namespace Service::AM::Applets {
 constexpr ResultCode ERROR_INVALID_PIN{ErrorModule::PCTL, 221};

 static void LogCurrentStorage(AppletDataBroker& broker, std::string_view prefix) {
-    std::unique_ptr<IStorage> storage = broker.PopNormalDataToApplet();
+    std::shared_ptr<IStorage> storage = broker.PopNormalDataToApplet();
    for (; storage != nullptr; storage = broker.PopNormalDataToApplet()) {
        const auto data = storage->GetData();
        LOG_INFO(Service_AM,
@@ -148,7 +148,7 @@ void Auth::AuthFinished(bool successful) {
    std::vector<u8> out(sizeof(Return));
    std::memcpy(out.data(), &return_, sizeof(Return));

-    broker.PushNormalDataFromApplet(IStorage{std::move(out)});
+    broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::move(out)));
    broker.SignalStateChanged();
 }

@@ -198,7 +198,7 @@ void PhotoViewer::Execute() {
 }

 void PhotoViewer::ViewFinished() {
-    broker.PushNormalDataFromApplet(IStorage{std::vector<u8>{}});
+    broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::vector<u8>{}));
    broker.SignalStateChanged();
 }

@@ -234,8 +234,8 @@ void StubApplet::ExecuteInteractive() {
    LOG_WARNING(Service_AM, "called (STUBBED)");
    LogCurrentStorage(broker, "ExecuteInteractive");

-    broker.PushNormalDataFromApplet(IStorage{std::vector<u8>(0x1000)});
-    broker.PushInteractiveDataFromApplet(IStorage{std::vector<u8>(0x1000)});
+    broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::vector<u8>(0x1000)));
+    broker.PushInteractiveDataFromApplet(std::make_shared<IStorage>(std::vector<u8>(0x1000)));
    broker.SignalStateChanged();
 }

@@ -243,8 +243,8 @@ void StubApplet::Execute() {
    LOG_WARNING(Service_AM, "called (STUBBED)");
    LogCurrentStorage(broker, "Execute");

-    broker.PushNormalDataFromApplet(IStorage{std::vector<u8>(0x1000)});
-    broker.PushInteractiveDataFromApplet(IStorage{std::vector<u8>(0x1000)});
+    broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::vector<u8>(0x1000)));
+    broker.PushInteractiveDataFromApplet(std::make_shared<IStorage>(std::vector<u8>(0x1000)));
    broker.SignalStateChanged();
 }

@@ -50,7 +50,7 @@ void ProfileSelect::ExecuteInteractive() {

 void ProfileSelect::Execute() {
    if (complete) {
-        broker.PushNormalDataFromApplet(IStorage{std::move(final_data)});
+        broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::move(final_data)));
        return;
    }

@@ -71,7 +71,7 @@ void ProfileSelect::SelectionComplete(std::optional<Common::UUID> uuid) {

    final_data = std::vector<u8>(sizeof(UserSelectionOutput));
    std::memcpy(final_data.data(), &output, final_data.size());
-    broker.PushNormalDataFromApplet(IStorage{std::move(final_data)});
+    broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::move(final_data)));
    broker.SignalStateChanged();
 }

@@ -102,7 +102,7 @@ void SoftwareKeyboard::ExecuteInteractive() {

 void SoftwareKeyboard::Execute() {
    if (complete) {
-        broker.PushNormalDataFromApplet(IStorage{std::move(final_data)});
+        broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::move(final_data)));
        broker.SignalStateChanged();
        return;
    }
@@ -145,15 +145,15 @@ void SoftwareKeyboard::WriteText(std::optional<std::u16string> text) {
        final_data = output_main;

        if (complete) {
-            broker.PushNormalDataFromApplet(IStorage{std::move(output_main)});
+            broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::move(output_main)));
            broker.SignalStateChanged();
        } else {
-            broker.PushInteractiveDataFromApplet(IStorage{std::move(output_sub)});
+            broker.PushInteractiveDataFromApplet(std::make_shared<IStorage>(std::move(output_sub)));
        }
    } else {
        output_main[0] = 1;
        complete = true;
-        broker.PushNormalDataFromApplet(IStorage{std::move(output_main)});
+        broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::move(output_main)));
        broker.SignalStateChanged();
    }
 }
@@ -284,7 +284,7 @@ void WebBrowser::Finalize() {
    std::vector<u8> data(sizeof(WebCommonReturnValue));
    std::memcpy(data.data(), &out, sizeof(WebCommonReturnValue));

-    broker.PushNormalDataFromApplet(IStorage{std::move(data)});
+    broker.PushNormalDataFromApplet(std::make_shared<IStorage>(std::move(data)));
    broker.SignalStateChanged();

    if (!temporary_dir.empty() && FileUtil::IsDirectory(temporary_dir)) {
@@ -170,8 +170,10 @@ public:
            {3, nullptr, "SetContextForMultiStream"},
            {4, &IHardwareOpusDecoderManager::DecodeInterleavedWithPerfOld, "DecodeInterleavedWithPerfOld"},
            {5, nullptr, "DecodeInterleavedForMultiStreamWithPerfOld"},
-            {6, &IHardwareOpusDecoderManager::DecodeInterleaved, "DecodeInterleaved"},
-            {7, nullptr, "DecodeInterleavedForMultiStream"},
+            {6, &IHardwareOpusDecoderManager::DecodeInterleaved, "DecodeInterleavedWithPerfAndResetOld"},
+            {7, nullptr, "DecodeInterleavedForMultiStreamWithPerfAndResetOld"},
+            {8, &IHardwareOpusDecoderManager::DecodeInterleaved, "DecodeInterleaved"},
+            {9, nullptr, "DecodeInterleavedForMultiStream"},
        };
        // clang-format on

@@ -117,13 +117,13 @@ bool NullBackend::SynchronizeDirectory(TitleIDVersion title, std::string name,
 }

 bool NullBackend::Clear(u64 title_id) {
-    LOG_DEBUG(Service_BCAT, "called, title_id={:016X}");
+    LOG_DEBUG(Service_BCAT, "called, title_id={:016X}", title_id);

    return true;
 }

 void NullBackend::SetPassphrase(u64 title_id, const Passphrase& passphrase) {
-    LOG_DEBUG(Service_BCAT, "called, title_id={:016X}, passphrase = {}", title_id,
+    LOG_DEBUG(Service_BCAT, "called, title_id={:016X}, passphrase={}", title_id,
              Common::HexToString(passphrase));
 }

@@ -200,7 +200,8 @@ private:
    DownloadResult DownloadInternal(const std::string& resolved_path, u32 timeout_seconds,
                                    const std::string& content_type_name) {
        if (client == nullptr) {
-            client = std::make_unique<httplib::SSLClient>(BOXCAT_HOSTNAME, PORT, timeout_seconds);
+            client = std::make_unique<httplib::SSLClient>(BOXCAT_HOSTNAME, PORT);
+            client->set_timeout_sec(timeout_seconds);
        }

        httplib::Headers headers{
@@ -448,8 +449,8 @@ std::optional<std::vector<u8>> Boxcat::GetLaunchParameter(TitleIDVersion title)

 Boxcat::StatusResult Boxcat::GetStatus(std::optional<std::string>& global,
                                       std::map<std::string, EventStatus>& games) {
-    httplib::SSLClient client{BOXCAT_HOSTNAME, static_cast<int>(PORT),
-                              static_cast<int>(TIMEOUT_SECONDS)};
+    httplib::SSLClient client{BOXCAT_HOSTNAME, static_cast<int>(PORT)};
+    client.set_timeout_sec(static_cast<int>(TIMEOUT_SECONDS));

    httplib::Headers headers{
        {std::string("Game-Assets-API-Version"), std::string(BOXCAT_API_VERSION)},
@@ -420,7 +420,7 @@ public:
            return;
        }

-        IFile file(result.Unwrap());
+        auto file = std::make_shared<IFile>(result.Unwrap());

        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
        rb.Push(RESULT_SUCCESS);
@@ -445,7 +445,7 @@ public:
            return;
        }

-        IDirectory directory(result.Unwrap());
+        auto directory = std::make_shared<IDirectory>(result.Unwrap());

        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
        rb.Push(RESULT_SUCCESS);
@@ -794,8 +794,8 @@ void FSP_SRV::OpenFileSystemWithPatch(Kernel::HLERequestContext& ctx) {
 void FSP_SRV::OpenSdCardFileSystem(Kernel::HLERequestContext& ctx) {
    LOG_DEBUG(Service_FS, "called");

-    IFileSystem filesystem(fsc.OpenSDMC().Unwrap(),
-                           SizeGetter::FromStorageId(fsc, FileSys::StorageId::SdCard));
+    auto filesystem = std::make_shared<IFileSystem>(
+        fsc.OpenSDMC().Unwrap(), SizeGetter::FromStorageId(fsc, FileSys::StorageId::SdCard));

    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
    rb.Push(RESULT_SUCCESS);
@@ -846,7 +846,8 @@ void FSP_SRV::OpenSaveDataFileSystem(Kernel::HLERequestContext& ctx) {
        id = FileSys::StorageId::NandSystem;
    }

-    IFileSystem filesystem(std::move(dir.Unwrap()), SizeGetter::FromStorageId(fsc, id));
+    auto filesystem =
+        std::make_shared<IFileSystem>(std::move(dir.Unwrap()), SizeGetter::FromStorageId(fsc, id));

    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
    rb.Push(RESULT_SUCCESS);
@@ -898,7 +899,7 @@ void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
        return;
    }

-    IStorage storage(std::move(romfs.Unwrap()));
+    auto storage = std::make_shared<IStorage>(std::move(romfs.Unwrap()));

    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
    rb.Push(RESULT_SUCCESS);
@@ -937,7 +938,8 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) {

    FileSys::PatchManager pm{title_id};

-    IStorage storage(pm.PatchRomFS(std::move(data.Unwrap()), 0, FileSys::ContentRecordType::Data));
+    auto storage = std::make_shared<IStorage>(
+        pm.PatchRomFS(std::move(data.Unwrap()), 0, FileSys::ContentRecordType::Data));

    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
    rb.Push(RESULT_SUCCESS);
@@ -287,13 +287,13 @@ void Controller_NPad::RequestPadStateUpdate(u32 npad_id) {
        analog_state[static_cast<std::size_t>(JoystickId::Joystick_Left)]->GetAnalogDirectionStatus(
            Input::AnalogDirection::DOWN));

-    pad_state.r_stick_up.Assign(analog_state[static_cast<std::size_t>(JoystickId::Joystick_Right)]
-                                    ->GetAnalogDirectionStatus(Input::AnalogDirection::RIGHT));
-    pad_state.r_stick_left.Assign(analog_state[static_cast<std::size_t>(JoystickId::Joystick_Right)]
-                                      ->GetAnalogDirectionStatus(Input::AnalogDirection::LEFT));
    pad_state.r_stick_right.Assign(
        analog_state[static_cast<std::size_t>(JoystickId::Joystick_Right)]
-            ->GetAnalogDirectionStatus(Input::AnalogDirection::UP));
+            ->GetAnalogDirectionStatus(Input::AnalogDirection::RIGHT));
+    pad_state.r_stick_left.Assign(analog_state[static_cast<std::size_t>(JoystickId::Joystick_Right)]
+                                      ->GetAnalogDirectionStatus(Input::AnalogDirection::LEFT));
+    pad_state.r_stick_up.Assign(analog_state[static_cast<std::size_t>(JoystickId::Joystick_Right)]
+                                    ->GetAnalogDirectionStatus(Input::AnalogDirection::UP));
    pad_state.r_stick_down.Assign(analog_state[static_cast<std::size_t>(JoystickId::Joystick_Right)]
                                      ->GetAnalogDirectionStatus(Input::AnalogDirection::DOWN));

@@ -10,6 +10,7 @@
 #include "core/core_timing_util.h"
 #include "core/frontend/emu_window.h"
 #include "core/frontend/input.h"
+#include "core/hardware_properties.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
@@ -37,11 +38,11 @@ namespace Service::HID {

 // Updating period for each HID device.
 // TODO(ogniK): Find actual polling rate of hid
-constexpr s64 pad_update_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 66);
+constexpr s64 pad_update_ticks = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 66);
 [[maybe_unused]] constexpr s64 accelerometer_update_ticks =
-    static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 100);
+    static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 100);
 [[maybe_unused]] constexpr s64 gyroscope_update_ticks =
-    static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 100);
+    static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 100);
 constexpr std::size_t SHARED_MEMORY_SIZE = 0x40000;

 IAppletResource::IAppletResource(Core::System& system)
@@ -129,12 +129,20 @@ public:
            {304, nullptr, "Disconnect"},
            {400, nullptr, "Initialize"},
            {401, nullptr, "Finalize"},
-            {402, nullptr, "SetOperationMode"},
+            {402, &IUserLocalCommunicationService::Initialize2, "Initialize2"}, // 7.0.0+
        };
        // clang-format on

        RegisterHandlers(functions);
    }
+
+    void Initialize2(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_LDN, "(STUBBED) called");
+        // Result success seem make this services start network and continue.
+        // If we just pass result error then it will stop and maybe try again and again.
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_UNKNOWN);
+    }
 };

 class LDNS final : public ServiceFramework<LDNS> {
@@ -44,6 +44,8 @@ u32 nvhost_gpu::ioctl(Ioctl command, const std::vector<u8>& input, const std::ve
        return GetWaitbase(input, output);
    case IoctlCommand::IocChannelSetTimeoutCommand:
        return ChannelSetTimeout(input, output);
+    case IoctlCommand::IocChannelSetTimeslice:
+        return ChannelSetTimeslice(input, output);
    default:
        break;
    }
@@ -228,4 +230,14 @@ u32 nvhost_gpu::ChannelSetTimeout(const std::vector<u8>& input, std::vector<u8>&
    return 0;
 }

+u32 nvhost_gpu::ChannelSetTimeslice(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlSetTimeslice params{};
+    std::memcpy(&params, input.data(), sizeof(IoctlSetTimeslice));
+    LOG_INFO(Service_NVDRV, "called, timeslice=0x{:X}", params.timeslice);
+
+    channel_timeslice = params.timeslice;
+
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices
@@ -48,6 +48,7 @@ private:
        IocAllocObjCtxCommand = 0xC0104809,
        IocChannelGetWaitbaseCommand = 0xC0080003,
        IocChannelSetTimeoutCommand = 0x40044803,
+        IocChannelSetTimeslice = 0xC004481D,
    };

    enum class CtxObjects : u32_le {
@@ -101,6 +102,11 @@ private:
    static_assert(sizeof(IoctlChannelSetPriority) == 4,
                  "IoctlChannelSetPriority is incorrect size");

+    struct IoctlSetTimeslice {
+        u32_le timeslice;
+    };
+    static_assert(sizeof(IoctlSetTimeslice) == 4, "IoctlSetTimeslice is incorrect size");
+
    struct IoctlEventIdControl {
        u32_le cmd; // 0=disable, 1=enable, 2=clear
        u32_le id;
@@ -174,6 +180,7 @@ private:
    u64_le user_data{};
    IoctlZCullBind zcull_params{};
    u32_le channel_priority{};
+    u32_le channel_timeslice{};

    u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output);
    u32 SetClientData(const std::vector<u8>& input, std::vector<u8>& output);
@@ -188,6 +195,7 @@ private:
                  const std::vector<u8>& input2, IoctlVersion version);
    u32 GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output);
    u32 ChannelSetTimeout(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 ChannelSetTimeslice(const std::vector<u8>& input, std::vector<u8>& output);

    std::shared_ptr<nvmap> nvmap_dev;
    u32 assigned_syncpoints{};
@@ -12,6 +12,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/readable_event.h"
 #include "core/hle/service/nvdrv/devices/nvdisp_disp0.h"
@@ -26,8 +27,8 @@

 namespace Service::NVFlinger {

-constexpr s64 frame_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 60);
-constexpr s64 frame_ticks_30fps = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 30);
+constexpr s64 frame_ticks = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 60);
+constexpr s64 frame_ticks_30fps = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 30);

 NVFlinger::NVFlinger(Core::System& system) : system(system) {
    displays.emplace_back(0, "Default", system);
@@ -222,7 +223,7 @@ void NVFlinger::Compose() {

 s64 NVFlinger::GetNextTicks() const {
    constexpr s64 max_hertz = 120LL;
-    return (Core::Timing::BASE_CLOCK_RATE * (1LL << swap_interval)) / max_hertz;
+    return (Core::Hardware::BASE_CLOCK_RATE * (1LL << swap_interval)) / max_hertz;
 }

 } // namespace Service::NVFlinger
@@ -5,6 +5,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/service/time/standard_steady_clock_core.h"

 namespace Service::Time::Clock {
@@ -12,7 +13,7 @@ namespace Service::Time::Clock {
 TimeSpanType StandardSteadyClockCore::GetCurrentRawTimePoint(Core::System& system) {
    const TimeSpanType ticks_time_span{TimeSpanType::FromTicks(
        Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-        Core::Timing::CNTFREQ)};
+        Core::Hardware::CNTFREQ)};
    TimeSpanType raw_time_point{setup_value.nanoseconds + ticks_time_span.nanoseconds};

    if (raw_time_point.nanoseconds < cached_raw_time_point.nanoseconds) {
@@ -5,6 +5,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/service/time/tick_based_steady_clock_core.h"

 namespace Service::Time::Clock {
@@ -12,7 +13,7 @@ namespace Service::Time::Clock {
 SteadyClockTimePoint TickBasedSteadyClockCore::GetTimePoint(Core::System& system) {
    const TimeSpanType ticks_time_span{TimeSpanType::FromTicks(
        Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-        Core::Timing::CNTFREQ)};
+        Core::Hardware::CNTFREQ)};

    return {ticks_time_span.ToSeconds(), GetClockSourceId()};
 }
@@ -6,6 +6,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
@@ -233,7 +234,7 @@ void Module::Interface::CalculateMonotonicSystemClockBaseTimePoint(Kernel::HLERe
    if (current_time_point.clock_source_id == context.steady_time_point.clock_source_id) {
        const auto ticks{Clock::TimeSpanType::FromTicks(
            Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-            Core::Timing::CNTFREQ)};
+            Core::Hardware::CNTFREQ)};
        const s64 base_time_point{context.offset + current_time_point.time_point -
                                  ticks.ToSeconds()};
        IPC::ResponseBuilder rb{ctx, (sizeof(s64) / 4) + 2};
@@ -5,6 +5,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/service/time/clock_types.h"
 #include "core/hle/service/time/steady_clock_core.h"
 #include "core/hle/service/time/time_sharedmemory.h"
@@ -31,7 +32,7 @@ void SharedMemory::SetupStandardSteadyClock(Core::System& system,
                                            Clock::TimeSpanType current_time_point) {
    const Clock::TimeSpanType ticks_time_span{Clock::TimeSpanType::FromTicks(
        Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-        Core::Timing::CNTFREQ)};
+        Core::Hardware::CNTFREQ)};
    const Clock::SteadyClockContext context{
        static_cast<u64>(current_time_point.nanoseconds - ticks_time_span.nanoseconds),
        clock_source_id};
@@ -9,6 +9,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/service/hid/controllers/npad.h"
 #include "core/hle/service/hid/hid.h"
@@ -17,7 +18,7 @@

 namespace Memory {

-constexpr s64 CHEAT_ENGINE_TICKS = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 12);
+constexpr s64 CHEAT_ENGINE_TICKS = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 12);
 constexpr u32 KEYPAD_BITMASK = 0x3FFFFFF;

 StandardVmCallbacks::StandardVmCallbacks(Core::System& system, const CheatProcessMetadata& metadata)
@@ -94,7 +94,7 @@ void LogSettings() {
    LogSetting("Renderer_UseAccurateGpuEmulation", Settings::values.use_accurate_gpu_emulation);
    LogSetting("Renderer_UseAsynchronousGpuEmulation",
               Settings::values.use_asynchronous_gpu_emulation);
-    LogSetting("Renderer_AspectRatio", Settings::values.aspect_ratio);
+    LogSetting("Renderer_UseVsync", Settings::values.use_vsync);
    LogSetting("Audio_OutputEngine", Settings::values.sink_id);
    LogSetting("Audio_EnableAudioStretching", Settings::values.enable_audio_stretching);
    LogSetting("Audio_OutputDevice", Settings::values.audio_device_id);
@@ -435,6 +435,7 @@ struct Values {
    bool use_disk_shader_cache;
    bool use_accurate_gpu_emulation;
    bool use_asynchronous_gpu_emulation;
+    bool use_vsync;
    bool force_30fps_mode;

    float bg_red;
@@ -188,7 +188,7 @@ void TelemetrySession::AddInitialInfo(Loader::AppLoader& app_loader) {
             Settings::values.use_accurate_gpu_emulation);
    AddField(field_type, "Renderer_UseAsynchronousGpuEmulation",
             Settings::values.use_asynchronous_gpu_emulation);
-    AddField(field_type, "Renderer_AspectRatio", Settings::values.aspect_ratio);
+    AddField(field_type, "Renderer_UseVsync", Settings::values.use_vsync);
    AddField(field_type, "System_UseDockedMode", Settings::values.use_docked_mode);
 }

@@ -7,13 +7,14 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/hardware_properties.h"
 #include "core/memory.h"
 #include "core/tools/freezer.h"

 namespace Tools {
 namespace {

-constexpr s64 MEMORY_FREEZER_TICKS = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 60);
+constexpr s64 MEMORY_FREEZER_TICKS = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 60);

 u64 MemoryReadWidth(Memory::Memory& memory, u32 width, VAddr addr) {
    switch (width) {
@@ -34,6 +34,20 @@ public:
                               y * coef * (x == 0 ? 1.0f : SQRT_HALF));
    }

+    bool GetAnalogDirectionStatus(Input::AnalogDirection direction) const override {
+        switch (direction) {
+        case Input::AnalogDirection::RIGHT:
+            return right->GetStatus();
+        case Input::AnalogDirection::LEFT:
+            return left->GetStatus();
+        case Input::AnalogDirection::UP:
+            return up->GetStatus();
+        case Input::AnalogDirection::DOWN:
+            return down->GetStatus();
+        }
+        return false;
+    }
+
 private:
    Button up;
    Button down;
@@ -37,6 +37,7 @@ add_library(video_core STATIC
    memory_manager.h
    morton.cpp
    morton.h
+    query_cache.h
    rasterizer_accelerated.cpp
    rasterizer_accelerated.h
    rasterizer_cache.cpp
@@ -74,6 +75,8 @@ add_library(video_core STATIC
    renderer_opengl/gl_stream_buffer.h
    renderer_opengl/gl_texture_cache.cpp
    renderer_opengl/gl_texture_cache.h
+    renderer_opengl/gl_query_cache.cpp
+    renderer_opengl/gl_query_cache.h
    renderer_opengl/maxwell_to_gl.h
    renderer_opengl/renderer_opengl.cpp
    renderer_opengl/renderer_opengl.h
@@ -177,6 +180,8 @@ if (ENABLE_VULKAN)
        renderer_vulkan/vk_memory_manager.h
        renderer_vulkan/vk_pipeline_cache.cpp
        renderer_vulkan/vk_pipeline_cache.h
+        renderer_vulkan/vk_query_cache.cpp
+        renderer_vulkan/vk_query_cache.h
        renderer_vulkan/vk_rasterizer.cpp
        renderer_vulkan/vk_rasterizer.h
        renderer_vulkan/vk_renderpass_cache.cpp
@@ -4,17 +4,21 @@

 #include <cinttypes>
 #include <cstring>
+#include <optional>
 #include "common/assert.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/shader_type.h"
+#include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/textures/texture.h"

 namespace Tegra::Engines {

+using VideoCore::QueryType;
+
 /// First register id that is actually a Macro call.
 constexpr u32 MacroRegistersStart = 0xE00;

@@ -399,6 +403,10 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
        ProcessQueryCondition();
        break;
    }
+    case MAXWELL3D_REG_INDEX(counter_reset): {
+        ProcessCounterReset();
+        break;
+    }
    case MAXWELL3D_REG_INDEX(sync_info): {
        ProcessSyncPoint();
        break;
@@ -481,7 +489,7 @@ void Maxwell3D::FlushMMEInlineDraw() {

    const bool is_indexed = mme_draw.current_mode == MMEDrawMode::Indexed;
    if (ShouldExecute()) {
-        rasterizer.DrawMultiBatch(is_indexed);
+        rasterizer.Draw(is_indexed, true);
    }

    // TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if
@@ -519,61 +527,51 @@ void Maxwell3D::ProcessFirmwareCall4() {
    regs.reg_array[0xd00] = 1;
 }

-void Maxwell3D::ProcessQueryGet() {
-    const GPUVAddr sequence_address{regs.query.QueryAddress()};
-    // Since the sequence address is given as a GPU VAddr, we have to convert it to an application
-    // VAddr before writing.
-
-    // TODO(Subv): Support the other query units.
-    ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop,
-               "Units other than CROP are unimplemented");
-
-    u64 result = 0;
-
-    // TODO(Subv): Support the other query variables
-    switch (regs.query.query_get.select) {
-    case Regs::QuerySelect::Zero:
-        // This seems to actually write the query sequence to the query address.
-        result = regs.query.query_sequence;
-        break;
-    default:
-        result = 1;
-        UNIMPLEMENTED_MSG("Unimplemented query select type {}",
-                          static_cast<u32>(regs.query.query_get.select.Value()));
-    }
-
-    // TODO(Subv): Research and implement how query sync conditions work.
-
+void Maxwell3D::StampQueryResult(u64 payload, bool long_query) {
    struct LongQueryResult {
        u64_le value;
        u64_le timestamp;
    };
    static_assert(sizeof(LongQueryResult) == 16, "LongQueryResult has wrong size");
+    const GPUVAddr sequence_address{regs.query.QueryAddress()};
+    if (long_query) {
+        // Write the 128-bit result structure in long mode. Note: We emulate an infinitely fast
+        // GPU, this command may actually take a while to complete in real hardware due to GPU
+        // wait queues.
+        LongQueryResult query_result{payload, system.GPU().GetTicks()};
+        memory_manager.WriteBlock(sequence_address, &query_result, sizeof(query_result));
+    } else {
+        memory_manager.Write<u32>(sequence_address, static_cast<u32>(payload));
+    }
+}

-    switch (regs.query.query_get.mode) {
-    case Regs::QueryMode::Write:
-    case Regs::QueryMode::Write2: {
-        u32 sequence = regs.query.query_sequence;
-        if (regs.query.query_get.short_query) {
-            // Write the current query sequence to the sequence address.
-            // TODO(Subv): Find out what happens if you use a long query type but mark it as a short
-            // query.
-            memory_manager.Write<u32>(sequence_address, sequence);
-        } else {
-            // Write the 128-bit result structure in long mode. Note: We emulate an infinitely fast
-            // GPU, this command may actually take a while to complete in real hardware due to GPU
-            // wait queues.
-            LongQueryResult query_result{};
-            query_result.value = result;
-            // TODO(Subv): Generate a real GPU timestamp and write it here instead of CoreTiming
-            query_result.timestamp = system.CoreTiming().GetTicks();
-            memory_manager.WriteBlock(sequence_address, &query_result, sizeof(query_result));
+void Maxwell3D::ProcessQueryGet() {
+    // TODO(Subv): Support the other query units.
+    ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop,
+               "Units other than CROP are unimplemented");
+
+    switch (regs.query.query_get.operation) {
+    case Regs::QueryOperation::Release:
+        StampQueryResult(regs.query.query_sequence, regs.query.query_get.short_query == 0);
+        break;
+    case Regs::QueryOperation::Acquire:
+        // TODO(Blinkhawk): Under this operation, the GPU waits for the CPU to write a value that
+        // matches the current payload.
+        UNIMPLEMENTED_MSG("Unimplemented query operation ACQUIRE");
+        break;
+    case Regs::QueryOperation::Counter:
+        if (const std::optional<u64> result = GetQueryResult()) {
+            // If the query returns an empty optional it means it's cached and deferred.
+            // In this case we have a non-empty result, so we stamp it immediately.
+            StampQueryResult(*result, regs.query.query_get.short_query == 0);
        }
        break;
-    }
+    case Regs::QueryOperation::Trap:
+        UNIMPLEMENTED_MSG("Unimplemented query operation TRAP");
+        break;
    default:
-        UNIMPLEMENTED_MSG("Query mode {} not implemented",
-                          static_cast<u32>(regs.query.query_get.mode.Value()));
+        UNIMPLEMENTED_MSG("Unknown query operation");
+        break;
    }
 }

@@ -590,20 +588,20 @@ void Maxwell3D::ProcessQueryCondition() {
    }
    case Regs::ConditionMode::ResNonZero: {
        Regs::QueryCompare cmp;
-        memory_manager.ReadBlockUnsafe(condition_address, &cmp, sizeof(cmp));
+        memory_manager.ReadBlock(condition_address, &cmp, sizeof(cmp));
        execute_on = cmp.initial_sequence != 0U && cmp.initial_mode != 0U;
        break;
    }
    case Regs::ConditionMode::Equal: {
        Regs::QueryCompare cmp;
-        memory_manager.ReadBlockUnsafe(condition_address, &cmp, sizeof(cmp));
+        memory_manager.ReadBlock(condition_address, &cmp, sizeof(cmp));
        execute_on =
            cmp.initial_sequence == cmp.current_sequence && cmp.initial_mode == cmp.current_mode;
        break;
    }
    case Regs::ConditionMode::NotEqual: {
        Regs::QueryCompare cmp;
-        memory_manager.ReadBlockUnsafe(condition_address, &cmp, sizeof(cmp));
+        memory_manager.ReadBlock(condition_address, &cmp, sizeof(cmp));
        execute_on =
            cmp.initial_sequence != cmp.current_sequence || cmp.initial_mode != cmp.current_mode;
        break;
@@ -616,6 +614,18 @@ void Maxwell3D::ProcessQueryCondition() {
    }
 }

+void Maxwell3D::ProcessCounterReset() {
+    switch (regs.counter_reset) {
+    case Regs::CounterReset::SampleCnt:
+        rasterizer.ResetCounter(QueryType::SamplesPassed);
+        break;
+    default:
+        LOG_WARNING(Render_OpenGL, "Unimplemented counter reset={}",
+                    static_cast<int>(regs.counter_reset));
+        break;
+    }
+}
+
 void Maxwell3D::ProcessSyncPoint() {
    const u32 sync_point = regs.sync_info.sync_point.Value();
    const u32 increment = regs.sync_info.increment.Value();
@@ -644,7 +654,7 @@ void Maxwell3D::DrawArrays() {

    const bool is_indexed{regs.index_array.count && !regs.vertex_buffer.count};
    if (ShouldExecute()) {
-        rasterizer.DrawBatch(is_indexed);
+        rasterizer.Draw(is_indexed, false);
    }

    // TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if
@@ -658,6 +668,22 @@ void Maxwell3D::DrawArrays() {
    }
 }

+std::optional<u64> Maxwell3D::GetQueryResult() {
+    switch (regs.query.query_get.select) {
+    case Regs::QuerySelect::Zero:
+        return 0;
+    case Regs::QuerySelect::SamplesPassed:
+        // Deferred.
+        rasterizer.Query(regs.query.QueryAddress(), VideoCore::QueryType::SamplesPassed,
+                         system.GPU().GetTicks());
+        return {};
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented query select type {}",
+                          static_cast<u32>(regs.query.query_get.select.Value()));
+        return 1;
+    }
+}
+
 void Maxwell3D::ProcessCBBind(std::size_t stage_index) {
    // Bind the buffer currently in CB_ADDRESS to the specified index in the desired shader stage.
    auto& shader = state.shader_stages[stage_index];
@@ -6,6 +6,7 @@

 #include <array>
 #include <bitset>
+#include <optional>
 #include <type_traits>
 #include <unordered_map>
 #include <vector>
@@ -71,12 +72,11 @@ public:
        static constexpr std::size_t MaxConstBuffers = 18;
        static constexpr std::size_t MaxConstBufferSize = 0x10000;

-        enum class QueryMode : u32 {
-            Write = 0,
-            Sync = 1,
-            // TODO(Subv): It is currently unknown what the difference between method 2 and method 0
-            // is.
-            Write2 = 2,
+        enum class QueryOperation : u32 {
+            Release = 0,
+            Acquire = 1,
+            Counter = 2,
+            Trap = 3,
        };

        enum class QueryUnit : u32 {
@@ -410,6 +410,27 @@ public:
            Linear = 1,
        };

+        enum class CounterReset : u32 {
+            SampleCnt = 0x01,
+            Unk02 = 0x02,
+            Unk03 = 0x03,
+            Unk04 = 0x04,
+            EmittedPrimitives = 0x10, // Not tested
+            Unk11 = 0x11,
+            Unk12 = 0x12,
+            Unk13 = 0x13,
+            Unk15 = 0x15,
+            Unk16 = 0x16,
+            Unk17 = 0x17,
+            Unk18 = 0x18,
+            Unk1A = 0x1A,
+            Unk1B = 0x1B,
+            Unk1C = 0x1C,
+            Unk1D = 0x1D,
+            Unk1E = 0x1E,
+            GeneratedPrimitives = 0x1F,
+        };
+
        struct Cull {
            enum class FrontFace : u32 {
                ClockWise = 0x0900,
@@ -521,7 +542,7 @@ public:
                BitField<12, 1, InvMemoryLayout> type;
            } memory_layout;
            union {
-                BitField<0, 16, u32> array_mode;
+                BitField<0, 16, u32> layers;
                BitField<16, 1, u32> volume;
            };
            u32 layer_stride;
@@ -704,8 +725,8 @@ public:
                INSERT_UNION_PADDING_WORDS(0x15);

                s32 stencil_back_func_ref;
-                u32 stencil_back_func_mask;
                u32 stencil_back_mask;
+                u32 stencil_back_func_mask;

                INSERT_UNION_PADDING_WORDS(0xC);

@@ -779,8 +800,12 @@ public:

                u32 zeta_width;
                u32 zeta_height;
+                union {
+                    BitField<0, 16, u32> zeta_layers;
+                    BitField<16, 1, u32> zeta_volume;
+                };

-                INSERT_UNION_PADDING_WORDS(0x27);
+                INSERT_UNION_PADDING_WORDS(0x26);

                u32 depth_test_enable;

@@ -858,11 +883,19 @@ public:
                    BitField<7, 1, u32> c7;
                } clip_distance_enabled;

-                INSERT_UNION_PADDING_WORDS(0x1);
+                u32 samplecnt_enable;

                float point_size;

-                INSERT_UNION_PADDING_WORDS(0x7);
+                INSERT_UNION_PADDING_WORDS(0x1);
+
+                u32 point_sprite_enable;
+
+                INSERT_UNION_PADDING_WORDS(0x3);
+
+                CounterReset counter_reset;
+
+                INSERT_UNION_PADDING_WORDS(0x1);

                u32 zeta_enable;

@@ -1077,7 +1110,7 @@ public:
                    u32 query_sequence;
                    union {
                        u32 raw;
-                        BitField<0, 2, QueryMode> mode;
+                        BitField<0, 2, QueryOperation> operation;
                        BitField<4, 1, u32> fence;
                        BitField<12, 4, QueryUnit> unit;
                        BitField<16, 1, QuerySyncCondition> sync_cond;
@@ -1409,9 +1442,15 @@ private:
    /// Handles a write to the QUERY_GET register.
    void ProcessQueryGet();

-    // Handles Conditional Rendering
+    /// Writes the query result accordingly.
+    void StampQueryResult(u64 payload, bool long_query);
+
+    /// Handles conditional rendering.
    void ProcessQueryCondition();

+    /// Handles counter resets.
+    void ProcessCounterReset();
+
    /// Handles writes to syncing register.
    void ProcessSyncPoint();

@@ -1428,6 +1467,9 @@ private:

    // Handles a instance drawcall from MME
    void StepInstance(MMEDrawMode expected_mode, u32 count);
+
+    /// Returns a query's value or an empty object if the value will be deferred through a cache.
+    std::optional<u64> GetQueryResult();
 };

 #define ASSERT_REG_POSITION(field_name, position)                                                  \
@@ -1458,8 +1500,8 @@ ASSERT_REG_POSITION(polygon_offset_fill_enable, 0x372);
 ASSERT_REG_POSITION(patch_vertices, 0x373);
 ASSERT_REG_POSITION(scissor_test, 0x380);
 ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
-ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D6);
-ASSERT_REG_POSITION(stencil_back_mask, 0x3D7);
+ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
+ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);
 ASSERT_REG_POSITION(color_mask_common, 0x3E4);
 ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
 ASSERT_REG_POSITION(depth_bounds, 0x3E7);
@@ -1469,6 +1511,7 @@ ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
 ASSERT_REG_POSITION(rt_control, 0x487);
 ASSERT_REG_POSITION(zeta_width, 0x48a);
 ASSERT_REG_POSITION(zeta_height, 0x48b);
+ASSERT_REG_POSITION(zeta_layers, 0x48c);
 ASSERT_REG_POSITION(depth_test_enable, 0x4B3);
 ASSERT_REG_POSITION(independent_blend_enable, 0x4B9);
 ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
@@ -1493,7 +1536,10 @@ ASSERT_REG_POSITION(screen_y_control, 0x4EB);
 ASSERT_REG_POSITION(vb_element_base, 0x50D);
 ASSERT_REG_POSITION(vb_base_instance, 0x50E);
 ASSERT_REG_POSITION(clip_distance_enabled, 0x544);
+ASSERT_REG_POSITION(samplecnt_enable, 0x545);
 ASSERT_REG_POSITION(point_size, 0x546);
+ASSERT_REG_POSITION(point_sprite_enable, 0x548);
+ASSERT_REG_POSITION(counter_reset, 0x54C);
 ASSERT_REG_POSITION(zeta_enable, 0x54E);
 ASSERT_REG_POSITION(multisample_control, 0x54F);
 ASSERT_REG_POSITION(condition, 0x554);
@@ -624,6 +624,19 @@ enum class ShuffleOperation : u64 {
    Bfly = 3, // shuffleXorNV
 };

+enum class ShfType : u64 {
+    Bits32 = 0,
+    U64 = 2,
+    S64 = 3,
+};
+
+enum class ShfXmode : u64 {
+    None = 0,
+    HI = 1,
+    X = 2,
+    XHI = 3,
+};
+
 union Instruction {
    constexpr Instruction& operator=(const Instruction& instr) {
        value = instr.value;
@@ -775,6 +788,13 @@ union Instruction {
        BitField<39, 1, u64> wrap;
    } shr;

+    union {
+        BitField<37, 2, ShfType> type;
+        BitField<48, 2, ShfXmode> xmode;
+        BitField<50, 1, u64> wrap;
+        BitField<20, 6, u64> immediate;
+    } shf;
+
    union {
        BitField<39, 5, u64> shift_amount;
        BitField<48, 1, u64> negate_b;
@@ -1657,11 +1677,11 @@ union Instruction {
    } xmad;

    union {
-        BitField<20, 14, u64> offset;
+        BitField<20, 14, u64> shifted_offset;
        BitField<34, 5, u64> index;

        u64 GetOffset() const {
-            return offset * 4;
+            return shifted_offset * 4;
        }
    } cbuf34;

@@ -6,6 +6,7 @@
 #include "common/microprofile.h"
 #include "core/core.h"
 #include "core/core_timing.h"
+#include "core/core_timing_util.h"
 #include "core/memory.h"
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/kepler_compute.h"
@@ -122,75 +123,23 @@ bool GPU::CancelSyncptInterrupt(const u32 syncpoint_id, const u32 value) {
    return true;
 }

+u64 GPU::GetTicks() const {
+    // This values were reversed engineered by fincs from NVN
+    // The gpu clock is reported in units of 385/625 nanoseconds
+    constexpr u64 gpu_ticks_num = 384;
+    constexpr u64 gpu_ticks_den = 625;
+
+    const u64 cpu_ticks = system.CoreTiming().GetTicks();
+    const u64 nanoseconds = Core::Timing::CyclesToNs(cpu_ticks).count();
+    const u64 nanoseconds_num = nanoseconds / gpu_ticks_den;
+    const u64 nanoseconds_rem = nanoseconds % gpu_ticks_den;
+    return nanoseconds_num * gpu_ticks_num + (nanoseconds_rem * gpu_ticks_num) / gpu_ticks_den;
+}
+
 void GPU::FlushCommands() {
    renderer.Rasterizer().FlushCommands();
 }

-u32 RenderTargetBytesPerPixel(RenderTargetFormat format) {
-    ASSERT(format != RenderTargetFormat::NONE);
-
-    switch (format) {
-    case RenderTargetFormat::RGBA32_FLOAT:
-    case RenderTargetFormat::RGBA32_UINT:
-        return 16;
-    case RenderTargetFormat::RGBA16_UINT:
-    case RenderTargetFormat::RGBA16_UNORM:
-    case RenderTargetFormat::RGBA16_FLOAT:
-    case RenderTargetFormat::RGBX16_FLOAT:
-    case RenderTargetFormat::RG32_FLOAT:
-    case RenderTargetFormat::RG32_UINT:
-        return 8;
-    case RenderTargetFormat::RGBA8_UNORM:
-    case RenderTargetFormat::RGBA8_SNORM:
-    case RenderTargetFormat::RGBA8_SRGB:
-    case RenderTargetFormat::RGBA8_UINT:
-    case RenderTargetFormat::RGB10_A2_UNORM:
-    case RenderTargetFormat::BGRA8_UNORM:
-    case RenderTargetFormat::BGRA8_SRGB:
-    case RenderTargetFormat::RG16_UNORM:
-    case RenderTargetFormat::RG16_SNORM:
-    case RenderTargetFormat::RG16_UINT:
-    case RenderTargetFormat::RG16_SINT:
-    case RenderTargetFormat::RG16_FLOAT:
-    case RenderTargetFormat::R32_FLOAT:
-    case RenderTargetFormat::R11G11B10_FLOAT:
-    case RenderTargetFormat::R32_UINT:
-        return 4;
-    case RenderTargetFormat::R16_UNORM:
-    case RenderTargetFormat::R16_SNORM:
-    case RenderTargetFormat::R16_UINT:
-    case RenderTargetFormat::R16_SINT:
-    case RenderTargetFormat::R16_FLOAT:
-    case RenderTargetFormat::RG8_UNORM:
-    case RenderTargetFormat::RG8_SNORM:
-        return 2;
-    case RenderTargetFormat::R8_UNORM:
-    case RenderTargetFormat::R8_UINT:
-        return 1;
-    default:
-        UNIMPLEMENTED_MSG("Unimplemented render target format {}", static_cast<u32>(format));
-        return 1;
-    }
-}
-
-u32 DepthFormatBytesPerPixel(DepthFormat format) {
-    switch (format) {
-    case DepthFormat::Z32_S8_X24_FLOAT:
-        return 8;
-    case DepthFormat::Z32_FLOAT:
-    case DepthFormat::S8_Z24_UNORM:
-    case DepthFormat::Z24_X8_UNORM:
-    case DepthFormat::Z24_S8_UNORM:
-    case DepthFormat::Z24_C8_UNORM:
-        return 4;
-    case DepthFormat::Z16_UNORM:
-        return 2;
-    default:
-        UNIMPLEMENTED_MSG("Unimplemented Depth format {}", static_cast<u32>(format));
-        return 1;
-    }
-}
-
 // Note that, traditionally, methods are treated as 4-byte addressable locations, and hence
 // their numbers are written down multiplied by 4 in Docs. Here we are not multiply by 4.
 // So the values you see in docs might be multiplied by 4.
@@ -340,7 +289,7 @@ void GPU::ProcessSemaphoreTriggerMethod() {
        block.sequence = regs.semaphore_sequence;
        // TODO(Kmather73): Generate a real GPU timestamp and write it here instead of
        // CoreTiming
-        block.timestamp = system.CoreTiming().GetTicks();
+        block.timestamp = GetTicks();
        memory_manager->WriteBlock(regs.semaphore_address.SemaphoreAddress(), &block,
                                   sizeof(block));
    } else {
@@ -57,6 +57,7 @@ enum class RenderTargetFormat : u32 {
    RG16_UINT = 0xDD,
    RG16_FLOAT = 0xDE,
    R11G11B10_FLOAT = 0xE0,
+    R32_SINT = 0xE3,
    R32_UINT = 0xE4,
    R32_FLOAT = 0xE5,
    B5G6R5_UNORM = 0xE8,
@@ -82,12 +83,6 @@ enum class DepthFormat : u32 {
    Z32_S8_X24_FLOAT = 0x19,
 };

-/// Returns the number of bytes per pixel of each rendertarget format.
-u32 RenderTargetBytesPerPixel(RenderTargetFormat format);
-
-/// Returns the number of bytes per pixel of each depth format.
-u32 DepthFormatBytesPerPixel(DepthFormat format);
-
 struct CommandListHeader;
 class DebugContext;

@@ -192,6 +187,8 @@ public:

    bool CancelSyncptInterrupt(u32 syncpoint_id, u32 value);

+    u64 GetTicks() const;
+
    std::unique_lock<std::mutex> LockSync() {
        return std::unique_lock{sync_mutex};
    }
@@ -5,7 +5,7 @@
 #include "common/assert.h"
 #include "common/microprofile.h"
 #include "core/core.h"
-#include "core/frontend/scope_acquire_window_context.h"
+#include "core/frontend/scope_acquire_context.h"
 #include "video_core/dma_pusher.h"
 #include "video_core/gpu.h"
 #include "video_core/gpu_thread.h"
@@ -27,7 +27,7 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
        return;
    }

-    Core::Frontend::ScopeAcquireWindowContext acquire_context{renderer.GetRenderWindow()};
+    Core::Frontend::ScopeAcquireContext acquire_context{renderer.GetRenderWindow()};

    CommandDataContainer next;
    while (state.is_running) {
@@ -86,7 +86,7 @@ struct CommandDataContainer {
 struct SynchState final {
    std::atomic_bool is_running{true};

-    using CommandQueue = Common::SPSCQueue<CommandDataContainer>;
+    using CommandQueue = Common::MPSCQueue<CommandDataContainer>;
    CommandQueue queue;
    u64 last_fence{};
    std::atomic<u64> signaled_fence{};
@@ -9,6 +9,7 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
+#include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"

@@ -84,7 +85,9 @@ GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
    const auto cpu_addr = GpuToCpuAddress(gpu_addr);
    ASSERT(cpu_addr);

-    rasterizer.FlushAndInvalidateRegion(cache_addr, aligned_size);
+    // Flush and invalidate through the GPU interface, to be asynchronous if possible.
+    system.GPU().FlushAndInvalidateRegion(cache_addr, aligned_size);
+
    UnmapRange(gpu_addr, aligned_size);
    ASSERT(system.CurrentProcess()
               ->VMManager()
@@ -242,6 +245,8 @@ void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, const std::s
        switch (page_table.attributes[page_index]) {
        case Common::PageType::Memory: {
            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
+            // Flush must happen on the rasterizer interface, such that memory is always synchronous
+            // when it is read (even when in asynchronous GPU mode). Fixes Dead Cells title menu.
            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
            std::memcpy(dest_buffer, src_ptr, copy_amount);
            break;
@@ -292,6 +297,8 @@ void MemoryManager::WriteBlock(GPUVAddr dest_addr, const void* src_buffer, const
        switch (page_table.attributes[page_index]) {
        case Common::PageType::Memory: {
            u8* dest_ptr{page_table.pointers[page_index] + page_offset};
+            // Invalidate must happen on the rasterizer interface, such that memory is always
+            // synchronous when it is written (even when in asynchronous GPU mode).
            rasterizer.InvalidateRegion(ToCacheAddr(dest_ptr), copy_amount);
            std::memcpy(dest_ptr, src_buffer, copy_amount);
            break;
@@ -339,6 +346,8 @@ void MemoryManager::CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, const std::

        switch (page_table.attributes[page_index]) {
        case Common::PageType::Memory: {
+            // Flush must happen on the rasterizer interface, such that memory is always synchronous
+            // when it is copied (even when in asynchronous GPU mode).
            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
            WriteBlock(dest_addr, src_ptr, copy_amount);
@@ -85,6 +85,7 @@ static constexpr ConversionArray morton_to_linear_fns = {
    MortonCopy<true, PixelFormat::RG32UI>,
    MortonCopy<true, PixelFormat::RGBX16F>,
    MortonCopy<true, PixelFormat::R32UI>,
+    MortonCopy<true, PixelFormat::R32I>,
    MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
    MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
    MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
@@ -166,6 +167,7 @@ static constexpr ConversionArray linear_to_morton_fns = {
    MortonCopy<false, PixelFormat::RG32UI>,
    MortonCopy<false, PixelFormat::RGBX16F>,
    MortonCopy<false, PixelFormat::R32UI>,
+    MortonCopy<false, PixelFormat::R32I>,
    nullptr,
    nullptr,
    nullptr,
@@ -0,0 +1,359 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <array>
+#include <cstring>
+#include <iterator>
+#include <memory>
+#include <mutex>
+#include <optional>
+#include <unordered_map>
+#include <vector>
+
+#include "common/assert.h"
+#include "core/core.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/gpu.h"
+#include "video_core/memory_manager.h"
+#include "video_core/rasterizer_interface.h"
+
+namespace VideoCommon {
+
+template <class QueryCache, class HostCounter>
+class CounterStreamBase {
+public:
+    explicit CounterStreamBase(QueryCache& cache, VideoCore::QueryType type)
+        : cache{cache}, type{type} {}
+
+    /// Updates the state of the stream, enabling or disabling as needed.
+    void Update(bool enabled) {
+        if (enabled) {
+            Enable();
+        } else {
+            Disable();
+        }
+    }
+
+    /// Resets the stream to zero. It doesn't disable the query after resetting.
+    void Reset() {
+        if (current) {
+            current->EndQuery();
+
+            // Immediately start a new query to avoid disabling its state.
+            current = cache.Counter(nullptr, type);
+        }
+        last = nullptr;
+    }
+
+    /// Returns the current counter slicing as needed.
+    std::shared_ptr<HostCounter> Current() {
+        if (!current) {
+            return nullptr;
+        }
+        current->EndQuery();
+        last = std::move(current);
+        current = cache.Counter(last, type);
+        return last;
+    }
+
+    /// Returns true when the counter stream is enabled.
+    bool IsEnabled() const {
+        return current != nullptr;
+    }
+
+private:
+    /// Enables the stream.
+    void Enable() {
+        if (current) {
+            return;
+        }
+        current = cache.Counter(last, type);
+    }
+
+    // Disables the stream.
+    void Disable() {
+        if (current) {
+            current->EndQuery();
+        }
+        last = std::exchange(current, nullptr);
+    }
+
+    QueryCache& cache;
+    const VideoCore::QueryType type;
+
+    std::shared_ptr<HostCounter> current;
+    std::shared_ptr<HostCounter> last;
+};
+
+template <class QueryCache, class CachedQuery, class CounterStream, class HostCounter,
+          class QueryPool>
+class QueryCacheBase {
+public:
+    explicit QueryCacheBase(Core::System& system, VideoCore::RasterizerInterface& rasterizer)
+        : system{system}, rasterizer{rasterizer}, streams{{CounterStream{
+                                                      static_cast<QueryCache&>(*this),
+                                                      VideoCore::QueryType::SamplesPassed}}} {}
+
+    void InvalidateRegion(CacheAddr addr, std::size_t size) {
+        std::unique_lock lock{mutex};
+        FlushAndRemoveRegion(addr, size);
+    }
+
+    void FlushRegion(CacheAddr addr, std::size_t size) {
+        std::unique_lock lock{mutex};
+        FlushAndRemoveRegion(addr, size);
+    }
+
+    /**
+     * Records a query in GPU mapped memory, potentially marked with a timestamp.
+     * @param gpu_addr  GPU address to flush to when the mapped memory is read.
+     * @param type      Query type, e.g. SamplesPassed.
+     * @param timestamp Timestamp, when empty the flushed query is assumed to be short.
+     */
+    void Query(GPUVAddr gpu_addr, VideoCore::QueryType type, std::optional<u64> timestamp) {
+        std::unique_lock lock{mutex};
+        auto& memory_manager = system.GPU().MemoryManager();
+        const auto host_ptr = memory_manager.GetPointer(gpu_addr);
+
+        CachedQuery* query = TryGet(ToCacheAddr(host_ptr));
+        if (!query) {
+            const auto cpu_addr = memory_manager.GpuToCpuAddress(gpu_addr);
+            ASSERT_OR_EXECUTE(cpu_addr, return;);
+
+            query = Register(type, *cpu_addr, host_ptr, timestamp.has_value());
+        }
+
+        query->BindCounter(Stream(type).Current(), timestamp);
+    }
+
+    /// Updates counters from GPU state. Expected to be called once per draw, clear or dispatch.
+    void UpdateCounters() {
+        std::unique_lock lock{mutex};
+        const auto& regs = system.GPU().Maxwell3D().regs;
+        Stream(VideoCore::QueryType::SamplesPassed).Update(regs.samplecnt_enable);
+    }
+
+    /// Resets a counter to zero. It doesn't disable the query after resetting.
+    void ResetCounter(VideoCore::QueryType type) {
+        std::unique_lock lock{mutex};
+        Stream(type).Reset();
+    }
+
+    /// Disable all active streams. Expected to be called at the end of a command buffer.
+    void DisableStreams() {
+        std::unique_lock lock{mutex};
+        for (auto& stream : streams) {
+            stream.Update(false);
+        }
+    }
+
+    /// Returns a new host counter.
+    std::shared_ptr<HostCounter> Counter(std::shared_ptr<HostCounter> dependency,
+                                         VideoCore::QueryType type) {
+        return std::make_shared<HostCounter>(static_cast<QueryCache&>(*this), std::move(dependency),
+                                             type);
+    }
+
+    /// Returns the counter stream of the specified type.
+    CounterStream& Stream(VideoCore::QueryType type) {
+        return streams[static_cast<std::size_t>(type)];
+    }
+
+    /// Returns the counter stream of the specified type.
+    const CounterStream& Stream(VideoCore::QueryType type) const {
+        return streams[static_cast<std::size_t>(type)];
+    }
+
+protected:
+    std::array<QueryPool, VideoCore::NumQueryTypes> query_pools;
+
+private:
+    /// Flushes a memory range to guest memory and removes it from the cache.
+    void FlushAndRemoveRegion(CacheAddr addr, std::size_t size) {
+        const u64 addr_begin = static_cast<u64>(addr);
+        const u64 addr_end = addr_begin + static_cast<u64>(size);
+        const auto in_range = [addr_begin, addr_end](CachedQuery& query) {
+            const u64 cache_begin = query.GetCacheAddr();
+            const u64 cache_end = cache_begin + query.SizeInBytes();
+            return cache_begin < addr_end && addr_begin < cache_end;
+        };
+
+        const u64 page_end = addr_end >> PAGE_SHIFT;
+        for (u64 page = addr_begin >> PAGE_SHIFT; page <= page_end; ++page) {
+            const auto& it = cached_queries.find(page);
+            if (it == std::end(cached_queries)) {
+                continue;
+            }
+            auto& contents = it->second;
+            for (auto& query : contents) {
+                if (!in_range(query)) {
+                    continue;
+                }
+                rasterizer.UpdatePagesCachedCount(query.CpuAddr(), query.SizeInBytes(), -1);
+                query.Flush();
+            }
+            contents.erase(std::remove_if(std::begin(contents), std::end(contents), in_range),
+                           std::end(contents));
+        }
+    }
+
+    /// Registers the passed parameters as cached and returns a pointer to the stored cached query.
+    CachedQuery* Register(VideoCore::QueryType type, VAddr cpu_addr, u8* host_ptr, bool timestamp) {
+        rasterizer.UpdatePagesCachedCount(cpu_addr, CachedQuery::SizeInBytes(timestamp), 1);
+        const u64 page = static_cast<u64>(ToCacheAddr(host_ptr)) >> PAGE_SHIFT;
+        return &cached_queries[page].emplace_back(static_cast<QueryCache&>(*this), type, cpu_addr,
+                                                  host_ptr);
+    }
+
+    /// Tries to a get a cached query. Returns nullptr on failure.
+    CachedQuery* TryGet(CacheAddr addr) {
+        const u64 page = static_cast<u64>(addr) >> PAGE_SHIFT;
+        const auto it = cached_queries.find(page);
+        if (it == std::end(cached_queries)) {
+            return nullptr;
+        }
+        auto& contents = it->second;
+        const auto found =
+            std::find_if(std::begin(contents), std::end(contents),
+                         [addr](auto& query) { return query.GetCacheAddr() == addr; });
+        return found != std::end(contents) ? &*found : nullptr;
+    }
+
+    static constexpr std::uintptr_t PAGE_SIZE = 4096;
+    static constexpr unsigned PAGE_SHIFT = 12;
+
+    Core::System& system;
+    VideoCore::RasterizerInterface& rasterizer;
+
+    std::recursive_mutex mutex;
+
+    std::unordered_map<u64, std::vector<CachedQuery>> cached_queries;
+
+    std::array<CounterStream, VideoCore::NumQueryTypes> streams;
+};
+
+template <class QueryCache, class HostCounter>
+class HostCounterBase {
+public:
+    explicit HostCounterBase(std::shared_ptr<HostCounter> dependency_)
+        : dependency{std::move(dependency_)}, depth{dependency ? (dependency->Depth() + 1) : 0} {
+        // Avoid nesting too many dependencies to avoid a stack overflow when these are deleted.
+        constexpr u64 depth_threshold = 96;
+        if (depth > depth_threshold) {
+            depth = 0;
+            base_result = dependency->Query();
+            dependency = nullptr;
+        }
+    }
+    virtual ~HostCounterBase() = default;
+
+    /// Returns the current value of the query.
+    u64 Query() {
+        if (result) {
+            return *result;
+        }
+
+        u64 value = BlockingQuery() + base_result;
+        if (dependency) {
+            value += dependency->Query();
+            dependency = nullptr;
+        }
+
+        result = value;
+        return *result;
+    }
+
+    /// Returns true when flushing this query will potentially wait.
+    bool WaitPending() const noexcept {
+        return result.has_value();
+    }
+
+    u64 Depth() const noexcept {
+        return depth;
+    }
+
+protected:
+    /// Returns the value of query from the backend API blocking as needed.
+    virtual u64 BlockingQuery() const = 0;
+
+private:
+    std::shared_ptr<HostCounter> dependency; ///< Counter to add to this value.
+    std::optional<u64> result;               ///< Filled with the already returned value.
+    u64 depth;                               ///< Number of nested dependencies.
+    u64 base_result = 0;                     ///< Equivalent to nested dependencies value.
+};
+
+template <class HostCounter>
+class CachedQueryBase {
+public:
+    explicit CachedQueryBase(VAddr cpu_addr, u8* host_ptr)
+        : cpu_addr{cpu_addr}, host_ptr{host_ptr} {}
+    virtual ~CachedQueryBase() = default;
+
+    CachedQueryBase(CachedQueryBase&&) noexcept = default;
+    CachedQueryBase(const CachedQueryBase&) = delete;
+
+    CachedQueryBase& operator=(CachedQueryBase&&) noexcept = default;
+    CachedQueryBase& operator=(const CachedQueryBase&) = delete;
+
+    /// Flushes the query to guest memory.
+    virtual void Flush() {
+        // When counter is nullptr it means that it's just been reseted. We are supposed to write a
+        // zero in these cases.
+        const u64 value = counter ? counter->Query() : 0;
+        std::memcpy(host_ptr, &value, sizeof(u64));
+
+        if (timestamp) {
+            std::memcpy(host_ptr + TIMESTAMP_OFFSET, &*timestamp, sizeof(u64));
+        }
+    }
+
+    /// Binds a counter to this query.
+    void BindCounter(std::shared_ptr<HostCounter> counter_, std::optional<u64> timestamp_) {
+        if (counter) {
+            // If there's an old counter set it means the query is being rewritten by the game.
+            // To avoid losing the data forever, flush here.
+            Flush();
+        }
+        counter = std::move(counter_);
+        timestamp = timestamp_;
+    }
+
+    VAddr CpuAddr() const noexcept {
+        return cpu_addr;
+    }
+
+    CacheAddr GetCacheAddr() const noexcept {
+        return ToCacheAddr(host_ptr);
+    }
+
+    u64 SizeInBytes() const noexcept {
+        return SizeInBytes(timestamp.has_value());
+    }
+
+    static constexpr u64 SizeInBytes(bool with_timestamp) noexcept {
+        return with_timestamp ? LARGE_QUERY_SIZE : SMALL_QUERY_SIZE;
+    }
+
+protected:
+    /// Returns true when querying the counter may potentially block.
+    bool WaitPending() const noexcept {
+        return counter && counter->WaitPending();
+    }
+
+private:
+    static constexpr std::size_t SMALL_QUERY_SIZE = 8;   // Query size without timestamp.
+    static constexpr std::size_t LARGE_QUERY_SIZE = 16;  // Query size with timestamp.
+    static constexpr std::intptr_t TIMESTAMP_OFFSET = 8; // Timestamp offset in a large query.
+
+    VAddr cpu_addr;                       ///< Guest CPU address.
+    u8* host_ptr;                         ///< Writable host pointer.
+    std::shared_ptr<HostCounter> counter; ///< Host counter to query, owns the dependency tree.
+    std::optional<u64> timestamp;         ///< Timestamp to flush to guest memory.
+};
+
+} // namespace VideoCommon
@@ -6,6 +6,7 @@

 #include <atomic>
 #include <functional>
+#include <optional>
 #include "common/common_types.h"
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/gpu.h"
@@ -17,6 +18,11 @@ class MemoryManager;

 namespace VideoCore {

+enum class QueryType {
+    SamplesPassed,
+};
+constexpr std::size_t NumQueryTypes = 1;
+
 enum class LoadCallbackStage {
    Prepare,
    Decompile,
@@ -29,11 +35,8 @@ class RasterizerInterface {
 public:
    virtual ~RasterizerInterface() {}

-    /// Draw the current batch of vertex arrays
-    virtual bool DrawBatch(bool is_indexed) = 0;
-
-    /// Draw the current batch of multiple instances of vertex arrays
-    virtual bool DrawMultiBatch(bool is_indexed) = 0;
+    /// Dispatches a draw invocation
+    virtual void Draw(bool is_indexed, bool is_instanced) = 0;

    /// Clear the current framebuffer
    virtual void Clear() = 0;
@@ -41,6 +44,12 @@ public:
    /// Dispatches a compute shader invocation
    virtual void DispatchCompute(GPUVAddr code_addr) = 0;

+    /// Resets the counter of a query
+    virtual void ResetCounter(QueryType type) = 0;
+
+    /// Records a GPU query and caches it
+    virtual void Query(GPUVAddr gpu_addr, QueryType type, std::optional<u64> timestamp) = 0;
+
    /// Notify rasterizer that all caches should be flushed to Switch memory
    virtual void FlushAll() = 0;

@@ -35,15 +35,19 @@ public:
    explicit RendererBase(Core::Frontend::EmuWindow& window);
    virtual ~RendererBase();

-    /// Swap buffers (render frame)
-    virtual void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) = 0;
-
    /// Initialize the renderer
    virtual bool Init() = 0;

    /// Shutdown the renderer
    virtual void ShutDown() = 0;

+    /// Finalize rendering the guest frame and draw into the presentation texture
+    virtual void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) = 0;
+
+    /// Draws the latest frame to the window waiting timeout_ms for a frame to arrive (Renderer
+    /// specific implementation)
+    virtual void TryPresent(int timeout_ms) = 0;
+
    // Getter/setter functions:
    // ------------------------

@@ -0,0 +1,120 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <cstring>
+#include <memory>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+
+#include <glad/glad.h>
+
+#include "common/assert.h"
+#include "core/core.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/memory_manager.h"
+#include "video_core/renderer_opengl/gl_query_cache.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
+
+namespace OpenGL {
+
+namespace {
+
+constexpr std::array<GLenum, VideoCore::NumQueryTypes> QueryTargets = {GL_SAMPLES_PASSED};
+
+constexpr GLenum GetTarget(VideoCore::QueryType type) {
+    return QueryTargets[static_cast<std::size_t>(type)];
+}
+
+} // Anonymous namespace
+
+QueryCache::QueryCache(Core::System& system, RasterizerOpenGL& gl_rasterizer)
+    : VideoCommon::QueryCacheBase<
+          QueryCache, CachedQuery, CounterStream, HostCounter,
+          std::vector<OGLQuery>>{system,
+                                 static_cast<VideoCore::RasterizerInterface&>(gl_rasterizer)},
+      gl_rasterizer{gl_rasterizer} {}
+
+QueryCache::~QueryCache() = default;
+
+OGLQuery QueryCache::AllocateQuery(VideoCore::QueryType type) {
+    auto& reserve = query_pools[static_cast<std::size_t>(type)];
+    OGLQuery query;
+    if (reserve.empty()) {
+        query.Create(GetTarget(type));
+        return query;
+    }
+
+    query = std::move(reserve.back());
+    reserve.pop_back();
+    return query;
+}
+
+void QueryCache::Reserve(VideoCore::QueryType type, OGLQuery&& query) {
+    query_pools[static_cast<std::size_t>(type)].push_back(std::move(query));
+}
+
+bool QueryCache::AnyCommandQueued() const noexcept {
+    return gl_rasterizer.AnyCommandQueued();
+}
+
+HostCounter::HostCounter(QueryCache& cache, std::shared_ptr<HostCounter> dependency,
+                         VideoCore::QueryType type)
+    : VideoCommon::HostCounterBase<QueryCache, HostCounter>{std::move(dependency)}, cache{cache},
+      type{type}, query{cache.AllocateQuery(type)} {
+    glBeginQuery(GetTarget(type), query.handle);
+}
+
+HostCounter::~HostCounter() {
+    cache.Reserve(type, std::move(query));
+}
+
+void HostCounter::EndQuery() {
+    if (!cache.AnyCommandQueued()) {
+        // There are chances a query waited on without commands (glDraw, glClear, glDispatch). Not
+        // having any of these causes a lock. glFlush is considered a command, so we can safely wait
+        // for this. Insert to the OpenGL command stream a flush.
+        glFlush();
+    }
+    glEndQuery(GetTarget(type));
+}
+
+u64 HostCounter::BlockingQuery() const {
+    GLint64 value;
+    glGetQueryObjecti64v(query.handle, GL_QUERY_RESULT, &value);
+    return static_cast<u64>(value);
+}
+
+CachedQuery::CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr, u8* host_ptr)
+    : VideoCommon::CachedQueryBase<HostCounter>{cpu_addr, host_ptr}, cache{&cache}, type{type} {}
+
+CachedQuery::CachedQuery(CachedQuery&& rhs) noexcept
+    : VideoCommon::CachedQueryBase<HostCounter>(std::move(rhs)), cache{rhs.cache}, type{rhs.type} {}
+
+CachedQuery& CachedQuery::operator=(CachedQuery&& rhs) noexcept {
+    VideoCommon::CachedQueryBase<HostCounter>::operator=(std::move(rhs));
+    cache = rhs.cache;
+    type = rhs.type;
+    return *this;
+}
+
+void CachedQuery::Flush() {
+    // Waiting for a query while another query of the same target is enabled locks Nvidia's driver.
+    // To avoid this disable and re-enable keeping the dependency stream.
+    // But we only have to do this if we have pending waits to be done.
+    auto& stream = cache->Stream(type);
+    const bool slice_counter = WaitPending() && stream.IsEnabled();
+    if (slice_counter) {
+        stream.Update(false);
+    }
+
+    VideoCommon::CachedQueryBase<HostCounter>::Flush();
+
+    if (slice_counter) {
+        stream.Update(true);
+    }
+}
+
+} // namespace OpenGL
@@ -0,0 +1,78 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <memory>
+#include <vector>
+
+#include "common/common_types.h"
+#include "video_core/query_cache.h"
+#include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+
+namespace Core {
+class System;
+}
+
+namespace OpenGL {
+
+class CachedQuery;
+class HostCounter;
+class QueryCache;
+class RasterizerOpenGL;
+
+using CounterStream = VideoCommon::CounterStreamBase<QueryCache, HostCounter>;
+
+class QueryCache final : public VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream,
+                                                            HostCounter, std::vector<OGLQuery>> {
+public:
+    explicit QueryCache(Core::System& system, RasterizerOpenGL& rasterizer);
+    ~QueryCache();
+
+    OGLQuery AllocateQuery(VideoCore::QueryType type);
+
+    void Reserve(VideoCore::QueryType type, OGLQuery&& query);
+
+    bool AnyCommandQueued() const noexcept;
+
+private:
+    RasterizerOpenGL& gl_rasterizer;
+};
+
+class HostCounter final : public VideoCommon::HostCounterBase<QueryCache, HostCounter> {
+public:
+    explicit HostCounter(QueryCache& cache, std::shared_ptr<HostCounter> dependency,
+                         VideoCore::QueryType type);
+    ~HostCounter();
+
+    void EndQuery();
+
+private:
+    u64 BlockingQuery() const override;
+
+    QueryCache& cache;
+    const VideoCore::QueryType type;
+    OGLQuery query;
+};
+
+class CachedQuery final : public VideoCommon::CachedQueryBase<HostCounter> {
+public:
+    explicit CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr,
+                         u8* host_ptr);
+    CachedQuery(CachedQuery&& rhs) noexcept;
+    CachedQuery(const CachedQuery&) = delete;
+
+    CachedQuery& operator=(CachedQuery&& rhs) noexcept;
+    CachedQuery& operator=(const CachedQuery&) = delete;
+
+    void Flush() override;
+
+private:
+    QueryCache* cache;
+    VideoCore::QueryType type;
+};
+
+} // namespace OpenGL
@@ -25,6 +25,7 @@
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/shader_type.h"
 #include "video_core/memory_manager.h"
+#include "video_core/renderer_opengl/gl_query_cache.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
@@ -92,8 +93,8 @@ std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
 RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
                                   ScreenInfo& info)
    : RasterizerAccelerated{system.Memory()}, texture_cache{system, *this, device},
-      shader_cache{*this, system, emu_window, device}, system{system}, screen_info{info},
-      buffer_cache{*this, system, device, STREAM_BUFFER_SIZE} {
+      shader_cache{*this, system, emu_window, device}, query_cache{system, *this}, system{system},
+      screen_info{info}, buffer_cache{*this, system, device, STREAM_BUFFER_SIZE} {
    shader_program_manager = std::make_unique<GLShader::ProgramManager>();
    state.draw.shader_program = 0;
    state.Apply();
@@ -541,11 +542,16 @@ void RasterizerOpenGL::Clear() {
    } else if (use_stencil) {
        glClearBufferiv(GL_STENCIL, 0, &regs.clear_stencil);
    }
+
+    ++num_queued_commands;
 }

 void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
    MICROPROFILE_SCOPE(OpenGL_Drawing);
    auto& gpu = system.GPU().Maxwell3D();
+    const auto& regs = gpu.regs;
+
+    query_cache.UpdateCounters();

    SyncRasterizeEnable(state);
    SyncColorMask();
@@ -611,7 +617,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {

    // Setup shaders and their used resources.
    texture_cache.GuardSamplers(true);
-    const auto primitive_mode = MaxwellToGL::PrimitiveTopology(gpu.regs.draw.topology);
+    const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(gpu.regs.draw.topology);
    SetupShaders(primitive_mode);
    texture_cache.GuardSamplers(false);

@@ -638,35 +644,47 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
        glTextureBarrier();
    }

+    ++num_queued_commands;
+
    const GLuint base_instance = static_cast<GLuint>(gpu.regs.vb_base_instance);
    const GLsizei num_instances =
        static_cast<GLsizei>(is_instanced ? gpu.mme_draw.instance_count : 1);
    if (is_indexed) {
-        const GLenum index_format = MaxwellToGL::IndexFormat(gpu.regs.index_array.format);
        const GLint base_vertex = static_cast<GLint>(gpu.regs.vb_element_base);
        const GLsizei num_vertices = static_cast<GLsizei>(gpu.regs.index_array.count);
-        glDrawElementsInstancedBaseVertexBaseInstance(
-            primitive_mode, num_vertices, index_format,
-            reinterpret_cast<const void*>(index_buffer_offset), num_instances, base_vertex,
-            base_instance);
+        const GLvoid* offset = reinterpret_cast<const GLvoid*>(index_buffer_offset);
+        const GLenum format = MaxwellToGL::IndexFormat(gpu.regs.index_array.format);
+        if (num_instances == 1 && base_instance == 0 && base_vertex == 0) {
+            glDrawElements(primitive_mode, num_vertices, format, offset);
+        } else if (num_instances == 1 && base_instance == 0) {
+            glDrawElementsBaseVertex(primitive_mode, num_vertices, format, offset, base_vertex);
+        } else if (base_vertex == 0 && base_instance == 0) {
+            glDrawElementsInstanced(primitive_mode, num_vertices, format, offset, num_instances);
+        } else if (base_vertex == 0) {
+            glDrawElementsInstancedBaseInstance(primitive_mode, num_vertices, format, offset,
+                                                num_instances, base_instance);
+        } else if (base_instance == 0) {
+            glDrawElementsInstancedBaseVertex(primitive_mode, num_vertices, format, offset,
+                                              num_instances, base_vertex);
+        } else {
+            glDrawElementsInstancedBaseVertexBaseInstance(primitive_mode, num_vertices, format,
+                                                          offset, num_instances, base_vertex,
+                                                          base_instance);
+        }
    } else {
        const GLint base_vertex = static_cast<GLint>(gpu.regs.vertex_buffer.first);
        const GLsizei num_vertices = static_cast<GLsizei>(gpu.regs.vertex_buffer.count);
-        glDrawArraysInstancedBaseInstance(primitive_mode, base_vertex, num_vertices, num_instances,
-                                          base_instance);
+        if (num_instances == 1 && base_instance == 0) {
+            glDrawArrays(primitive_mode, base_vertex, num_vertices);
+        } else if (base_instance == 0) {
+            glDrawArraysInstanced(primitive_mode, base_vertex, num_vertices, num_instances);
+        } else {
+            glDrawArraysInstancedBaseInstance(primitive_mode, base_vertex, num_vertices,
+                                              num_instances, base_instance);
+        }
    }
 }

-bool RasterizerOpenGL::DrawBatch(bool is_indexed) {
-    Draw(is_indexed, false);
-    return true;
-}
-
-bool RasterizerOpenGL::DrawMultiBatch(bool is_indexed) {
-    Draw(is_indexed, true);
-    return true;
-}
-
 void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
    if (device.HasBrokenCompute()) {
        return;
@@ -707,6 +725,16 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
    state.ApplyProgramPipeline();

    glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z);
+    ++num_queued_commands;
+}
+
+void RasterizerOpenGL::ResetCounter(VideoCore::QueryType type) {
+    query_cache.ResetCounter(type);
+}
+
+void RasterizerOpenGL::Query(GPUVAddr gpu_addr, VideoCore::QueryType type,
+                             std::optional<u64> timestamp) {
+    query_cache.Query(gpu_addr, type, timestamp);
 }

 void RasterizerOpenGL::FlushAll() {}
@@ -718,6 +746,7 @@ void RasterizerOpenGL::FlushRegion(CacheAddr addr, u64 size) {
    }
    texture_cache.FlushRegion(addr, size);
    buffer_cache.FlushRegion(addr, size);
+    query_cache.FlushRegion(addr, size);
 }

 void RasterizerOpenGL::InvalidateRegion(CacheAddr addr, u64 size) {
@@ -728,6 +757,7 @@ void RasterizerOpenGL::InvalidateRegion(CacheAddr addr, u64 size) {
    texture_cache.InvalidateRegion(addr, size);
    shader_cache.InvalidateRegion(addr, size);
    buffer_cache.InvalidateRegion(addr, size);
+    query_cache.InvalidateRegion(addr, size);
 }

 void RasterizerOpenGL::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
@@ -738,10 +768,18 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
 }

 void RasterizerOpenGL::FlushCommands() {
+    // Only flush when we have commands queued to OpenGL.
+    if (num_queued_commands == 0) {
+        return;
+    }
+    num_queued_commands = 0;
    glFlush();
 }

 void RasterizerOpenGL::TickFrame() {
+    // Ticking a frame means that buffers will be swapped, calling glFlush implicitly.
+    num_queued_commands = 0;
+
    buffer_cache.TickFrame();
 }

@@ -1220,6 +1258,7 @@ void RasterizerOpenGL::SyncPointState() {
    // Limit the point size to 1 since nouveau sometimes sets a point size of 0 (and that's invalid
    // in OpenGL).
    state.point.program_control = regs.vp_point_size.enable != 0;
+    state.point.sprite = regs.point_sprite_enable != 0;
    state.point.size = std::max(1.0f, regs.point_size);
 }

@@ -24,6 +24,7 @@
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_framebuffer_cache.h"
+#include "video_core/renderer_opengl/gl_query_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_sampler_cache.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -57,10 +58,11 @@ public:
                              ScreenInfo& info);
    ~RasterizerOpenGL() override;

-    bool DrawBatch(bool is_indexed) override;
-    bool DrawMultiBatch(bool is_indexed) override;
+    void Draw(bool is_indexed, bool is_instanced) override;
    void Clear() override;
    void DispatchCompute(GPUVAddr code_addr) override;
+    void ResetCounter(VideoCore::QueryType type) override;
+    void Query(GPUVAddr gpu_addr, VideoCore::QueryType type, std::optional<u64> timestamp) override;
    void FlushAll() override;
    void FlushRegion(CacheAddr addr, u64 size) override;
    void InvalidateRegion(CacheAddr addr, u64 size) override;
@@ -75,6 +77,11 @@ public:
    void LoadDiskResources(const std::atomic_bool& stop_loading,
                           const VideoCore::DiskResourceLoadCallback& callback) override;

+    /// Returns true when there are commands queued to the OpenGL server.
+    bool AnyCommandQueued() const {
+        return num_queued_commands > 0;
+    }
+
 private:
    /// Configures the color and depth framebuffer states.
    void ConfigureFramebuffers();
@@ -102,9 +109,6 @@ private:
    void SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr,
                           std::size_t size);

-    /// Syncs all the state, shaders, render targets and textures setting before a draw call.
-    void Draw(bool is_indexed, bool is_instanced);
-
    /// Configures the current textures to use for the draw command.
    void SetupDrawTextures(std::size_t stage_index, const Shader& shader);

@@ -180,10 +184,23 @@ private:
    /// Syncs the alpha test state to match the guest state
    void SyncAlphaTest();

-    /// Check for extension that are not strictly required
-    /// but are needed for correct emulation
+    /// Check for extension that are not strictly required but are needed for correct emulation
    void CheckExtensions();

+    std::size_t CalculateVertexArraysSize() const;
+
+    std::size_t CalculateIndexBufferSize() const;
+
+    /// Updates and returns a vertex array object representing current vertex format
+    GLuint SetupVertexFormat();
+
+    void SetupVertexBuffer(GLuint vao);
+    void SetupVertexInstances(GLuint vao);
+
+    GLintptr SetupIndexBuffer();
+
+    void SetupShaders(GLenum primitive_mode);
+
    const Device device;
    OpenGLState state;

@@ -191,6 +208,7 @@ private:
    ShaderCacheOpenGL shader_cache;
    SamplerCacheOpenGL sampler_cache;
    FramebufferCacheOpenGL framebuffer_cache;
+    QueryCache query_cache;

    Core::System& system;
    ScreenInfo& screen_info;
@@ -208,19 +226,8 @@ private:
    BindBuffersRangePushBuffer bind_ubo_pushbuffer{GL_UNIFORM_BUFFER};
    BindBuffersRangePushBuffer bind_ssbo_pushbuffer{GL_SHADER_STORAGE_BUFFER};

-    std::size_t CalculateVertexArraysSize() const;
-
-    std::size_t CalculateIndexBufferSize() const;
-
-    /// Updates and returns a vertex array object representing current vertex format
-    GLuint SetupVertexFormat();
-
-    void SetupVertexBuffer(GLuint vao);
-    void SetupVertexInstances(GLuint vao);
-
-    GLintptr SetupIndexBuffer();
-
-    void SetupShaders(GLenum primitive_mode);
+    /// Number of commands queued to the OpenGL driver. Reseted on flush.
+    std::size_t num_queued_commands = 0;
 };

 } // namespace OpenGL
@@ -15,6 +15,24 @@ MICROPROFILE_DEFINE(OpenGL_ResourceDeletion, "OpenGL", "Resource Deletion", MP_R

 namespace OpenGL {

+void OGLRenderbuffer::Create() {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glGenRenderbuffers(1, &handle);
+}
+
+void OGLRenderbuffer::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteRenderbuffers(1, &handle);
+    OpenGLState::GetCurState().ResetRenderbuffer(handle).Apply();
+    handle = 0;
+}
+
 void OGLTexture::Create(GLenum target) {
    if (handle != 0)
        return;
@@ -207,4 +225,21 @@ void OGLFramebuffer::Release() {
    handle = 0;
 }

+void OGLQuery::Create(GLenum target) {
+    if (handle != 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceCreation);
+    glCreateQueries(target, 1, &handle);
+}
+
+void OGLQuery::Release() {
+    if (handle == 0)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_ResourceDeletion);
+    glDeleteQueries(1, &handle);
+    handle = 0;
+}
+
 } // namespace OpenGL
@@ -11,6 +11,31 @@

 namespace OpenGL {

+class OGLRenderbuffer : private NonCopyable {
+public:
+    OGLRenderbuffer() = default;
+
+    OGLRenderbuffer(OGLRenderbuffer&& o) noexcept : handle(std::exchange(o.handle, 0)) {}
+
+    ~OGLRenderbuffer() {
+        Release();
+    }
+
+    OGLRenderbuffer& operator=(OGLRenderbuffer&& o) noexcept {
+        Release();
+        handle = std::exchange(o.handle, 0);
+        return *this;
+    }
+
+    /// Creates a new internal OpenGL resource and stores the handle
+    void Create();
+
+    /// Deletes the internal OpenGL resource
+    void Release();
+
+    GLuint handle = 0;
+};
+
 class OGLTexture : private NonCopyable {
 public:
    OGLTexture() = default;
@@ -266,4 +291,29 @@ public:
    GLuint handle = 0;
 };

+class OGLQuery : private NonCopyable {
+public:
+    OGLQuery() = default;
+
+    OGLQuery(OGLQuery&& o) noexcept : handle(std::exchange(o.handle, 0)) {}
+
+    ~OGLQuery() {
+        Release();
+    }
+
+    OGLQuery& operator=(OGLQuery&& o) noexcept {
+        Release();
+        handle = std::exchange(o.handle, 0);
+        return *this;
+    }
+
+    /// Creates a new internal OpenGL resource and stores the handle
+    void Create(GLenum target);
+
+    /// Deletes the internal OpenGL resource
+    void Release();
+
+    GLuint handle = 0;
+};
+
 } // namespace OpenGL
@@ -128,6 +128,7 @@ void OpenGLState::ApplyClipDistances() {

 void OpenGLState::ApplyPointSize() {
    Enable(GL_PROGRAM_POINT_SIZE, cur_state.point.program_control, point.program_control);
+    Enable(GL_POINT_SPRITE, cur_state.point.sprite, point.sprite);
    if (UpdateValue(cur_state.point.size, point.size)) {
        glPointSize(point.size);
    }
@@ -422,6 +423,13 @@ void OpenGLState::ApplyClipControl() {
    }
 }

+void OpenGLState::ApplyRenderBuffer() {
+    if (cur_state.renderbuffer != renderbuffer) {
+        cur_state.renderbuffer = renderbuffer;
+        glBindRenderbuffer(GL_RENDERBUFFER, renderbuffer);
+    }
+}
+
 void OpenGLState::ApplyTextures() {
    const std::size_t size = std::size(textures);
    for (std::size_t i = 0; i < size; ++i) {
@@ -477,6 +485,7 @@ void OpenGLState::Apply() {
    ApplyPolygonOffset();
    ApplyAlphaTest();
    ApplyClipControl();
+    ApplyRenderBuffer();
 }

 void OpenGLState::EmulateViewportWithScissor() {
@@ -550,4 +559,11 @@ OpenGLState& OpenGLState::ResetFramebuffer(GLuint handle) {
    return *this;
 }

+OpenGLState& OpenGLState::ResetRenderbuffer(GLuint handle) {
+    if (renderbuffer == handle) {
+        renderbuffer = 0;
+    }
+    return *this;
+}
+
 } // namespace OpenGL
--- a/Show More
+++ b/Show More