summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--externals/microprofile/microprofile.h6
-rw-r--r--src/core/core.cpp2
-rw-r--r--src/core/gdbstub/gdbstub.cpp9
-rw-r--r--src/core/gdbstub/gdbstub.h7
-rw-r--r--src/core/hle/service/am/applets/web_browser.cpp6
-rw-r--r--src/core/hle/service/set/set.cpp10
-rw-r--r--src/core/hle/service/set/set.h1
-rw-r--r--src/core/hle/service/sm/controller.cpp2
-rw-r--r--src/core/hle/service/time/time_zone_content_manager.cpp2
-rw-r--r--src/core/settings.cpp1
-rw-r--r--src/core/settings.h2
-rw-r--r--src/input_common/udp/client.cpp2
-rw-r--r--src/input_common/udp/udp.cpp3
-rw-r--r--src/video_core/dirty_flags.cpp8
-rw-r--r--src/video_core/dirty_flags.h2
-rw-r--r--src/video_core/engines/kepler_compute.cpp8
-rw-r--r--src/video_core/engines/maxwell_3d.cpp25
-rw-r--r--src/video_core/engines/maxwell_3d.h23
-rw-r--r--src/video_core/engines/shader_bytecode.h11
-rw-r--r--src/video_core/gpu.h1
-rw-r--r--src/video_core/morton.cpp2
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp19
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp34
-rw-r--r--src/video_core/renderer_opengl/gl_state_tracker.cpp1
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.cpp1
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp288
-rw-r--r--src/video_core/renderer_vulkan/maxwell_to_vk.cpp9
-rw-r--r--src/video_core/renderer_vulkan/vk_device.cpp1
-rw-r--r--src/video_core/renderer_vulkan/vk_pipeline_cache.cpp6
-rw-r--r--src/video_core/renderer_vulkan/vk_pipeline_cache.h6
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.cpp90
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.h1
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_decompiler.cpp2
-rw-r--r--src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp2
-rw-r--r--src/video_core/renderer_vulkan/vk_state_tracker.cpp2
-rw-r--r--src/video_core/renderer_vulkan/vk_texture_cache.cpp6
-rw-r--r--src/video_core/shader/decode/bfe.cpp69
-rw-r--r--src/video_core/shader/decode/xmad.cpp63
-rw-r--r--src/video_core/shader/node_helper.cpp2
-rw-r--r--src/video_core/surface.cpp2
-rw-r--r--src/video_core/surface.h142
-rw-r--r--src/video_core/texture_cache/format_lookup_table.cpp3
-rw-r--r--src/video_core/textures/astc.cpp1074
-rw-r--r--src/yuzu/configuration/config.cpp9
-rw-r--r--src/yuzu/configuration/configure_system.cpp4
-rw-r--r--src/yuzu/configuration/configure_system.h1
-rw-r--r--src/yuzu/configuration/configure_system.ui62
-rw-r--r--src/yuzu_cmd/config.cpp2
48 files changed, 1128 insertions, 906 deletions
diff --git a/externals/microprofile/microprofile.h b/externals/microprofile/microprofile.h
index c1556d10c..9d830f7bf 100644
--- a/externals/microprofile/microprofile.h
+++ b/externals/microprofile/microprofile.h
@@ -828,7 +828,7 @@ inline MicroProfileLogEntry MicroProfileMakeLogIndex(uint64_t nBegin, MicroProfi
MicroProfileLogEntry Entry = (nBegin<<62) | ((0x3fff&nToken)<<48) | (MP_LOG_TICK_MASK&nTick);
int t = MicroProfileLogType(Entry);
uint64_t nTimerIndex = MicroProfileLogTimerIndex(Entry);
- MP_ASSERT(t == nBegin);
+ MP_ASSERT((uint64_t)t == nBegin);
MP_ASSERT(nTimerIndex == (nToken&0x3fff));
return Entry;
@@ -1556,10 +1556,10 @@ void MicroProfileFlip()
pFramePut->nFrameStartCpu = MP_TICK();
pFramePut->nFrameStartGpu = (uint32_t)MicroProfileGpuInsertTimeStamp();
- if(pFrameNext->nFrameStartGpu != (uint64_t)-1)
+ if(pFrameNext->nFrameStartGpu != -1)
pFrameNext->nFrameStartGpu = MicroProfileGpuGetTimeStamp((uint32_t)pFrameNext->nFrameStartGpu);
- if(pFrameCurrent->nFrameStartGpu == (uint64_t)-1)
+ if(pFrameCurrent->nFrameStartGpu == -1)
pFrameCurrent->nFrameStartGpu = pFrameNext->nFrameStartGpu + 1;
uint64_t nFrameStartCpu = pFrameCurrent->nFrameStartCpu;
diff --git a/src/core/core.cpp b/src/core/core.cpp
index 218508126..d1bc9340d 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -166,7 +166,7 @@ struct System::Impl {
service_manager = std::make_shared<Service::SM::ServiceManager>();
Service::Init(service_manager, system);
- GDBStub::Init();
+ GDBStub::DeferStart();
renderer = VideoCore::CreateRenderer(emu_window, system);
if (!renderer->Init()) {
diff --git a/src/core/gdbstub/gdbstub.cpp b/src/core/gdbstub/gdbstub.cpp
index e8d8871a7..6d15aeed9 100644
--- a/src/core/gdbstub/gdbstub.cpp
+++ b/src/core/gdbstub/gdbstub.cpp
@@ -141,6 +141,7 @@ constexpr char target_xml[] =
)";
int gdbserver_socket = -1;
+bool defer_start = false;
u8 command_buffer[GDB_BUFFER_SIZE];
u32 command_length;
@@ -1166,6 +1167,9 @@ static void RemoveBreakpoint() {
void HandlePacket() {
if (!IsConnected()) {
+ if (defer_start) {
+ ToggleServer(true);
+ }
return;
}
@@ -1256,6 +1260,10 @@ void ToggleServer(bool status) {
}
}
+void DeferStart() {
+ defer_start = true;
+}
+
static void Init(u16 port) {
if (!server_enabled) {
// Set the halt loop to false in case the user enabled the gdbstub mid-execution.
@@ -1341,6 +1349,7 @@ void Shutdown() {
if (!server_enabled) {
return;
}
+ defer_start = false;
LOG_INFO(Debug_GDBStub, "Stopping GDB ...");
if (gdbserver_socket != -1) {
diff --git a/src/core/gdbstub/gdbstub.h b/src/core/gdbstub/gdbstub.h
index 5a36524b2..8fe3c320b 100644
--- a/src/core/gdbstub/gdbstub.h
+++ b/src/core/gdbstub/gdbstub.h
@@ -43,6 +43,13 @@ void ToggleServer(bool status);
/// Start the gdbstub server.
void Init();
+/**
+ * Defer initialization of the gdbstub to the first packet processing functions.
+ * This avoids a case where the gdbstub thread is frozen after initialization
+ * and fails to respond in time to packets.
+ */
+void DeferStart();
+
/// Stop gdbstub server.
void Shutdown();
diff --git a/src/core/hle/service/am/applets/web_browser.cpp b/src/core/hle/service/am/applets/web_browser.cpp
index 12443c910..9f30e167d 100644
--- a/src/core/hle/service/am/applets/web_browser.cpp
+++ b/src/core/hle/service/am/applets/web_browser.cpp
@@ -254,6 +254,12 @@ void WebBrowser::Execute() {
if (status != RESULT_SUCCESS) {
complete = true;
+
+ // This is a workaround in order not to softlock yuzu when an error happens during the
+ // webapplet init. In order to avoid an svcBreak, the status is set to RESULT_SUCCESS
+ Finalize();
+ status = RESULT_SUCCESS;
+
return;
}
diff --git a/src/core/hle/service/set/set.cpp b/src/core/hle/service/set/set.cpp
index 5bcc0b588..9e12c76fc 100644
--- a/src/core/hle/service/set/set.cpp
+++ b/src/core/hle/service/set/set.cpp
@@ -111,6 +111,14 @@ void SET::GetLanguageCode(Kernel::HLERequestContext& ctx) {
rb.PushEnum(available_language_codes[Settings::values.language_index]);
}
+void SET::GetRegionCode(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_SET, "called");
+
+ IPC::ResponseBuilder rb{ctx, 3};
+ rb.Push(RESULT_SUCCESS);
+ rb.Push(Settings::values.region_index);
+}
+
SET::SET() : ServiceFramework("set") {
// clang-format off
static const FunctionInfo functions[] = {
@@ -118,7 +126,7 @@ SET::SET() : ServiceFramework("set") {
{1, &SET::GetAvailableLanguageCodes, "GetAvailableLanguageCodes"},
{2, &SET::MakeLanguageCode, "MakeLanguageCode"},
{3, &SET::GetAvailableLanguageCodeCount, "GetAvailableLanguageCodeCount"},
- {4, nullptr, "GetRegionCode"},
+ {4, &SET::GetRegionCode, "GetRegionCode"},
{5, &SET::GetAvailableLanguageCodes2, "GetAvailableLanguageCodes2"},
{6, &SET::GetAvailableLanguageCodeCount2, "GetAvailableLanguageCodeCount2"},
{7, nullptr, "GetKeyCodeMap"},
diff --git a/src/core/hle/service/set/set.h b/src/core/hle/service/set/set.h
index b154e08aa..6084b345d 100644
--- a/src/core/hle/service/set/set.h
+++ b/src/core/hle/service/set/set.h
@@ -43,6 +43,7 @@ private:
void GetAvailableLanguageCodeCount(Kernel::HLERequestContext& ctx);
void GetAvailableLanguageCodeCount2(Kernel::HLERequestContext& ctx);
void GetQuestFlag(Kernel::HLERequestContext& ctx);
+ void GetRegionCode(Kernel::HLERequestContext& ctx);
};
} // namespace Service::Set
diff --git a/src/core/hle/service/sm/controller.cpp b/src/core/hle/service/sm/controller.cpp
index c45b285f8..9cca84b31 100644
--- a/src/core/hle/service/sm/controller.cpp
+++ b/src/core/hle/service/sm/controller.cpp
@@ -44,7 +44,7 @@ void Controller::QueryPointerBufferSize(Kernel::HLERequestContext& ctx) {
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(RESULT_SUCCESS);
- rb.Push<u16>(0x500);
+ rb.Push<u16>(0x1000);
}
Controller::Controller() : ServiceFramework("IpcController") {
diff --git a/src/core/hle/service/time/time_zone_content_manager.cpp b/src/core/hle/service/time/time_zone_content_manager.cpp
index 57b1a2bca..78d4acd95 100644
--- a/src/core/hle/service/time/time_zone_content_manager.cpp
+++ b/src/core/hle/service/time/time_zone_content_manager.cpp
@@ -53,7 +53,7 @@ static std::vector<std::string> BuildLocationNameCache(Core::System& system) {
return {};
}
- std::vector<char> raw_data(binary_list->GetSize());
+ std::vector<char> raw_data(binary_list->GetSize() + 1);
binary_list->ReadBytes<char>(raw_data.data(), binary_list->GetSize());
std::stringstream data_stream{raw_data.data()};
diff --git a/src/core/settings.cpp b/src/core/settings.cpp
index 7c0303684..c1282cb80 100644
--- a/src/core/settings.cpp
+++ b/src/core/settings.cpp
@@ -86,6 +86,7 @@ void LogSettings() {
LogSetting("System_RngSeed", Settings::values.rng_seed.value_or(0));
LogSetting("System_CurrentUser", Settings::values.current_user);
LogSetting("System_LanguageIndex", Settings::values.language_index);
+ LogSetting("System_RegionIndex", Settings::values.region_index);
LogSetting("Core_UseMultiCore", Settings::values.use_multi_core);
LogSetting("Renderer_UseResolutionFactor", Settings::values.resolution_factor);
LogSetting("Renderer_UseFrameLimit", Settings::values.use_frame_limit);
diff --git a/src/core/settings.h b/src/core/settings.h
index cb5979e6f..79ec01731 100644
--- a/src/core/settings.h
+++ b/src/core/settings.h
@@ -387,6 +387,8 @@ struct Values {
s32 current_user;
s32 language_index;
+ s32 region_index;
+ s32 sound_index;
// Controls
std::array<PlayerInput, 10> players;
diff --git a/src/input_common/udp/client.cpp b/src/input_common/udp/client.cpp
index e82ae7ef1..da5227058 100644
--- a/src/input_common/udp/client.cpp
+++ b/src/input_common/udp/client.cpp
@@ -35,7 +35,7 @@ public:
pad_index(pad_index) {
boost::system::error_code ec{};
auto ipv4 = boost::asio::ip::make_address_v4(host, ec);
- if (ec.failed()) {
+ if (ec.value() != boost::system::errc::success) {
LOG_ERROR(Input, "Invalid IPv4 address \"{}\" provided to socket", host);
ipv4 = boost::asio::ip::address_v4{};
}
diff --git a/src/input_common/udp/udp.cpp b/src/input_common/udp/udp.cpp
index ca99cc22f..8c6ef1394 100644
--- a/src/input_common/udp/udp.cpp
+++ b/src/input_common/udp/udp.cpp
@@ -3,6 +3,7 @@
// Refer to the license.txt file included.
#include <mutex>
+#include <optional>
#include <tuple>
#include "common/param_package.h"
@@ -44,7 +45,7 @@ public:
std::unique_ptr<Input::TouchDevice> Create(const Common::ParamPackage& params) override {
{
std::lock_guard guard(status->update_mutex);
- status->touch_calibration.emplace();
+ status->touch_calibration = DeviceStatus::CalibrationData{};
// These default values work well for DS4 but probably not other touch inputs
status->touch_calibration->min_x = params.Get("min_x", 100);
status->touch_calibration->min_y = params.Get("min_y", 50);
diff --git a/src/video_core/dirty_flags.cpp b/src/video_core/dirty_flags.cpp
index 4429f3405..e16075993 100644
--- a/src/video_core/dirty_flags.cpp
+++ b/src/video_core/dirty_flags.cpp
@@ -15,14 +15,6 @@ namespace VideoCommon::Dirty {
using Tegra::Engines::Maxwell3D;
-void SetupCommonOnWriteStores(Tegra::Engines::Maxwell3D::DirtyState::Flags& store) {
- store[RenderTargets] = true;
- store[ZetaBuffer] = true;
- for (std::size_t i = 0; i < Maxwell3D::Regs::NumRenderTargets; ++i) {
- store[ColorBuffer0 + i] = true;
- }
-}
-
void SetupDirtyRenderTargets(Tegra::Engines::Maxwell3D::DirtyState::Tables& tables) {
static constexpr std::size_t num_per_rt = NUM(rt[0]);
static constexpr std::size_t begin = OFF(rt);
diff --git a/src/video_core/dirty_flags.h b/src/video_core/dirty_flags.h
index 0dbafd3ef..3f6c1d83a 100644
--- a/src/video_core/dirty_flags.h
+++ b/src/video_core/dirty_flags.h
@@ -44,8 +44,6 @@ void FillBlock(Tegra::Engines::Maxwell3D::DirtyState::Tables& tables, std::size_
FillBlock(tables[1], begin, num, index_b);
}
-void SetupCommonOnWriteStores(Tegra::Engines::Maxwell3D::DirtyState::Flags& store);
-
void SetupDirtyRenderTargets(Tegra::Engines::Maxwell3D::DirtyState::Tables& tables);
} // namespace VideoCommon::Dirty
diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp
index 1ecd65925..368c75a66 100644
--- a/src/video_core/engines/kepler_compute.cpp
+++ b/src/video_core/engines/kepler_compute.cpp
@@ -119,14 +119,6 @@ Texture::TICEntry KeplerCompute::GetTICEntry(u32 tic_index) const {
Texture::TICEntry tic_entry;
memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
- const auto r_type{tic_entry.r_type.Value()};
- const auto g_type{tic_entry.g_type.Value()};
- const auto b_type{tic_entry.b_type.Value()};
- const auto a_type{tic_entry.a_type.Value()};
-
- // TODO(Subv): Different data types for separate components are not supported
- DEBUG_ASSERT(r_type == g_type && r_type == b_type && r_type == a_type);
-
return tic_entry;
}
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index ce536e29b..ba63b44b4 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -98,6 +98,8 @@ void Maxwell3D::InitializeRegisterDefaults() {
regs.framebuffer_srgb = 1;
regs.front_face = Maxwell3D::Regs::FrontFace::ClockWise;
+ shadow_state = regs;
+
mme_inline[MAXWELL3D_REG_INDEX(draw.vertex_end_gl)] = true;
mme_inline[MAXWELL3D_REG_INDEX(draw.vertex_begin_gl)] = true;
mme_inline[MAXWELL3D_REG_INDEX(vertex_buffer.count)] = true;
@@ -160,8 +162,17 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
ASSERT_MSG(method < Regs::NUM_REGS,
"Invalid Maxwell3D register, increase the size of the Regs structure");
- if (regs.reg_array[method] != method_call.argument) {
- regs.reg_array[method] = method_call.argument;
+ u32 arg = method_call.argument;
+ // Keep track of the register value in shadow_state when requested.
+ if (shadow_state.shadow_ram_control == Regs::ShadowRamControl::Track ||
+ shadow_state.shadow_ram_control == Regs::ShadowRamControl::TrackWithFilter) {
+ shadow_state.reg_array[method] = arg;
+ } else if (shadow_state.shadow_ram_control == Regs::ShadowRamControl::Replay) {
+ arg = shadow_state.reg_array[method];
+ }
+
+ if (regs.reg_array[method] != arg) {
+ regs.reg_array[method] = arg;
for (const auto& table : dirty.tables) {
dirty.flags[table[method]] = true;
@@ -169,12 +180,16 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
}
switch (method) {
+ case MAXWELL3D_REG_INDEX(shadow_ram_control): {
+ shadow_state.shadow_ram_control = static_cast<Regs::ShadowRamControl>(method_call.argument);
+ break;
+ }
case MAXWELL3D_REG_INDEX(macros.data): {
- ProcessMacroUpload(method_call.argument);
+ ProcessMacroUpload(arg);
break;
}
case MAXWELL3D_REG_INDEX(macros.bind): {
- ProcessMacroBind(method_call.argument);
+ ProcessMacroBind(arg);
break;
}
case MAXWELL3D_REG_INDEX(firmware[4]): {
@@ -250,7 +265,7 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
}
case MAXWELL3D_REG_INDEX(data_upload): {
const bool is_last_call = method_call.IsLastCall();
- upload_state.ProcessData(method_call.argument, is_last_call);
+ upload_state.ProcessData(arg, is_last_call);
if (is_last_call) {
OnMemoryWrite();
}
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index 8a9e9992e..d24c9f657 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -531,6 +531,17 @@ public:
Fill = 0x1b02,
};
+ enum class ShadowRamControl : u32 {
+ // write value to shadow ram
+ Track = 0,
+ // write value to shadow ram ( with validation ??? )
+ TrackWithFilter = 1,
+ // only write to real hw register
+ Passthrough = 2,
+ // write value from shadow ram to real hw register
+ Replay = 3,
+ };
+
struct RenderTargetConfig {
u32 address_high;
u32 address_low;
@@ -674,7 +685,9 @@ public:
u32 bind;
} macros;
- INSERT_UNION_PADDING_WORDS(0x17);
+ ShadowRamControl shadow_ram_control;
+
+ INSERT_UNION_PADDING_WORDS(0x16);
Upload::Registers upload;
struct {
@@ -1263,7 +1276,12 @@ public:
};
std::array<u32, NUM_REGS> reg_array;
};
- } regs{};
+ };
+
+ Regs regs{};
+
+ /// Store temporary hw register values, used by some calls to restore state after a operation
+ Regs shadow_state;
static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32), "Maxwell3D Regs has wrong size");
static_assert(std::is_trivially_copyable_v<Regs>, "Maxwell3D Regs must be trivially copyable");
@@ -1458,6 +1476,7 @@ private:
"Field " #field_name " has invalid position")
ASSERT_REG_POSITION(macros, 0x45);
+ASSERT_REG_POSITION(shadow_ram_control, 0x49);
ASSERT_REG_POSITION(upload, 0x60);
ASSERT_REG_POSITION(exec_upload, 0x6C);
ASSERT_REG_POSITION(data_upload, 0x6D);
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index 4cd0c07a7..49dc5abe0 100644
--- a/src/video_core/engines/shader_bytecode.h
+++ b/src/video_core/engines/shader_bytecode.h
@@ -917,14 +917,9 @@ union Instruction {
} fadd32i;
union {
- BitField<20, 8, u64> shift_position;
- BitField<28, 8, u64> shift_length;
- BitField<48, 1, u64> negate_b;
- BitField<49, 1, u64> negate_a;
-
- u64 GetLeftShiftValue() const {
- return 32 - (shift_position + shift_length);
- }
+ BitField<40, 1, u64> brev;
+ BitField<47, 1, u64> rd_cc;
+ BitField<48, 1, u64> is_signed;
} bfe;
union {
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index ba8c9d665..64acb17df 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -39,6 +39,7 @@ enum class RenderTargetFormat : u32 {
RGBA32_FLOAT = 0xC0,
RGBA32_UINT = 0xC2,
RGBA16_UNORM = 0xC6,
+ RGBA16_SNORM = 0xC7,
RGBA16_UINT = 0xC9,
RGBA16_FLOAT = 0xCA,
RG32_FLOAT = 0xCB,
diff --git a/src/video_core/morton.cpp b/src/video_core/morton.cpp
index f2c83266e..6d522c318 100644
--- a/src/video_core/morton.cpp
+++ b/src/video_core/morton.cpp
@@ -51,6 +51,7 @@ static constexpr ConversionArray morton_to_linear_fns = {
MortonCopy<true, PixelFormat::R8UI>,
MortonCopy<true, PixelFormat::RGBA16F>,
MortonCopy<true, PixelFormat::RGBA16U>,
+ MortonCopy<true, PixelFormat::RGBA16S>,
MortonCopy<true, PixelFormat::RGBA16UI>,
MortonCopy<true, PixelFormat::R11FG11FB10F>,
MortonCopy<true, PixelFormat::RGBA32UI>,
@@ -131,6 +132,7 @@ static constexpr ConversionArray linear_to_morton_fns = {
MortonCopy<false, PixelFormat::R8U>,
MortonCopy<false, PixelFormat::R8UI>,
MortonCopy<false, PixelFormat::RGBA16F>,
+ MortonCopy<false, PixelFormat::RGBA16S>,
MortonCopy<false, PixelFormat::RGBA16U>,
MortonCopy<false, PixelFormat::RGBA16UI>,
MortonCopy<false, PixelFormat::R11FG11FB10F>,
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 1af4268a4..826eee7df 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -93,10 +93,6 @@ void oglEnable(GLenum cap, bool state) {
(state ? glEnable : glDisable)(cap);
}
-void oglEnablei(GLenum cap, bool state, GLuint index) {
- (state ? glEnablei : glDisablei)(cap, index);
-}
-
} // Anonymous namespace
RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
@@ -478,7 +474,6 @@ void RasterizerOpenGL::Clear() {
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();
@@ -529,7 +524,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
// Upload vertex and index data.
SetupVertexBuffer();
SetupVertexInstances();
- GLintptr index_buffer_offset;
+ GLintptr index_buffer_offset = 0;
if (is_indexed) {
index_buffer_offset = SetupIndexBuffer();
}
@@ -555,7 +550,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
ConfigureFramebuffers();
// Signal the buffer cache that we are not going to upload more things.
- const bool invalidate = buffer_cache.Unmap();
+ buffer_cache.Unmap();
// Now that we are no longer uploading data, we can safely bind the buffers to OpenGL.
vertex_array_pushbuffer.Bind();
@@ -938,13 +933,15 @@ void RasterizerOpenGL::SyncViewport() {
}
flags[Dirty::Viewport0 + i] = false;
- const Common::Rectangle<f32> rect{regs.viewport_transform[i].GetRect()};
+ const auto& src = regs.viewport_transform[i];
+ const Common::Rectangle<f32> rect{src.GetRect()};
glViewportIndexedf(static_cast<GLuint>(i), rect.left, rect.bottom, rect.GetWidth(),
rect.GetHeight());
- const auto& src = regs.viewports[i];
- glDepthRangeIndexed(static_cast<GLuint>(i), static_cast<GLdouble>(src.depth_range_near),
- static_cast<GLdouble>(src.depth_range_far));
+ const GLdouble reduce_z = regs.depth_mode == Maxwell::DepthMode::MinusOneToOne;
+ const GLdouble near_depth = src.translate_z - src.scale_z * reduce_z;
+ const GLdouble far_depth = src.translate_z + src.scale_z;
+ glDepthRangeIndexed(static_cast<GLuint>(i), near_depth, far_depth);
}
}
}
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index e6ae8041b..8aa4a7ac9 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -402,10 +402,6 @@ std::string FlowStackTopName(MetaStackClass stack) {
return fmt::format("{}_flow_stack_top", GetFlowStackPrefix(stack));
}
-[[deprecated]] constexpr bool IsVertexShader(ShaderType stage) {
- return stage == ShaderType::Vertex;
-}
-
struct GenericVaryingDescription {
std::string name;
u8 first_element = 0;
@@ -538,8 +534,9 @@ private:
}
void DeclareVertex() {
- if (!IsVertexShader(stage))
+ if (stage != ShaderType::Vertex) {
return;
+ }
DeclareVertexRedeclarations();
}
@@ -621,14 +618,14 @@ private:
break;
}
}
- if (!IsVertexShader(stage) || device.HasVertexViewportLayer()) {
+ if (stage != ShaderType::Vertex || device.HasVertexViewportLayer()) {
if (ir.UsesLayer()) {
code.AddLine("int gl_Layer;");
}
if (ir.UsesViewportIndex()) {
code.AddLine("int gl_ViewportIndex;");
}
- } else if ((ir.UsesLayer() || ir.UsesViewportIndex()) && IsVertexShader(stage) &&
+ } else if ((ir.UsesLayer() || ir.UsesViewportIndex()) && stage == ShaderType::Vertex &&
!device.HasVertexViewportLayer()) {
LOG_ERROR(
Render_OpenGL,
@@ -1170,7 +1167,7 @@ private:
// TODO(Subv): Find out what the values are for the first two elements when inside a
// vertex shader, and what's the value of the fourth element when inside a Tess Eval
// shader.
- ASSERT(IsVertexShader(stage));
+ ASSERT(stage == ShaderType::Vertex);
switch (element) {
case 2:
// Config pack's first value is instance_id.
@@ -1248,12 +1245,12 @@ private:
UNIMPLEMENTED();
return {};
case 1:
- if (IsVertexShader(stage) && !device.HasVertexViewportLayer()) {
+ if (stage == ShaderType::Vertex && !device.HasVertexViewportLayer()) {
return {};
}
return {{"gl_Layer", Type::Int}};
case 2:
- if (IsVertexShader(stage) && !device.HasVertexViewportLayer()) {
+ if (stage == ShaderType::Vertex && !device.HasVertexViewportLayer()) {
return {};
}
return {{"gl_ViewportIndex", Type::Int}};
@@ -2052,16 +2049,19 @@ private:
expr += GetSampler(meta->sampler);
expr += ", ";
- expr += constructors.at(operation.GetOperandsCount() - 1);
+ expr += constructors.at(operation.GetOperandsCount() + (meta->array ? 1 : 0) - 1);
expr += '(';
for (std::size_t i = 0; i < count; ++i) {
- expr += VisitOperand(operation, i).AsInt();
- const std::size_t next = i + 1;
- if (next == count)
- expr += ')';
- else if (next < count)
+ if (i > 0) {
expr += ", ";
+ }
+ expr += VisitOperand(operation, i).AsInt();
}
+ if (meta->array) {
+ expr += ", ";
+ expr += Visit(meta->array).AsInt();
+ }
+ expr += ')';
if (meta->lod && !meta->sampler.IsBuffer()) {
expr += ", ";
@@ -2572,7 +2572,7 @@ private:
}
u32 GetNumPhysicalInputAttributes() const {
- return IsVertexShader(stage) ? GetNumPhysicalAttributes() : GetNumPhysicalVaryings();
+ return stage == ShaderType::Vertex ? GetNumPhysicalAttributes() : GetNumPhysicalVaryings();
}
u32 GetNumPhysicalAttributes() const {
diff --git a/src/video_core/renderer_opengl/gl_state_tracker.cpp b/src/video_core/renderer_opengl/gl_state_tracker.cpp
index 3f3bdf812..255ac3147 100644
--- a/src/video_core/renderer_opengl/gl_state_tracker.cpp
+++ b/src/video_core/renderer_opengl/gl_state_tracker.cpp
@@ -238,7 +238,6 @@ void StateTracker::Initialize() {
SetupDirtyMisc(tables);
auto& store = dirty.on_write_stores;
- SetupCommonOnWriteStores(store);
store[VertexBuffers] = true;
for (std::size_t i = 0; i < Regs::NumVertexArrays; ++i) {
store[VertexBuffer0 + i] = true;
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index 2d3838a7a..f424e3000 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -53,6 +53,7 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format
{GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE, false}, // R8UI
{GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, false}, // RGBA16F
{GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT, false}, // RGBA16U
+ {GL_RGBA16_SNORM, GL_RGBA, GL_SHORT, false}, // RGBA16S
{GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT, false}, // RGBA16UI
{GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, false}, // R11FG11FB10F
{GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT, false}, // RGBA32UI
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index 12333e8c9..fca5e3ec0 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -5,8 +5,11 @@
#include <algorithm>
#include <cstddef>
#include <cstdlib>
+#include <cstring>
#include <memory>
+
#include <glad/glad.h>
+
#include "common/assert.h"
#include "common/logging/log.h"
#include "common/microprofile.h"
@@ -25,6 +28,8 @@
namespace OpenGL {
+namespace {
+
// If the size of this is too small, it ends up creating a soft cap on FPS as the renderer will have
// to wait on available presentation frames.
constexpr std::size_t SWAP_CHAIN_SIZE = 3;
@@ -41,124 +46,6 @@ struct Frame {
bool is_srgb{}; /// Framebuffer is sRGB or RGB
};
-/**
- * For smooth Vsync rendering, we want to always present the latest frame that the core generates,
- * but also make sure that rendering happens at the pace that the frontend dictates. This is a
- * helper class that the renderer uses to sync frames between the render thread and the presentation
- * thread
- */
-class FrameMailbox {
-public:
- std::mutex swap_chain_lock;
- std::condition_variable present_cv;
- std::array<Frame, SWAP_CHAIN_SIZE> swap_chain{};
- std::queue<Frame*> free_queue;
- std::deque<Frame*> present_queue;
- Frame* previous_frame{};
-
- FrameMailbox() {
- for (auto& frame : swap_chain) {
- free_queue.push(&frame);
- }
- }
-
- ~FrameMailbox() {
- // lock the mutex and clear out the present and free_queues and notify any people who are
- // blocked to prevent deadlock on shutdown
- std::scoped_lock lock{swap_chain_lock};
- std::queue<Frame*>().swap(free_queue);
- present_queue.clear();
- present_cv.notify_all();
- }
-
- void ReloadPresentFrame(Frame* frame, u32 height, u32 width) {
- frame->present.Release();
- frame->present.Create();
- GLint previous_draw_fbo{};
- glGetIntegerv(GL_DRAW_FRAMEBUFFER_BINDING, &previous_draw_fbo);
- glBindFramebuffer(GL_FRAMEBUFFER, frame->present.handle);
- glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER,
- frame->color.handle);
- if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
- LOG_CRITICAL(Render_OpenGL, "Failed to recreate present FBO!");
- }
- glBindFramebuffer(GL_DRAW_FRAMEBUFFER, previous_draw_fbo);
- frame->color_reloaded = false;
- }
-
- void ReloadRenderFrame(Frame* frame, u32 width, u32 height) {
- // Recreate the color texture attachment
- frame->color.Release();
- frame->color.Create();
- const GLenum internal_format = frame->is_srgb ? GL_SRGB8 : GL_RGB8;
- glNamedRenderbufferStorage(frame->color.handle, internal_format, width, height);
-
- // Recreate the FBO for the render target
- frame->render.Release();
- frame->render.Create();
- glBindFramebuffer(GL_FRAMEBUFFER, frame->render.handle);
- glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER,
- frame->color.handle);
- if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
- LOG_CRITICAL(Render_OpenGL, "Failed to recreate render FBO!");
- }
-
- frame->width = width;
- frame->height = height;
- frame->color_reloaded = true;
- }
-
- Frame* GetRenderFrame() {
- std::unique_lock lock{swap_chain_lock};
-
- // If theres no free frames, we will reuse the oldest render frame
- if (free_queue.empty()) {
- auto frame = present_queue.back();
- present_queue.pop_back();
- return frame;
- }
-
- Frame* frame = free_queue.front();
- free_queue.pop();
- return frame;
- }
-
- void ReleaseRenderFrame(Frame* frame) {
- std::unique_lock lock{swap_chain_lock};
- present_queue.push_front(frame);
- present_cv.notify_one();
- }
-
- Frame* TryGetPresentFrame(int timeout_ms) {
- std::unique_lock lock{swap_chain_lock};
- // wait for new entries in the present_queue
- present_cv.wait_for(lock, std::chrono::milliseconds(timeout_ms),
- [&] { return !present_queue.empty(); });
- if (present_queue.empty()) {
- // timed out waiting for a frame to draw so return the previous frame
- return previous_frame;
- }
-
- // free the previous frame and add it back to the free queue
- if (previous_frame) {
- free_queue.push(previous_frame);
- }
-
- // the newest entries are pushed to the front of the queue
- Frame* frame = present_queue.front();
- present_queue.pop_front();
- // remove all old entries from the present queue and move them back to the free_queue
- for (auto f : present_queue) {
- free_queue.push(f);
- }
- present_queue.clear();
- previous_frame = frame;
- return frame;
- }
-};
-
-namespace {
-
constexpr char VERTEX_SHADER[] = R"(
#version 430 core
@@ -211,6 +98,24 @@ struct ScreenRectVertex {
std::array<GLfloat, 2> tex_coord;
};
+/// Returns true if any debug tool is attached
+bool HasDebugTool() {
+ const bool nsight = std::getenv("NVTX_INJECTION64_PATH") || std::getenv("NSIGHT_LAUNCHED");
+ if (nsight) {
+ return true;
+ }
+
+ GLint num_extensions;
+ glGetIntegerv(GL_NUM_EXTENSIONS, &num_extensions);
+ for (GLuint index = 0; index < static_cast<GLuint>(num_extensions); ++index) {
+ const auto name = reinterpret_cast<const char*>(glGetStringi(GL_EXTENSIONS, index));
+ if (!std::strcmp(name, "GL_EXT_debug_tool")) {
+ return true;
+ }
+ }
+ return false;
+}
+
/**
* Defines a 1:1 pixel ortographic projection matrix with (0,0) on the top-left
* corner and (width, height) on the lower-bottom.
@@ -294,6 +199,153 @@ void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum severit
} // Anonymous namespace
+/**
+ * For smooth Vsync rendering, we want to always present the latest frame that the core generates,
+ * but also make sure that rendering happens at the pace that the frontend dictates. This is a
+ * helper class that the renderer uses to sync frames between the render thread and the presentation
+ * thread
+ */
+class FrameMailbox {
+public:
+ std::mutex swap_chain_lock;
+ std::condition_variable present_cv;
+ std::array<Frame, SWAP_CHAIN_SIZE> swap_chain{};
+ std::queue<Frame*> free_queue;
+ std::deque<Frame*> present_queue;
+ Frame* previous_frame{};
+
+ FrameMailbox() : has_debug_tool{HasDebugTool()} {
+ for (auto& frame : swap_chain) {
+ free_queue.push(&frame);
+ }
+ }
+
+ ~FrameMailbox() {
+ // lock the mutex and clear out the present and free_queues and notify any people who are
+ // blocked to prevent deadlock on shutdown
+ std::scoped_lock lock{swap_chain_lock};
+ std::queue<Frame*>().swap(free_queue);
+ present_queue.clear();
+ present_cv.notify_all();
+ }
+
+ void ReloadPresentFrame(Frame* frame, u32 height, u32 width) {
+ frame->present.Release();
+ frame->present.Create();
+ GLint previous_draw_fbo{};
+ glGetIntegerv(GL_DRAW_FRAMEBUFFER_BINDING, &previous_draw_fbo);
+ glBindFramebuffer(GL_FRAMEBUFFER, frame->present.handle);
+ glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER,
+ frame->color.handle);
+ if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
+ LOG_CRITICAL(Render_OpenGL, "Failed to recreate present FBO!");
+ }
+ glBindFramebuffer(GL_DRAW_FRAMEBUFFER, previous_draw_fbo);
+ frame->color_reloaded = false;
+ }
+
+ void ReloadRenderFrame(Frame* frame, u32 width, u32 height) {
+ // Recreate the color texture attachment
+ frame->color.Release();
+ frame->color.Create();
+ const GLenum internal_format = frame->is_srgb ? GL_SRGB8 : GL_RGB8;
+ glNamedRenderbufferStorage(frame->color.handle, internal_format, width, height);
+
+ // Recreate the FBO for the render target
+ frame->render.Release();
+ frame->render.Create();
+ glBindFramebuffer(GL_FRAMEBUFFER, frame->render.handle);
+ glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER,
+ frame->color.handle);
+ if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
+ LOG_CRITICAL(Render_OpenGL, "Failed to recreate render FBO!");
+ }
+
+ frame->width = width;
+ frame->height = height;
+ frame->color_reloaded = true;
+ }
+
+ Frame* GetRenderFrame() {
+ std::unique_lock lock{swap_chain_lock};
+
+ // If theres no free frames, we will reuse the oldest render frame
+ if (free_queue.empty()) {
+ auto frame = present_queue.back();
+ present_queue.pop_back();
+ return frame;
+ }
+
+ Frame* frame = free_queue.front();
+ free_queue.pop();
+ return frame;
+ }
+
+ void ReleaseRenderFrame(Frame* frame) {
+ std::unique_lock lock{swap_chain_lock};
+ present_queue.push_front(frame);
+ present_cv.notify_one();
+
+ DebugNotifyNextFrame();
+ }
+
+ Frame* TryGetPresentFrame(int timeout_ms) {
+ DebugWaitForNextFrame();
+
+ std::unique_lock lock{swap_chain_lock};
+ // wait for new entries in the present_queue
+ present_cv.wait_for(lock, std::chrono::milliseconds(timeout_ms),
+ [&] { return !present_queue.empty(); });
+ if (present_queue.empty()) {
+ // timed out waiting for a frame to draw so return the previous frame
+ return previous_frame;
+ }
+
+ // free the previous frame and add it back to the free queue
+ if (previous_frame) {
+ free_queue.push(previous_frame);
+ }
+
+ // the newest entries are pushed to the front of the queue
+ Frame* frame = present_queue.front();
+ present_queue.pop_front();
+ // remove all old entries from the present queue and move them back to the free_queue
+ for (auto f : present_queue) {
+ free_queue.push(f);
+ }
+ present_queue.clear();
+ previous_frame = frame;
+ return frame;
+ }
+
+private:
+ std::mutex debug_synch_mutex;
+ std::condition_variable debug_synch_condition;
+ std::atomic_int frame_for_debug{};
+ const bool has_debug_tool; // When true, using a GPU debugger, so keep frames in lock-step
+
+ /// Signal that a new frame is available (called from GPU thread)
+ void DebugNotifyNextFrame() {
+ if (!has_debug_tool) {
+ return;
+ }
+ frame_for_debug++;
+ std::lock_guard lock{debug_synch_mutex};
+ debug_synch_condition.notify_one();
+ }
+
+ /// Wait for a new frame to be available (called from presentation thread)
+ void DebugWaitForNextFrame() {
+ if (!has_debug_tool) {
+ return;
+ }
+ const int last_frame = frame_for_debug;
+ std::unique_lock lock{debug_synch_mutex};
+ debug_synch_condition.wait(lock,
+ [this, last_frame] { return frame_for_debug > last_frame; });
+ }
+};
+
RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::System& system)
: VideoCore::RendererBase{emu_window}, emu_window{emu_window}, system{system},
frame_mailbox{std::make_unique<FrameMailbox>()} {}
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
index df3ac707c..f93447610 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
@@ -125,6 +125,7 @@ struct FormatTuple {
{vk::Format::eR8Uint, Attachable | Storage}, // R8UI
{vk::Format::eR16G16B16A16Sfloat, Attachable | Storage}, // RGBA16F
{vk::Format::eR16G16B16A16Unorm, Attachable | Storage}, // RGBA16U
+ {vk::Format::eR16G16B16A16Snorm, Attachable | Storage}, // RGBA16S
{vk::Format::eR16G16B16A16Uint, Attachable | Storage}, // RGBA16UI
{vk::Format::eB10G11R11UfloatPack32, Attachable | Storage}, // R11FG11FB10F
{vk::Format::eR32G32B32A32Uint, Attachable | Storage}, // RGBA32UI
@@ -256,6 +257,8 @@ vk::ShaderStageFlagBits ShaderStage(Tegra::Engines::ShaderType stage) {
return vk::ShaderStageFlagBits::eGeometry;
case Tegra::Engines::ShaderType::Fragment:
return vk::ShaderStageFlagBits::eFragment;
+ case Tegra::Engines::ShaderType::Compute:
+ return vk::ShaderStageFlagBits::eCompute;
}
UNIMPLEMENTED_MSG("Unimplemented shader stage={}", static_cast<u32>(stage));
return {};
@@ -331,6 +334,8 @@ vk::Format VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttr
return vk::Format::eR16G16B16Unorm;
case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
return vk::Format::eR16G16B16A16Unorm;
+ case Maxwell::VertexAttribute::Size::Size_10_10_10_2:
+ return vk::Format::eA2B10G10R10UnormPack32;
default:
break;
}
@@ -364,6 +369,10 @@ vk::Format VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttr
return vk::Format::eR8G8B8A8Uint;
case Maxwell::VertexAttribute::Size::Size_32:
return vk::Format::eR32Uint;
+ case Maxwell::VertexAttribute::Size::Size_32_32:
+ return vk::Format::eR32G32Uint;
+ case Maxwell::VertexAttribute::Size::Size_32_32_32:
+ return vk::Format::eR32G32B32Uint;
case Maxwell::VertexAttribute::Size::Size_32_32_32_32:
return vk::Format::eR32G32B32A32Uint;
default:
diff --git a/src/video_core/renderer_vulkan/vk_device.cpp b/src/video_core/renderer_vulkan/vk_device.cpp
index 3847bd722..28d2fbc4f 100644
--- a/src/video_core/renderer_vulkan/vk_device.cpp
+++ b/src/video_core/renderer_vulkan/vk_device.cpp
@@ -535,6 +535,7 @@ std::unordered_map<vk::Format, vk::FormatProperties> VKDevice::GetFormatProperti
vk::Format::eR32G32Sfloat,
vk::Format::eR32G32Uint,
vk::Format::eR16G16B16A16Uint,
+ vk::Format::eR16G16B16A16Snorm,
vk::Format::eR16G16B16A16Unorm,
vk::Format::eR16G16Unorm,
vk::Format::eR16G16Snorm,
diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
index 056ef495c..557b9d662 100644
--- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
@@ -179,10 +179,11 @@ Tegra::Engines::ConstBufferEngineInterface& CachedShader::GetEngine(
VKPipelineCache::VKPipelineCache(Core::System& system, RasterizerVulkan& rasterizer,
const VKDevice& device, VKScheduler& scheduler,
VKDescriptorPool& descriptor_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue)
+ VKUpdateDescriptorQueue& update_descriptor_queue,
+ VKRenderPassCache& renderpass_cache)
: RasterizerCache{rasterizer}, system{system}, device{device}, scheduler{scheduler},
descriptor_pool{descriptor_pool}, update_descriptor_queue{update_descriptor_queue},
- renderpass_cache(device) {}
+ renderpass_cache{renderpass_cache} {}
VKPipelineCache::~VKPipelineCache() = default;
@@ -191,7 +192,6 @@ std::array<Shader, Maxwell::MaxShaderProgram> VKPipelineCache::GetShaders() {
std::array<Shader, Maxwell::MaxShaderProgram> shaders;
for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
- const auto& shader_config = gpu.regs.shader_config[index];
const auto program{static_cast<Maxwell::ShaderProgram>(index)};
// Skip stages that are not enabled
diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h
index 21340c9a4..c4c112290 100644
--- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h
+++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h
@@ -161,7 +161,8 @@ public:
explicit VKPipelineCache(Core::System& system, RasterizerVulkan& rasterizer,
const VKDevice& device, VKScheduler& scheduler,
VKDescriptorPool& descriptor_pool,
- VKUpdateDescriptorQueue& update_descriptor_queue);
+ VKUpdateDescriptorQueue& update_descriptor_queue,
+ VKRenderPassCache& renderpass_cache);
~VKPipelineCache();
std::array<Shader, Maxwell::MaxShaderProgram> GetShaders();
@@ -184,8 +185,7 @@ private:
VKScheduler& scheduler;
VKDescriptorPool& descriptor_pool;
VKUpdateDescriptorQueue& update_descriptor_queue;
-
- VKRenderPassCache renderpass_cache;
+ VKRenderPassCache& renderpass_cache;
std::array<Shader, Maxwell::MaxShaderProgram> last_shaders;
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index f889019c1..58c69b786 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -287,12 +287,13 @@ RasterizerVulkan::RasterizerVulkan(Core::System& system, Core::Frontend::EmuWind
screen_info{screen_info}, device{device}, resource_manager{resource_manager},
memory_manager{memory_manager}, state_tracker{state_tracker}, scheduler{scheduler},
staging_pool(device, memory_manager, scheduler), descriptor_pool(device),
- update_descriptor_queue(device, scheduler),
+ update_descriptor_queue(device, scheduler), renderpass_cache(device),
quad_array_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
uint8_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue),
texture_cache(system, *this, device, resource_manager, memory_manager, scheduler,
staging_pool),
- pipeline_cache(system, *this, device, scheduler, descriptor_pool, update_descriptor_queue),
+ pipeline_cache(system, *this, device, scheduler, descriptor_pool, update_descriptor_queue,
+ renderpass_cache),
buffer_cache(*this, system, device, memory_manager, scheduler, staging_pool),
sampler_cache(device), query_cache(system, *this, device, scheduler) {
scheduler.SetQueryCache(query_cache);
@@ -365,13 +366,16 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) {
void RasterizerVulkan::Clear() {
MICROPROFILE_SCOPE(Vulkan_Clearing);
- query_cache.UpdateCounters();
-
const auto& gpu = system.GPU().Maxwell3D();
if (!system.GPU().Maxwell3D().ShouldExecute()) {
return;
}
+ sampled_views.clear();
+ image_views.clear();
+
+ query_cache.UpdateCounters();
+
const auto& regs = gpu.regs;
const bool use_color = regs.clear_buffers.R || regs.clear_buffers.G || regs.clear_buffers.B ||
regs.clear_buffers.A;
@@ -380,52 +384,54 @@ void RasterizerVulkan::Clear() {
if (!use_color && !use_depth && !use_stencil) {
return;
}
- // Clearing images requires to be out of a renderpass
- scheduler.RequestOutsideRenderPassOperationContext();
- // TODO(Rodrigo): Implement clears rendering a quad or using beginning a renderpass.
+ [[maybe_unused]] const auto texceptions = UpdateAttachments();
+ DEBUG_ASSERT(texceptions.none());
+ SetupImageTransitions(0, color_attachments, zeta_attachment);
- if (use_color) {
- View color_view;
- {
- MICROPROFILE_SCOPE(Vulkan_RenderTargets);
- color_view = texture_cache.GetColorBufferSurface(regs.clear_buffers.RT.Value(), false);
- }
+ const vk::RenderPass renderpass = renderpass_cache.GetRenderPass(GetRenderPassParams(0));
+ const auto [framebuffer, render_area] = ConfigureFramebuffers(renderpass);
+ scheduler.RequestRenderpass({renderpass, framebuffer, {{0, 0}, render_area}, 0, nullptr});
+
+ const auto& scissor = regs.scissor_test[0];
+ const vk::Offset2D scissor_offset(scissor.min_x, scissor.min_y);
+ vk::Extent2D scissor_extent{scissor.max_x - scissor.min_x, scissor.max_y - scissor.min_y};
+ scissor_extent.width = std::min(scissor_extent.width, render_area.width);
+ scissor_extent.height = std::min(scissor_extent.height, render_area.height);
- color_view->Transition(vk::ImageLayout::eTransferDstOptimal,
- vk::PipelineStageFlagBits::eTransfer,
- vk::AccessFlagBits::eTransferWrite);
+ const u32 layer = regs.clear_buffers.layer;
+ const vk::ClearRect clear_rect({scissor_offset, scissor_extent}, layer, 1);
+ if (use_color) {
const std::array clear_color = {regs.clear_color[0], regs.clear_color[1],
regs.clear_color[2], regs.clear_color[3]};
- const vk::ClearColorValue clear(clear_color);
- scheduler.Record([image = color_view->GetImage(),
- subresource = color_view->GetImageSubresourceRange(),
- clear](auto cmdbuf, auto& dld) {
- cmdbuf.clearColorImage(image, vk::ImageLayout::eTransferDstOptimal, clear, subresource,
- dld);
+ const vk::ClearValue clear_value{clear_color};
+ const u32 color_attachment = regs.clear_buffers.RT;
+ scheduler.Record([color_attachment, clear_value, clear_rect](auto cmdbuf, auto& dld) {
+ const vk::ClearAttachment attachment(vk::ImageAspectFlagBits::eColor, color_attachment,
+ clear_value);
+ cmdbuf.clearAttachments(1, &attachment, 1, &clear_rect, dld);
});
}
- if (use_depth || use_stencil) {
- View zeta_surface;
- {
- MICROPROFILE_SCOPE(Vulkan_RenderTargets);
- zeta_surface = texture_cache.GetDepthBufferSurface(false);
- }
- zeta_surface->Transition(vk::ImageLayout::eTransferDstOptimal,
- vk::PipelineStageFlagBits::eTransfer,
- vk::AccessFlagBits::eTransferWrite);
-
- const vk::ClearDepthStencilValue clear(regs.clear_depth,
- static_cast<u32>(regs.clear_stencil));
- scheduler.Record([image = zeta_surface->GetImage(),
- subresource = zeta_surface->GetImageSubresourceRange(),
- clear](auto cmdbuf, auto& dld) {
- cmdbuf.clearDepthStencilImage(image, vk::ImageLayout::eTransferDstOptimal, clear,
- subresource, dld);
- });
+ if (!use_depth && !use_stencil) {
+ return;
+ }
+ vk::ImageAspectFlags aspect_flags;
+ if (use_depth) {
+ aspect_flags |= vk::ImageAspectFlagBits::eDepth;
+ }
+ if (use_stencil) {
+ aspect_flags |= vk::ImageAspectFlagBits::eStencil;
}
+
+ scheduler.Record([clear_depth = regs.clear_depth, clear_stencil = regs.clear_stencil,
+ clear_rect, aspect_flags](auto cmdbuf, auto& dld) {
+ const vk::ClearDepthStencilValue clear_zeta(clear_depth, clear_stencil);
+ const vk::ClearValue clear_value{clear_zeta};
+ const vk::ClearAttachment attachment(aspect_flags, 0, clear_value);
+ cmdbuf.clearAttachments(1, &attachment, 1, &clear_rect, dld);
+ });
}
void RasterizerVulkan::DispatchCompute(GPUVAddr code_addr) {
@@ -542,8 +548,6 @@ bool RasterizerVulkan::AccelerateDisplay(const Tegra::FramebufferConfig& config,
// Verify that the cached surface is the same size and format as the requested framebuffer
const auto& params{surface->GetSurfaceParams()};
- const auto& pixel_format{
- VideoCore::Surface::PixelFormatFromGPUPixelFormat(config.pixel_format)};
ASSERT_MSG(params.width == config.width, "Framebuffer width is different");
ASSERT_MSG(params.height == config.height, "Framebuffer height is different");
@@ -1151,7 +1155,7 @@ std::size_t RasterizerVulkan::CalculateVertexArraysSize() const {
// This implementation assumes that all attributes are used in the shader.
const GPUVAddr start{regs.vertex_array[index].StartAddress()};
const GPUVAddr end{regs.vertex_array_limit[index].LimitAddress()};
- DEBUG_ASSERT(end > start);
+ DEBUG_ASSERT(end >= start);
size += (end - start + 1) * regs.vertex_array[index].enable;
}
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h
index b2e73d98d..3185868e9 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.h
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.h
@@ -253,6 +253,7 @@ private:
VKStagingBufferPool staging_pool;
VKDescriptorPool descriptor_pool;
VKUpdateDescriptorQueue update_descriptor_queue;
+ VKRenderPassCache renderpass_cache;
QuadArrayPass quad_array_pass;
Uint8Pass uint8_pass;
diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
index b2c298051..51ecb5567 100644
--- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
+++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
@@ -837,7 +837,7 @@ private:
Decorate(id, spv::Decoration::Offset, static_cast<u32>(tfb->offset));
}
- element += static_cast<u8>(num_components);
+ element = static_cast<u8>(static_cast<std::size_t>(element) + num_components);
}
}
diff --git a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
index d9ea3cc21..374959f82 100644
--- a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
+++ b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
@@ -100,7 +100,6 @@ void VKStagingBufferPool::ReleaseCache(bool host_visible) {
}
u64 VKStagingBufferPool::ReleaseLevel(StagingBuffersCache& cache, std::size_t log2) {
- static constexpr u64 epochs_to_destroy = 180;
static constexpr std::size_t deletions_per_tick = 16;
auto& staging = cache[log2];
@@ -108,6 +107,7 @@ u64 VKStagingBufferPool::ReleaseLevel(StagingBuffersCache& cache, std::size_t lo
const std::size_t old_size = entries.size();
const auto is_deleteable = [this](const auto& entry) {
+ static constexpr u64 epochs_to_destroy = 180;
return entry.last_epoch + epochs_to_destroy < epoch && !entry.watch.IsUsed();
};
const std::size_t begin_offset = staging.delete_index;
diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.cpp b/src/video_core/renderer_vulkan/vk_state_tracker.cpp
index d74e68b63..94a89e388 100644
--- a/src/video_core/renderer_vulkan/vk_state_tracker.cpp
+++ b/src/video_core/renderer_vulkan/vk_state_tracker.cpp
@@ -90,8 +90,6 @@ void StateTracker::Initialize() {
SetupDirtyBlendConstants(tables);
SetupDirtyDepthBounds(tables);
SetupDirtyStencilProperties(tables);
-
- SetupCommonOnWriteStores(dirty.on_write_stores);
}
void StateTracker::InvalidateCommandBufferState() {
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
index 73d92a5ae..26175921b 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
@@ -52,6 +52,9 @@ vk::ImageType SurfaceTargetToImage(SurfaceTarget target) {
return vk::ImageType::e2D;
case SurfaceTarget::Texture3D:
return vk::ImageType::e3D;
+ case SurfaceTarget::TextureBuffer:
+ UNREACHABLE();
+ return {};
}
UNREACHABLE_MSG("Unknown texture target={}", static_cast<u32>(target));
return {};
@@ -273,7 +276,6 @@ void CachedSurface::UploadImage(const std::vector<u8>& staging_buffer) {
for (u32 level = 0; level < params.num_levels; ++level) {
vk::BufferImageCopy copy = GetBufferImageCopy(level);
- const auto& dld = device.GetDispatchLoader();
if (image->GetAspectMask() ==
(vk::ImageAspectFlagBits::eDepth | vk::ImageAspectFlagBits::eStencil)) {
vk::BufferImageCopy depth = copy;
@@ -422,7 +424,6 @@ void VKTextureCache::ImageCopy(Surface& src_surface, Surface& dst_surface,
dst_base_layer, num_layers, copy_params.dest_level, 1, vk::PipelineStageFlagBits::eTransfer,
vk::AccessFlagBits::eTransferWrite, vk::ImageLayout::eTransferDstOptimal);
- const auto& dld{device.GetDispatchLoader()};
const vk::ImageSubresourceLayers src_subresource(
src_surface->GetAspectMask(), copy_params.source_level, copy_params.source_z, num_layers);
const vk::ImageSubresourceLayers dst_subresource(
@@ -458,7 +459,6 @@ void VKTextureCache::ImageBlit(View& src_view, View& dst_view,
dst_view->GetImageSubresourceLayers(), {dst_top_left, dst_bot_right});
const bool is_linear = copy_config.filter == Tegra::Engines::Fermi2D::Filter::Linear;
- const auto& dld{device.GetDispatchLoader()};
scheduler.Record([src_image = src_view->GetImage(), dst_image = dst_view->GetImage(), blit,
is_linear](auto cmdbuf, auto& dld) {
cmdbuf.blitImage(src_image, vk::ImageLayout::eTransferSrcOptimal, dst_image,
diff --git a/src/video_core/shader/decode/bfe.cpp b/src/video_core/shader/decode/bfe.cpp
index e02bcd097..8e3b46e8e 100644
--- a/src/video_core/shader/decode/bfe.cpp
+++ b/src/video_core/shader/decode/bfe.cpp
@@ -17,33 +17,60 @@ u32 ShaderIR::DecodeBfe(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
const auto opcode = OpCode::Decode(instr);
- UNIMPLEMENTED_IF(instr.bfe.negate_b);
-
Node op_a = GetRegister(instr.gpr8);
- op_a = GetOperandAbsNegInteger(op_a, false, instr.bfe.negate_a, false);
-
- switch (opcode->get().GetId()) {
- case OpCode::Id::BFE_IMM: {
- UNIMPLEMENTED_IF_MSG(instr.generates_cc,
- "Condition codes generation in BFE is not implemented");
+ Node op_b = [&] {
+ switch (opcode->get().GetId()) {
+ case OpCode::Id::BFE_R:
+ return GetRegister(instr.gpr20);
+ case OpCode::Id::BFE_C:
+ return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
+ case OpCode::Id::BFE_IMM:
+ return Immediate(instr.alu.GetSignedImm20_20());
+ default:
+ UNREACHABLE();
+ return Immediate(0);
+ }
+ }();
- const Node inner_shift_imm = Immediate(static_cast<u32>(instr.bfe.GetLeftShiftValue()));
- const Node outer_shift_imm =
- Immediate(static_cast<u32>(instr.bfe.GetLeftShiftValue() + instr.bfe.shift_position));
+ UNIMPLEMENTED_IF_MSG(instr.bfe.rd_cc, "Condition codes in BFE is not implemented");
- const Node inner_shift =
- Operation(OperationCode::ILogicalShiftLeft, NO_PRECISE, op_a, inner_shift_imm);
- const Node outer_shift =
- Operation(OperationCode::ILogicalShiftRight, NO_PRECISE, inner_shift, outer_shift_imm);
+ const bool is_signed = instr.bfe.is_signed;
- SetInternalFlagsFromInteger(bb, outer_shift, instr.generates_cc);
- SetRegister(bb, instr.gpr0, outer_shift);
- break;
- }
- default:
- UNIMPLEMENTED_MSG("Unhandled BFE instruction: {}", opcode->get().GetName());
+ // using reverse parallel method in
+ // https://graphics.stanford.edu/~seander/bithacks.html#ReverseParallel
+ // note for later if possible to implement faster method.
+ if (instr.bfe.brev) {
+ const auto swap = [&](u32 s, u32 mask) {
+ Node v1 =
+ SignedOperation(OperationCode::ILogicalShiftRight, is_signed, op_a, Immediate(s));
+ if (mask != 0) {
+ v1 = SignedOperation(OperationCode::IBitwiseAnd, is_signed, std::move(v1),
+ Immediate(mask));
+ }
+ Node v2 = op_a;
+ if (mask != 0) {
+ v2 = SignedOperation(OperationCode::IBitwiseAnd, is_signed, std::move(v2),
+ Immediate(mask));
+ }
+ v2 = SignedOperation(OperationCode::ILogicalShiftLeft, is_signed, std::move(v2),
+ Immediate(s));
+ return SignedOperation(OperationCode::IBitwiseOr, is_signed, std::move(v1),
+ std::move(v2));
+ };
+ op_a = swap(1, 0x55555555U);
+ op_a = swap(2, 0x33333333U);
+ op_a = swap(4, 0x0F0F0F0FU);
+ op_a = swap(8, 0x00FF00FFU);
+ op_a = swap(16, 0);
}
+ const auto offset = SignedOperation(OperationCode::IBitfieldExtract, is_signed, op_b,
+ Immediate(0), Immediate(8));
+ const auto bits = SignedOperation(OperationCode::IBitfieldExtract, is_signed, op_b,
+ Immediate(8), Immediate(8));
+ auto result = SignedOperation(OperationCode::IBitfieldExtract, is_signed, op_a, offset, bits);
+ SetRegister(bb, instr.gpr0, std::move(result));
+
return pc;
}
diff --git a/src/video_core/shader/decode/xmad.cpp b/src/video_core/shader/decode/xmad.cpp
index 206961909..fbd7e9a17 100644
--- a/src/video_core/shader/decode/xmad.cpp
+++ b/src/video_core/shader/decode/xmad.cpp
@@ -12,6 +12,7 @@ namespace VideoCommon::Shader {
using Tegra::Shader::Instruction;
using Tegra::Shader::OpCode;
+using Tegra::Shader::PredCondition;
u32 ShaderIR::DecodeXmad(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
@@ -63,15 +64,18 @@ u32 ShaderIR::DecodeXmad(NodeBlock& bb, u32 pc) {
}
}();
- op_a = BitfieldExtract(op_a, instr.xmad.high_a ? 16 : 0, 16);
+ op_a = SignedOperation(OperationCode::IBitfieldExtract, is_signed_a, std::move(op_a),
+ instr.xmad.high_a ? Immediate(16) : Immediate(0), Immediate(16));
const Node original_b = op_b;
- op_b = BitfieldExtract(op_b, is_high_b ? 16 : 0, 16);
+ op_b = SignedOperation(OperationCode::IBitfieldExtract, is_signed_b, std::move(op_b),
+ is_high_b ? Immediate(16) : Immediate(0), Immediate(16));
- // TODO(Rodrigo): Use an appropiate sign for this operation
- Node product = Operation(OperationCode::IMul, NO_PRECISE, op_a, op_b);
+ // we already check sign_a and sign_b is difference or not before so just use one in here.
+ Node product = SignedOperation(OperationCode::IMul, is_signed_a, op_a, op_b);
if (is_psl) {
- product = Operation(OperationCode::ILogicalShiftLeft, NO_PRECISE, product, Immediate(16));
+ product =
+ SignedOperation(OperationCode::ILogicalShiftLeft, is_signed_a, product, Immediate(16));
}
SetTemporary(bb, 0, product);
product = GetTemporary(0);
@@ -88,12 +92,40 @@ u32 ShaderIR::DecodeXmad(NodeBlock& bb, u32 pc) {
return BitfieldExtract(original_c, 16, 16);
case Tegra::Shader::XmadMode::CBcc: {
const Node shifted_b = SignedOperation(OperationCode::ILogicalShiftLeft, is_signed_b,
- NO_PRECISE, original_b, Immediate(16));
- return SignedOperation(OperationCode::IAdd, is_signed_c, NO_PRECISE, original_c,
- shifted_b);
+ original_b, Immediate(16));
+ return SignedOperation(OperationCode::IAdd, is_signed_c, original_c, shifted_b);
+ }
+ case Tegra::Shader::XmadMode::CSfu: {
+ const Node comp_a = GetPredicateComparisonInteger(PredCondition::Equal, is_signed_a,
+ op_a, Immediate(0));
+ const Node comp_b = GetPredicateComparisonInteger(PredCondition::Equal, is_signed_b,
+ op_b, Immediate(0));
+ const Node comp = Operation(OperationCode::LogicalOr, comp_a, comp_b);
+
+ const Node comp_minus_a = GetPredicateComparisonInteger(
+ PredCondition::NotEqual, is_signed_a,
+ SignedOperation(OperationCode::IBitwiseAnd, is_signed_a, op_a,
+ Immediate(0x80000000)),
+ Immediate(0));
+ const Node comp_minus_b = GetPredicateComparisonInteger(
+ PredCondition::NotEqual, is_signed_b,
+ SignedOperation(OperationCode::IBitwiseAnd, is_signed_b, op_b,
+ Immediate(0x80000000)),
+ Immediate(0));
+
+ Node new_c = Operation(
+ OperationCode::Select, comp_minus_a,
+ SignedOperation(OperationCode::IAdd, is_signed_c, original_c, Immediate(-65536)),
+ original_c);
+ new_c = Operation(
+ OperationCode::Select, comp_minus_b,
+ SignedOperation(OperationCode::IAdd, is_signed_c, new_c, Immediate(-65536)),
+ std::move(new_c));
+
+ return Operation(OperationCode::Select, comp, original_c, std::move(new_c));
}
default:
- UNIMPLEMENTED_MSG("Unhandled XMAD mode: {}", static_cast<u32>(instr.xmad.mode.Value()));
+ UNREACHABLE();
return Immediate(0);
}
}();
@@ -102,18 +134,19 @@ u32 ShaderIR::DecodeXmad(NodeBlock& bb, u32 pc) {
op_c = GetTemporary(1);
// TODO(Rodrigo): Use an appropiate sign for this operation
- Node sum = Operation(OperationCode::IAdd, product, op_c);
+ Node sum = SignedOperation(OperationCode::IAdd, is_signed_a, product, std::move(op_c));
SetTemporary(bb, 2, sum);
sum = GetTemporary(2);
if (is_merge) {
- const Node a = BitfieldExtract(sum, 0, 16);
- const Node b =
- Operation(OperationCode::ILogicalShiftLeft, NO_PRECISE, original_b, Immediate(16));
- sum = Operation(OperationCode::IBitwiseOr, NO_PRECISE, a, b);
+ const Node a = SignedOperation(OperationCode::IBitfieldExtract, is_signed_a, std::move(sum),
+ Immediate(0), Immediate(16));
+ const Node b = SignedOperation(OperationCode::ILogicalShiftLeft, is_signed_b, original_b,
+ Immediate(16));
+ sum = SignedOperation(OperationCode::IBitwiseOr, is_signed_a, a, b);
}
SetInternalFlagsFromInteger(bb, sum, instr.generates_cc);
- SetRegister(bb, instr.gpr0, sum);
+ SetRegister(bb, instr.gpr0, std::move(sum));
return pc;
}
diff --git a/src/video_core/shader/node_helper.cpp b/src/video_core/shader/node_helper.cpp
index b3dcd291c..76c56abb5 100644
--- a/src/video_core/shader/node_helper.cpp
+++ b/src/video_core/shader/node_helper.cpp
@@ -68,6 +68,8 @@ OperationCode SignedToUnsignedCode(OperationCode operation_code, bool is_signed)
return OperationCode::UBitwiseXor;
case OperationCode::IBitwiseNot:
return OperationCode::UBitwiseNot;
+ case OperationCode::IBitfieldExtract:
+ return OperationCode::UBitfieldExtract;
case OperationCode::IBitfieldInsert:
return OperationCode::UBitfieldInsert;
case OperationCode::IBitCount:
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
index 9707c353d..cc7181229 100644
--- a/src/video_core/surface.cpp
+++ b/src/video_core/surface.cpp
@@ -111,6 +111,8 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format)
return PixelFormat::RGBA16F;
case Tegra::RenderTargetFormat::RGBA16_UNORM:
return PixelFormat::RGBA16U;
+ case Tegra::RenderTargetFormat::RGBA16_SNORM:
+ return PixelFormat::RGBA16S;
case Tegra::RenderTargetFormat::RGBA16_UINT:
return PixelFormat::RGBA16UI;
case Tegra::RenderTargetFormat::RGBA32_FLOAT:
diff --git a/src/video_core/surface.h b/src/video_core/surface.h
index d88109e5a..ae8817465 100644
--- a/src/video_core/surface.h
+++ b/src/video_core/surface.h
@@ -25,82 +25,83 @@ enum class PixelFormat {
R8UI = 7,
RGBA16F = 8,
RGBA16U = 9,
- RGBA16UI = 10,
- R11FG11FB10F = 11,
- RGBA32UI = 12,
- DXT1 = 13,
- DXT23 = 14,
- DXT45 = 15,
- DXN1 = 16, // This is also known as BC4
- DXN2UNORM = 17,
- DXN2SNORM = 18,
- BC7U = 19,
- BC6H_UF16 = 20,
- BC6H_SF16 = 21,
- ASTC_2D_4X4 = 22,
- BGRA8 = 23,
- RGBA32F = 24,
- RG32F = 25,
- R32F = 26,
- R16F = 27,
- R16U = 28,
- R16S = 29,
- R16UI = 30,
- R16I = 31,
- RG16 = 32,
- RG16F = 33,
- RG16UI = 34,
- RG16I = 35,
- RG16S = 36,
- RGB32F = 37,
- RGBA8_SRGB = 38,
- RG8U = 39,
- RG8S = 40,
- RG32UI = 41,
- RGBX16F = 42,
- R32UI = 43,
- R32I = 44,
- ASTC_2D_8X8 = 45,
- ASTC_2D_8X5 = 46,
- ASTC_2D_5X4 = 47,
- BGRA8_SRGB = 48,
- DXT1_SRGB = 49,
- DXT23_SRGB = 50,
- DXT45_SRGB = 51,
- BC7U_SRGB = 52,
- R4G4B4A4U = 53,
- ASTC_2D_4X4_SRGB = 54,
- ASTC_2D_8X8_SRGB = 55,
- ASTC_2D_8X5_SRGB = 56,
- ASTC_2D_5X4_SRGB = 57,
- ASTC_2D_5X5 = 58,
- ASTC_2D_5X5_SRGB = 59,
- ASTC_2D_10X8 = 60,
- ASTC_2D_10X8_SRGB = 61,
- ASTC_2D_6X6 = 62,
- ASTC_2D_6X6_SRGB = 63,
- ASTC_2D_10X10 = 64,
- ASTC_2D_10X10_SRGB = 65,
- ASTC_2D_12X12 = 66,
- ASTC_2D_12X12_SRGB = 67,
- ASTC_2D_8X6 = 68,
- ASTC_2D_8X6_SRGB = 69,
- ASTC_2D_6X5 = 70,
- ASTC_2D_6X5_SRGB = 71,
- E5B9G9R9F = 72,
+ RGBA16S = 10,
+ RGBA16UI = 11,
+ R11FG11FB10F = 12,
+ RGBA32UI = 13,
+ DXT1 = 14,
+ DXT23 = 15,
+ DXT45 = 16,
+ DXN1 = 17, // This is also known as BC4
+ DXN2UNORM = 18,
+ DXN2SNORM = 19,
+ BC7U = 20,
+ BC6H_UF16 = 21,
+ BC6H_SF16 = 22,
+ ASTC_2D_4X4 = 23,
+ BGRA8 = 24,
+ RGBA32F = 25,
+ RG32F = 26,
+ R32F = 27,
+ R16F = 28,
+ R16U = 29,
+ R16S = 30,
+ R16UI = 31,
+ R16I = 32,
+ RG16 = 33,
+ RG16F = 34,
+ RG16UI = 35,
+ RG16I = 36,
+ RG16S = 37,
+ RGB32F = 38,
+ RGBA8_SRGB = 39,
+ RG8U = 40,
+ RG8S = 41,
+ RG32UI = 42,
+ RGBX16F = 43,
+ R32UI = 44,
+ R32I = 45,
+ ASTC_2D_8X8 = 46,
+ ASTC_2D_8X5 = 47,
+ ASTC_2D_5X4 = 48,
+ BGRA8_SRGB = 49,
+ DXT1_SRGB = 50,
+ DXT23_SRGB = 51,
+ DXT45_SRGB = 52,
+ BC7U_SRGB = 53,
+ R4G4B4A4U = 54,
+ ASTC_2D_4X4_SRGB = 55,
+ ASTC_2D_8X8_SRGB = 56,
+ ASTC_2D_8X5_SRGB = 57,
+ ASTC_2D_5X4_SRGB = 58,
+ ASTC_2D_5X5 = 59,
+ ASTC_2D_5X5_SRGB = 60,
+ ASTC_2D_10X8 = 61,
+ ASTC_2D_10X8_SRGB = 62,
+ ASTC_2D_6X6 = 63,
+ ASTC_2D_6X6_SRGB = 64,
+ ASTC_2D_10X10 = 65,
+ ASTC_2D_10X10_SRGB = 66,
+ ASTC_2D_12X12 = 67,
+ ASTC_2D_12X12_SRGB = 68,
+ ASTC_2D_8X6 = 69,
+ ASTC_2D_8X6_SRGB = 70,
+ ASTC_2D_6X5 = 71,
+ ASTC_2D_6X5_SRGB = 72,
+ E5B9G9R9F = 73,
MaxColorFormat,
// Depth formats
- Z32F = 73,
- Z16 = 74,
+ Z32F = 74,
+ Z16 = 75,
MaxDepthFormat,
// DepthStencil formats
- Z24S8 = 75,
- S8Z24 = 76,
- Z32FS8 = 77,
+ Z24S8 = 76,
+ S8Z24 = 77,
+ Z32FS8 = 78,
MaxDepthStencilFormat,
@@ -138,6 +139,7 @@ constexpr std::array<u32, MaxPixelFormat> compression_factor_shift_table = {{
0, // R8UI
0, // RGBA16F
0, // RGBA16U
+ 0, // RGBA16S
0, // RGBA16UI
0, // R11FG11FB10F
0, // RGBA32UI
@@ -235,6 +237,7 @@ constexpr std::array<u32, MaxPixelFormat> block_width_table = {{
1, // R8UI
1, // RGBA16F
1, // RGBA16U
+ 1, // RGBA16S
1, // RGBA16UI
1, // R11FG11FB10F
1, // RGBA32UI
@@ -324,6 +327,7 @@ constexpr std::array<u32, MaxPixelFormat> block_height_table = {{
1, // R8UI
1, // RGBA16F
1, // RGBA16U
+ 1, // RGBA16S
1, // RGBA16UI
1, // R11FG11FB10F
1, // RGBA32UI
@@ -413,6 +417,7 @@ constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
8, // R8UI
64, // RGBA16F
64, // RGBA16U
+ 64, // RGBA16S
64, // RGBA16UI
32, // R11FG11FB10F
128, // RGBA32UI
@@ -517,6 +522,7 @@ constexpr std::array<SurfaceCompression, MaxPixelFormat> compression_type_table
SurfaceCompression::None, // R8UI
SurfaceCompression::None, // RGBA16F
SurfaceCompression::None, // RGBA16U
+ SurfaceCompression::None, // RGBA16S
SurfaceCompression::None, // RGBA16UI
SurfaceCompression::None, // R11FG11FB10F
SurfaceCompression::None, // RGBA32UI
diff --git a/src/video_core/texture_cache/format_lookup_table.cpp b/src/video_core/texture_cache/format_lookup_table.cpp
index cc3ad8417..e151c26c4 100644
--- a/src/video_core/texture_cache/format_lookup_table.cpp
+++ b/src/video_core/texture_cache/format_lookup_table.cpp
@@ -41,7 +41,7 @@ struct Table {
ComponentType alpha_component;
bool is_srgb;
};
-constexpr std::array<Table, 75> DefinitionTable = {{
+constexpr std::array<Table, 76> DefinitionTable = {{
{TextureFormat::A8R8G8B8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ABGR8U},
{TextureFormat::A8R8G8B8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::ABGR8S},
{TextureFormat::A8R8G8B8, C, UINT, UINT, UINT, UINT, PixelFormat::ABGR8UI},
@@ -61,6 +61,7 @@ constexpr std::array<Table, 75> DefinitionTable = {{
{TextureFormat::G8R8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::RG8U},
{TextureFormat::G8R8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::RG8S},
+ {TextureFormat::R16_G16_B16_A16, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::RGBA16S},
{TextureFormat::R16_G16_B16_A16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::RGBA16U},
{TextureFormat::R16_G16_B16_A16, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::RGBA16F},
{TextureFormat::R16_G16_B16_A16, C, UINT, UINT, UINT, UINT, PixelFormat::RGBA16UI},
diff --git a/src/video_core/textures/astc.cpp b/src/video_core/textures/astc.cpp
index 33bd31865..062b4f252 100644
--- a/src/video_core/textures/astc.cpp
+++ b/src/video_core/textures/astc.cpp
@@ -17,26 +17,37 @@
#include <algorithm>
#include <cassert>
-#include <cstdint>
#include <cstring>
#include <vector>
+#include "common/common_types.h"
+
#include "video_core/textures/astc.h"
+namespace {
+
+/// Count the number of bits set in a number.
+constexpr u32 Popcnt(u32 n) {
+ u32 c = 0;
+ for (; n; c++) {
+ n &= n - 1;
+ }
+ return c;
+}
+
+} // Anonymous namespace
+
class InputBitStream {
public:
- explicit InputBitStream(const unsigned char* ptr, int start_offset = 0)
+ explicit InputBitStream(const u8* ptr, std::size_t start_offset = 0)
: m_CurByte(ptr), m_NextBit(start_offset % 8) {}
- ~InputBitStream() = default;
-
- int GetBitsRead() const {
+ std::size_t GetBitsRead() const {
return m_BitsRead;
}
- int ReadBit() {
-
- int bit = *m_CurByte >> m_NextBit++;
+ u32 ReadBit() {
+ u32 bit = *m_CurByte >> m_NextBit++;
while (m_NextBit >= 8) {
m_NextBit -= 8;
m_CurByte++;
@@ -46,57 +57,66 @@ public:
return bit & 1;
}
- unsigned int ReadBits(unsigned int nBits) {
- unsigned int ret = 0;
- for (unsigned int i = 0; i < nBits; i++) {
+ u32 ReadBits(std::size_t nBits) {
+ u32 ret = 0;
+ for (std::size_t i = 0; i < nBits; ++i) {
+ ret |= (ReadBit() & 1) << i;
+ }
+ return ret;
+ }
+
+ template <std::size_t nBits>
+ u32 ReadBits() {
+ u32 ret = 0;
+ for (std::size_t i = 0; i < nBits; ++i) {
ret |= (ReadBit() & 1) << i;
}
return ret;
}
private:
- const unsigned char* m_CurByte;
- int m_NextBit = 0;
- int m_BitsRead = 0;
+ const u8* m_CurByte;
+ std::size_t m_NextBit = 0;
+ std::size_t m_BitsRead = 0;
};
class OutputBitStream {
public:
- explicit OutputBitStream(unsigned char* ptr, int nBits = 0, int start_offset = 0)
+ explicit OutputBitStream(u8* ptr, s32 nBits = 0, s32 start_offset = 0)
: m_NumBits(nBits), m_CurByte(ptr), m_NextBit(start_offset % 8) {}
~OutputBitStream() = default;
- int GetBitsWritten() const {
+ s32 GetBitsWritten() const {
return m_BitsWritten;
}
- void WriteBitsR(unsigned int val, unsigned int nBits) {
- for (unsigned int i = 0; i < nBits; i++) {
+ void WriteBitsR(u32 val, u32 nBits) {
+ for (u32 i = 0; i < nBits; i++) {
WriteBit((val >> (nBits - i - 1)) & 1);
}
}
- void WriteBits(unsigned int val, unsigned int nBits) {
- for (unsigned int i = 0; i < nBits; i++) {
+ void WriteBits(u32 val, u32 nBits) {
+ for (u32 i = 0; i < nBits; i++) {
WriteBit((val >> i) & 1);
}
}
private:
- void WriteBit(int b) {
+ void WriteBit(s32 b) {
if (done)
return;
- const unsigned int mask = 1 << m_NextBit++;
+ const u32 mask = 1 << m_NextBit++;
// clear the bit
- *m_CurByte &= static_cast<unsigned char>(~mask);
+ *m_CurByte &= static_cast<u8>(~mask);
// Write the bit, if necessary
if (b)
- *m_CurByte |= static_cast<unsigned char>(mask);
+ *m_CurByte |= static_cast<u8>(mask);
// Next byte?
if (m_NextBit >= 8) {
@@ -107,10 +127,10 @@ private:
done = done || ++m_BitsWritten >= m_NumBits;
}
- int m_BitsWritten = 0;
- const int m_NumBits;
- unsigned char* m_CurByte;
- int m_NextBit = 0;
+ s32 m_BitsWritten = 0;
+ const s32 m_NumBits;
+ u8* m_CurByte;
+ s32 m_NextBit = 0;
bool done = false;
};
@@ -123,20 +143,20 @@ public:
Bits(const Bits&) = delete;
Bits& operator=(const Bits&) = delete;
- uint8_t operator[](uint32_t bitPos) const {
- return static_cast<uint8_t>((m_Bits >> bitPos) & 1);
+ u8 operator[](u32 bitPos) const {
+ return static_cast<u8>((m_Bits >> bitPos) & 1);
}
- IntType operator()(uint32_t start, uint32_t end) const {
+ IntType operator()(u32 start, u32 end) const {
if (start == end) {
return (*this)[start];
} else if (start > end) {
- uint32_t t = start;
+ u32 t = start;
start = end;
end = t;
}
- uint64_t mask = (1 << (end - start + 1)) - 1;
+ u64 mask = (1 << (end - start + 1)) - 1;
return (m_Bits >> start) & static_cast<IntType>(mask);
}
@@ -144,273 +164,236 @@ private:
const IntType& m_Bits;
};
-enum EIntegerEncoding { eIntegerEncoding_JustBits, eIntegerEncoding_Quint, eIntegerEncoding_Trit };
-
-class IntegerEncodedValue {
-private:
- const EIntegerEncoding m_Encoding;
- const uint32_t m_NumBits;
- uint32_t m_BitValue;
- union {
- uint32_t m_QuintValue;
- uint32_t m_TritValue;
- };
+enum class IntegerEncoding { JustBits, Qus32, Trit };
-public:
- // Jank, but we're not doing any heavy lifting in this class, so it's
- // probably OK. It allows us to use these in std::vectors...
- IntegerEncodedValue& operator=(const IntegerEncodedValue& other) {
- new (this) IntegerEncodedValue(other);
- return *this;
- }
+struct IntegerEncodedValue {
+ constexpr IntegerEncodedValue() = default;
- IntegerEncodedValue(EIntegerEncoding encoding, uint32_t numBits)
- : m_Encoding(encoding), m_NumBits(numBits) {}
+ constexpr IntegerEncodedValue(IntegerEncoding encoding_, u32 num_bits_)
+ : encoding{encoding_}, num_bits{num_bits_} {}
- EIntegerEncoding GetEncoding() const {
- return m_Encoding;
- }
- uint32_t BaseBitLength() const {
- return m_NumBits;
- }
-
- uint32_t GetBitValue() const {
- return m_BitValue;
- }
- void SetBitValue(uint32_t val) {
- m_BitValue = val;
- }
-
- uint32_t GetTritValue() const {
- return m_TritValue;
- }
- void SetTritValue(uint32_t val) {
- m_TritValue = val;
- }
-
- uint32_t GetQuintValue() const {
- return m_QuintValue;
- }
- void SetQuintValue(uint32_t val) {
- m_QuintValue = val;
- }
-
- bool MatchesEncoding(const IntegerEncodedValue& other) const {
- return m_Encoding == other.m_Encoding && m_NumBits == other.m_NumBits;
+ constexpr bool MatchesEncoding(const IntegerEncodedValue& other) const {
+ return encoding == other.encoding && num_bits == other.num_bits;
}
// Returns the number of bits required to encode nVals values.
- uint32_t GetBitLength(uint32_t nVals) const {
- uint32_t totalBits = m_NumBits * nVals;
- if (m_Encoding == eIntegerEncoding_Trit) {
+ u32 GetBitLength(u32 nVals) const {
+ u32 totalBits = num_bits * nVals;
+ if (encoding == IntegerEncoding::Trit) {
totalBits += (nVals * 8 + 4) / 5;
- } else if (m_Encoding == eIntegerEncoding_Quint) {
+ } else if (encoding == IntegerEncoding::Qus32) {
totalBits += (nVals * 7 + 2) / 3;
}
return totalBits;
}
- // Count the number of bits set in a number.
- static inline uint32_t Popcnt(uint32_t n) {
- uint32_t c;
- for (c = 0; n; c++) {
- n &= n - 1;
+ IntegerEncoding encoding{};
+ u32 num_bits = 0;
+ u32 bit_value = 0;
+ union {
+ u32 qus32_value = 0;
+ u32 trit_value;
+ };
+};
+
+static void DecodeTritBlock(InputBitStream& bits, std::vector<IntegerEncodedValue>& result,
+ u32 nBitsPerValue) {
+ // Implement the algorithm in section C.2.12
+ u32 m[5];
+ u32 t[5];
+ u32 T;
+
+ // Read the trit encoded block according to
+ // table C.2.14
+ m[0] = bits.ReadBits(nBitsPerValue);
+ T = bits.ReadBits<2>();
+ m[1] = bits.ReadBits(nBitsPerValue);
+ T |= bits.ReadBits<2>() << 2;
+ m[2] = bits.ReadBits(nBitsPerValue);
+ T |= bits.ReadBit() << 4;
+ m[3] = bits.ReadBits(nBitsPerValue);
+ T |= bits.ReadBits<2>() << 5;
+ m[4] = bits.ReadBits(nBitsPerValue);
+ T |= bits.ReadBit() << 7;
+
+ u32 C = 0;
+
+ Bits<u32> Tb(T);
+ if (Tb(2, 4) == 7) {
+ C = (Tb(5, 7) << 2) | Tb(0, 1);
+ t[4] = t[3] = 2;
+ } else {
+ C = Tb(0, 4);
+ if (Tb(5, 6) == 3) {
+ t[4] = 2;
+ t[3] = Tb[7];
+ } else {
+ t[4] = Tb[7];
+ t[3] = Tb(5, 6);
}
- return c;
}
- // Returns a new instance of this struct that corresponds to the
- // can take no more than maxval values
- static IntegerEncodedValue CreateEncoding(uint32_t maxVal) {
- while (maxVal > 0) {
- uint32_t check = maxVal + 1;
-
- // Is maxVal a power of two?
- if (!(check & (check - 1))) {
- return IntegerEncodedValue(eIntegerEncoding_JustBits, Popcnt(maxVal));
- }
-
- // Is maxVal of the type 3*2^n - 1?
- if ((check % 3 == 0) && !((check / 3) & ((check / 3) - 1))) {
- return IntegerEncodedValue(eIntegerEncoding_Trit, Popcnt(check / 3 - 1));
- }
+ Bits<u32> Cb(C);
+ if (Cb(0, 1) == 3) {
+ t[2] = 2;
+ t[1] = Cb[4];
+ t[0] = (Cb[3] << 1) | (Cb[2] & ~Cb[3]);
+ } else if (Cb(2, 3) == 3) {
+ t[2] = 2;
+ t[1] = 2;
+ t[0] = Cb(0, 1);
+ } else {
+ t[2] = Cb[4];
+ t[1] = Cb(2, 3);
+ t[0] = (Cb[1] << 1) | (Cb[0] & ~Cb[1]);
+ }
- // Is maxVal of the type 5*2^n - 1?
- if ((check % 5 == 0) && !((check / 5) & ((check / 5) - 1))) {
- return IntegerEncodedValue(eIntegerEncoding_Quint, Popcnt(check / 5 - 1));
- }
+ for (std::size_t i = 0; i < 5; ++i) {
+ IntegerEncodedValue& val = result.emplace_back(IntegerEncoding::Trit, nBitsPerValue);
+ val.bit_value = m[i];
+ val.trit_value = t[i];
+ }
+}
- // Apparently it can't be represented with a bounded integer sequence...
- // just iterate.
- maxVal--;
+static void DecodeQus32Block(InputBitStream& bits, std::vector<IntegerEncodedValue>& result,
+ u32 nBitsPerValue) {
+ // Implement the algorithm in section C.2.12
+ u32 m[3];
+ u32 q[3];
+ u32 Q;
+
+ // Read the trit encoded block according to
+ // table C.2.15
+ m[0] = bits.ReadBits(nBitsPerValue);
+ Q = bits.ReadBits<3>();
+ m[1] = bits.ReadBits(nBitsPerValue);
+ Q |= bits.ReadBits<2>() << 3;
+ m[2] = bits.ReadBits(nBitsPerValue);
+ Q |= bits.ReadBits<2>() << 5;
+
+ Bits<u32> Qb(Q);
+ if (Qb(1, 2) == 3 && Qb(5, 6) == 0) {
+ q[0] = q[1] = 4;
+ q[2] = (Qb[0] << 2) | ((Qb[4] & ~Qb[0]) << 1) | (Qb[3] & ~Qb[0]);
+ } else {
+ u32 C = 0;
+ if (Qb(1, 2) == 3) {
+ q[2] = 4;
+ C = (Qb(3, 4) << 3) | ((~Qb(5, 6) & 3) << 1) | Qb[0];
+ } else {
+ q[2] = Qb(5, 6);
+ C = Qb(0, 4);
}
- return IntegerEncodedValue(eIntegerEncoding_JustBits, 0);
- }
-
- // Fills result with the values that are encoded in the given
- // bitstream. We must know beforehand what the maximum possible
- // value is, and how many values we're decoding.
- static void DecodeIntegerSequence(std::vector<IntegerEncodedValue>& result,
- InputBitStream& bits, uint32_t maxRange, uint32_t nValues) {
- // Determine encoding parameters
- IntegerEncodedValue val = IntegerEncodedValue::CreateEncoding(maxRange);
-
- // Start decoding
- uint32_t nValsDecoded = 0;
- while (nValsDecoded < nValues) {
- switch (val.GetEncoding()) {
- case eIntegerEncoding_Quint:
- DecodeQuintBlock(bits, result, val.BaseBitLength());
- nValsDecoded += 3;
- break;
- case eIntegerEncoding_Trit:
- DecodeTritBlock(bits, result, val.BaseBitLength());
- nValsDecoded += 5;
- break;
-
- case eIntegerEncoding_JustBits:
- val.SetBitValue(bits.ReadBits(val.BaseBitLength()));
- result.push_back(val);
- nValsDecoded++;
- break;
- }
+ Bits<u32> Cb(C);
+ if (Cb(0, 2) == 5) {
+ q[1] = 4;
+ q[0] = Cb(3, 4);
+ } else {
+ q[1] = Cb(3, 4);
+ q[0] = Cb(0, 2);
}
}
-private:
- static void DecodeTritBlock(InputBitStream& bits, std::vector<IntegerEncodedValue>& result,
- uint32_t nBitsPerValue) {
- // Implement the algorithm in section C.2.12
- uint32_t m[5];
- uint32_t t[5];
- uint32_t T;
-
- // Read the trit encoded block according to
- // table C.2.14
- m[0] = bits.ReadBits(nBitsPerValue);
- T = bits.ReadBits(2);
- m[1] = bits.ReadBits(nBitsPerValue);
- T |= bits.ReadBits(2) << 2;
- m[2] = bits.ReadBits(nBitsPerValue);
- T |= bits.ReadBit() << 4;
- m[3] = bits.ReadBits(nBitsPerValue);
- T |= bits.ReadBits(2) << 5;
- m[4] = bits.ReadBits(nBitsPerValue);
- T |= bits.ReadBit() << 7;
-
- uint32_t C = 0;
-
- Bits<uint32_t> Tb(T);
- if (Tb(2, 4) == 7) {
- C = (Tb(5, 7) << 2) | Tb(0, 1);
- t[4] = t[3] = 2;
- } else {
- C = Tb(0, 4);
- if (Tb(5, 6) == 3) {
- t[4] = 2;
- t[3] = Tb[7];
- } else {
- t[4] = Tb[7];
- t[3] = Tb(5, 6);
- }
+ for (std::size_t i = 0; i < 3; ++i) {
+ IntegerEncodedValue& val = result.emplace_back(IntegerEncoding::Qus32, nBitsPerValue);
+ val.bit_value = m[i];
+ val.qus32_value = q[i];
+ }
+}
+
+// Returns a new instance of this struct that corresponds to the
+// can take no more than maxval values
+static constexpr IntegerEncodedValue CreateEncoding(u32 maxVal) {
+ while (maxVal > 0) {
+ u32 check = maxVal + 1;
+
+ // Is maxVal a power of two?
+ if (!(check & (check - 1))) {
+ return IntegerEncodedValue(IntegerEncoding::JustBits, Popcnt(maxVal));
}
- Bits<uint32_t> Cb(C);
- if (Cb(0, 1) == 3) {
- t[2] = 2;
- t[1] = Cb[4];
- t[0] = (Cb[3] << 1) | (Cb[2] & ~Cb[3]);
- } else if (Cb(2, 3) == 3) {
- t[2] = 2;
- t[1] = 2;
- t[0] = Cb(0, 1);
- } else {
- t[2] = Cb[4];
- t[1] = Cb(2, 3);
- t[0] = (Cb[1] << 1) | (Cb[0] & ~Cb[1]);
+ // Is maxVal of the type 3*2^n - 1?
+ if ((check % 3 == 0) && !((check / 3) & ((check / 3) - 1))) {
+ return IntegerEncodedValue(IntegerEncoding::Trit, Popcnt(check / 3 - 1));
}
- for (uint32_t i = 0; i < 5; i++) {
- IntegerEncodedValue val(eIntegerEncoding_Trit, nBitsPerValue);
- val.SetBitValue(m[i]);
- val.SetTritValue(t[i]);
- result.push_back(val);
+ // Is maxVal of the type 5*2^n - 1?
+ if ((check % 5 == 0) && !((check / 5) & ((check / 5) - 1))) {
+ return IntegerEncodedValue(IntegerEncoding::Qus32, Popcnt(check / 5 - 1));
}
+
+ // Apparently it can't be represented with a bounded integer sequence...
+ // just iterate.
+ maxVal--;
}
+ return IntegerEncodedValue(IntegerEncoding::JustBits, 0);
+}
- static void DecodeQuintBlock(InputBitStream& bits, std::vector<IntegerEncodedValue>& result,
- uint32_t nBitsPerValue) {
- // Implement the algorithm in section C.2.12
- uint32_t m[3];
- uint32_t q[3];
- uint32_t Q;
-
- // Read the trit encoded block according to
- // table C.2.15
- m[0] = bits.ReadBits(nBitsPerValue);
- Q = bits.ReadBits(3);
- m[1] = bits.ReadBits(nBitsPerValue);
- Q |= bits.ReadBits(2) << 3;
- m[2] = bits.ReadBits(nBitsPerValue);
- Q |= bits.ReadBits(2) << 5;
-
- Bits<uint32_t> Qb(Q);
- if (Qb(1, 2) == 3 && Qb(5, 6) == 0) {
- q[0] = q[1] = 4;
- q[2] = (Qb[0] << 2) | ((Qb[4] & ~Qb[0]) << 1) | (Qb[3] & ~Qb[0]);
- } else {
- uint32_t C = 0;
- if (Qb(1, 2) == 3) {
- q[2] = 4;
- C = (Qb(3, 4) << 3) | ((~Qb(5, 6) & 3) << 1) | Qb[0];
- } else {
- q[2] = Qb(5, 6);
- C = Qb(0, 4);
- }
+static constexpr std::array<IntegerEncodedValue, 256> MakeEncodedValues() {
+ std::array<IntegerEncodedValue, 256> encodings{};
+ for (std::size_t i = 0; i < encodings.size(); ++i) {
+ encodings[i] = CreateEncoding(static_cast<u32>(i));
+ }
+ return encodings;
+}
- Bits<uint32_t> Cb(C);
- if (Cb(0, 2) == 5) {
- q[1] = 4;
- q[0] = Cb(3, 4);
- } else {
- q[1] = Cb(3, 4);
- q[0] = Cb(0, 2);
- }
- }
+static constexpr std::array EncodingsValues = MakeEncodedValues();
+
+// Fills result with the values that are encoded in the given
+// bitstream. We must know beforehand what the maximum possible
+// value is, and how many values we're decoding.
+static void DecodeIntegerSequence(std::vector<IntegerEncodedValue>& result, InputBitStream& bits,
+ u32 maxRange, u32 nValues) {
+ // Determine encoding parameters
+ IntegerEncodedValue val = EncodingsValues[maxRange];
+
+ // Start decoding
+ u32 nValsDecoded = 0;
+ while (nValsDecoded < nValues) {
+ switch (val.encoding) {
+ case IntegerEncoding::Qus32:
+ DecodeQus32Block(bits, result, val.num_bits);
+ nValsDecoded += 3;
+ break;
+
+ case IntegerEncoding::Trit:
+ DecodeTritBlock(bits, result, val.num_bits);
+ nValsDecoded += 5;
+ break;
- for (uint32_t i = 0; i < 3; i++) {
- IntegerEncodedValue val(eIntegerEncoding_Quint, nBitsPerValue);
- val.m_BitValue = m[i];
- val.m_QuintValue = q[i];
+ case IntegerEncoding::JustBits:
+ val.bit_value = bits.ReadBits(val.num_bits);
result.push_back(val);
+ nValsDecoded++;
+ break;
}
}
-};
+}
namespace ASTCC {
struct TexelWeightParams {
- uint32_t m_Width = 0;
- uint32_t m_Height = 0;
+ u32 m_Width = 0;
+ u32 m_Height = 0;
bool m_bDualPlane = false;
- uint32_t m_MaxWeight = 0;
+ u32 m_MaxWeight = 0;
bool m_bError = false;
bool m_bVoidExtentLDR = false;
bool m_bVoidExtentHDR = false;
- uint32_t GetPackedBitSize() const {
+ u32 GetPackedBitSize() const {
// How many indices do we have?
- uint32_t nIdxs = m_Height * m_Width;
+ u32 nIdxs = m_Height * m_Width;
if (m_bDualPlane) {
nIdxs *= 2;
}
- return IntegerEncodedValue::CreateEncoding(m_MaxWeight).GetBitLength(nIdxs);
+ return EncodingsValues[m_MaxWeight].GetBitLength(nIdxs);
}
- uint32_t GetNumWeightValues() const {
- uint32_t ret = m_Width * m_Height;
+ u32 GetNumWeightValues() const {
+ u32 ret = m_Width * m_Height;
if (m_bDualPlane) {
ret *= 2;
}
@@ -422,7 +405,7 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
TexelWeightParams params;
// Read the entire block mode all at once
- uint16_t modeBits = static_cast<uint16_t>(strm.ReadBits(11));
+ u16 modeBits = static_cast<u16>(strm.ReadBits<11>());
// Does this match the void extent block mode?
if ((modeBits & 0x01FF) == 0x1FC) {
@@ -457,7 +440,7 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
// of the block mode. Layout is determined by a number
// between 0 and 9 corresponding to table C.2.8 of the
// ASTC spec.
- uint32_t layout = 0;
+ u32 layout = 0;
if ((modeBits & 0x1) || (modeBits & 0x2)) {
// layout is in [0-4]
@@ -509,7 +492,7 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
assert(layout < 10);
// Determine R
- uint32_t R = !!(modeBits & 0x10);
+ u32 R = !!(modeBits & 0x10);
if (layout < 5) {
R |= (modeBits & 0x3) << 1;
} else {
@@ -520,54 +503,54 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
// Determine width & height
switch (layout) {
case 0: {
- uint32_t A = (modeBits >> 5) & 0x3;
- uint32_t B = (modeBits >> 7) & 0x3;
+ u32 A = (modeBits >> 5) & 0x3;
+ u32 B = (modeBits >> 7) & 0x3;
params.m_Width = B + 4;
params.m_Height = A + 2;
break;
}
case 1: {
- uint32_t A = (modeBits >> 5) & 0x3;
- uint32_t B = (modeBits >> 7) & 0x3;
+ u32 A = (modeBits >> 5) & 0x3;
+ u32 B = (modeBits >> 7) & 0x3;
params.m_Width = B + 8;
params.m_Height = A + 2;
break;
}
case 2: {
- uint32_t A = (modeBits >> 5) & 0x3;
- uint32_t B = (modeBits >> 7) & 0x3;
+ u32 A = (modeBits >> 5) & 0x3;
+ u32 B = (modeBits >> 7) & 0x3;
params.m_Width = A + 2;
params.m_Height = B + 8;
break;
}
case 3: {
- uint32_t A = (modeBits >> 5) & 0x3;
- uint32_t B = (modeBits >> 7) & 0x1;
+ u32 A = (modeBits >> 5) & 0x3;
+ u32 B = (modeBits >> 7) & 0x1;
params.m_Width = A + 2;
params.m_Height = B + 6;
break;
}
case 4: {
- uint32_t A = (modeBits >> 5) & 0x3;
- uint32_t B = (modeBits >> 7) & 0x1;
+ u32 A = (modeBits >> 5) & 0x3;
+ u32 B = (modeBits >> 7) & 0x1;
params.m_Width = B + 2;
params.m_Height = A + 2;
break;
}
case 5: {
- uint32_t A = (modeBits >> 5) & 0x3;
+ u32 A = (modeBits >> 5) & 0x3;
params.m_Width = 12;
params.m_Height = A + 2;
break;
}
case 6: {
- uint32_t A = (modeBits >> 5) & 0x3;
+ u32 A = (modeBits >> 5) & 0x3;
params.m_Width = A + 2;
params.m_Height = 12;
break;
@@ -586,15 +569,15 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
}
case 9: {
- uint32_t A = (modeBits >> 5) & 0x3;
- uint32_t B = (modeBits >> 9) & 0x3;
+ u32 A = (modeBits >> 5) & 0x3;
+ u32 B = (modeBits >> 9) & 0x3;
params.m_Width = A + 6;
params.m_Height = B + 6;
break;
}
default:
- assert(!"Don't know this layout...");
+ assert(false && "Don't know this layout...");
params.m_bError = true;
break;
}
@@ -605,10 +588,10 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
bool H = (layout != 9) && (modeBits & 0x200);
if (H) {
- const uint32_t maxWeights[6] = {9, 11, 15, 19, 23, 31};
+ const u32 maxWeights[6] = {9, 11, 15, 19, 23, 31};
params.m_MaxWeight = maxWeights[R - 2];
} else {
- const uint32_t maxWeights[6] = {1, 2, 3, 4, 5, 7};
+ const u32 maxWeights[6] = {1, 2, 3, 4, 5, 7};
params.m_MaxWeight = maxWeights[R - 2];
}
@@ -617,32 +600,32 @@ static TexelWeightParams DecodeBlockInfo(InputBitStream& strm) {
return params;
}
-static void FillVoidExtentLDR(InputBitStream& strm, uint32_t* const outBuf, uint32_t blockWidth,
- uint32_t blockHeight) {
+static void FillVoidExtentLDR(InputBitStream& strm, u32* const outBuf, u32 blockWidth,
+ u32 blockHeight) {
// Don't actually care about the void extent, just read the bits...
- for (int i = 0; i < 4; ++i) {
- strm.ReadBits(13);
+ for (s32 i = 0; i < 4; ++i) {
+ strm.ReadBits<13>();
}
// Decode the RGBA components and renormalize them to the range [0, 255]
- uint16_t r = static_cast<uint16_t>(strm.ReadBits(16));
- uint16_t g = static_cast<uint16_t>(strm.ReadBits(16));
- uint16_t b = static_cast<uint16_t>(strm.ReadBits(16));
- uint16_t a = static_cast<uint16_t>(strm.ReadBits(16));
+ u16 r = static_cast<u16>(strm.ReadBits<16>());
+ u16 g = static_cast<u16>(strm.ReadBits<16>());
+ u16 b = static_cast<u16>(strm.ReadBits<16>());
+ u16 a = static_cast<u16>(strm.ReadBits<16>());
- uint32_t rgba = (r >> 8) | (g & 0xFF00) | (static_cast<uint32_t>(b) & 0xFF00) << 8 |
- (static_cast<uint32_t>(a) & 0xFF00) << 16;
+ u32 rgba = (r >> 8) | (g & 0xFF00) | (static_cast<u32>(b) & 0xFF00) << 8 |
+ (static_cast<u32>(a) & 0xFF00) << 16;
- for (uint32_t j = 0; j < blockHeight; j++) {
- for (uint32_t i = 0; i < blockWidth; i++) {
+ for (u32 j = 0; j < blockHeight; j++) {
+ for (u32 i = 0; i < blockWidth; i++) {
outBuf[j * blockWidth + i] = rgba;
}
}
}
-static void FillError(uint32_t* outBuf, uint32_t blockWidth, uint32_t blockHeight) {
- for (uint32_t j = 0; j < blockHeight; j++) {
- for (uint32_t i = 0; i < blockWidth; i++) {
+static void FillError(u32* outBuf, u32 blockWidth, u32 blockHeight) {
+ for (u32 j = 0; j < blockHeight; j++) {
+ for (u32 i = 0; i < blockWidth; i++) {
outBuf[j * blockWidth + i] = 0xFFFF00FF;
}
}
@@ -651,18 +634,18 @@ static void FillError(uint32_t* outBuf, uint32_t blockWidth, uint32_t blockHeigh
// Replicates low numBits such that [(toBit - 1):(toBit - 1 - fromBit)]
// is the same as [(numBits - 1):0] and repeats all the way down.
template <typename IntType>
-static IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
+static IntType Replicate(IntType val, u32 numBits, u32 toBit) {
if (numBits == 0)
return 0;
if (toBit == 0)
return 0;
IntType v = val & static_cast<IntType>((1 << numBits) - 1);
IntType res = v;
- uint32_t reslen = numBits;
+ u32 reslen = numBits;
while (reslen < toBit) {
- uint32_t comp = 0;
+ u32 comp = 0;
if (numBits > toBit - reslen) {
- uint32_t newshift = toBit - reslen;
+ u32 newshift = toBit - reslen;
comp = numBits - newshift;
numBits = newshift;
}
@@ -675,14 +658,14 @@ static IntType Replicate(const IntType& val, uint32_t numBits, uint32_t toBit) {
class Pixel {
protected:
- using ChannelType = int16_t;
- uint8_t m_BitDepth[4] = {8, 8, 8, 8};
- int16_t color[4] = {};
+ using ChannelType = s16;
+ u8 m_BitDepth[4] = {8, 8, 8, 8};
+ s16 color[4] = {};
public:
Pixel() = default;
- Pixel(uint32_t a, uint32_t r, uint32_t g, uint32_t b, unsigned bitDepth = 8)
- : m_BitDepth{uint8_t(bitDepth), uint8_t(bitDepth), uint8_t(bitDepth), uint8_t(bitDepth)},
+ Pixel(u32 a, u32 r, u32 g, u32 b, u32 bitDepth = 8)
+ : m_BitDepth{u8(bitDepth), u8(bitDepth), u8(bitDepth), u8(bitDepth)},
color{static_cast<ChannelType>(a), static_cast<ChannelType>(r),
static_cast<ChannelType>(g), static_cast<ChannelType>(b)} {}
@@ -691,22 +674,22 @@ public:
// significant bits when going from larger to smaller bit depth
// or by repeating the most significant bits when going from
// smaller to larger bit depths.
- void ChangeBitDepth(const uint8_t (&depth)[4]) {
- for (uint32_t i = 0; i < 4; i++) {
+ void ChangeBitDepth(const u8 (&depth)[4]) {
+ for (u32 i = 0; i < 4; i++) {
Component(i) = ChangeBitDepth(Component(i), m_BitDepth[i], depth[i]);
m_BitDepth[i] = depth[i];
}
}
template <typename IntType>
- static float ConvertChannelToFloat(IntType channel, uint8_t bitDepth) {
+ static float ConvertChannelToFloat(IntType channel, u8 bitDepth) {
float denominator = static_cast<float>((1 << bitDepth) - 1);
return static_cast<float>(channel) / denominator;
}
// Changes the bit depth of a single component. See the comment
// above for how we do this.
- static ChannelType ChangeBitDepth(Pixel::ChannelType val, uint8_t oldDepth, uint8_t newDepth) {
+ static ChannelType ChangeBitDepth(Pixel::ChannelType val, u8 oldDepth, u8 newDepth) {
assert(newDepth <= 8);
assert(oldDepth <= 8);
@@ -722,16 +705,15 @@ public:
if (newDepth == 0) {
return 0xFF;
} else {
- uint8_t bitsWasted = static_cast<uint8_t>(oldDepth - newDepth);
- uint16_t v = static_cast<uint16_t>(val);
- v = static_cast<uint16_t>((v + (1 << (bitsWasted - 1))) >> bitsWasted);
- v = ::std::min<uint16_t>(::std::max<uint16_t>(0, v),
- static_cast<uint16_t>((1 << newDepth) - 1));
- return static_cast<uint8_t>(v);
+ u8 bitsWasted = static_cast<u8>(oldDepth - newDepth);
+ u16 v = static_cast<u16>(val);
+ v = static_cast<u16>((v + (1 << (bitsWasted - 1))) >> bitsWasted);
+ v = ::std::min<u16>(::std::max<u16>(0, v), static_cast<u16>((1 << newDepth) - 1));
+ return static_cast<u8>(v);
}
}
- assert(!"We shouldn't get here.");
+ assert(false && "We shouldn't get here.");
return 0;
}
@@ -759,15 +741,15 @@ public:
ChannelType& B() {
return color[3];
}
- const ChannelType& Component(uint32_t idx) const {
+ const ChannelType& Component(u32 idx) const {
return color[idx];
}
- ChannelType& Component(uint32_t idx) {
+ ChannelType& Component(u32 idx) {
return color[idx];
}
- void GetBitDepth(uint8_t (&outDepth)[4]) const {
- for (int i = 0; i < 4; i++) {
+ void GetBitDepth(u8 (&outDepth)[4]) const {
+ for (s32 i = 0; i < 4; i++) {
outDepth[i] = m_BitDepth[i];
}
}
@@ -776,12 +758,12 @@ public:
// and then pack each channel into an R8G8B8A8 32-bit integer. We assume
// that the architecture is little-endian, so the alpha channel will end
// up in the most-significant byte.
- uint32_t Pack() const {
+ u32 Pack() const {
Pixel eightBit(*this);
- const uint8_t eightBitDepth[4] = {8, 8, 8, 8};
+ const u8 eightBitDepth[4] = {8, 8, 8, 8};
eightBit.ChangeBitDepth(eightBitDepth);
- uint32_t r = 0;
+ u32 r = 0;
r |= eightBit.A();
r <<= 8;
r |= eightBit.B();
@@ -794,7 +776,7 @@ public:
// Clamps the pixel to the range [0,255]
void ClampByte() {
- for (uint32_t i = 0; i < 4; i++) {
+ for (u32 i = 0; i < 4; i++) {
color[i] = (color[i] < 0) ? 0 : ((color[i] > 255) ? 255 : color[i]);
}
}
@@ -804,24 +786,24 @@ public:
}
};
-static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* modes,
- const uint32_t nPartitions, const uint32_t nBitsForColorData) {
+static void DecodeColorValues(u32* out, u8* data, const u32* modes, const u32 nPartitions,
+ const u32 nBitsForColorData) {
// First figure out how many color values we have
- uint32_t nValues = 0;
- for (uint32_t i = 0; i < nPartitions; i++) {
+ u32 nValues = 0;
+ for (u32 i = 0; i < nPartitions; i++) {
nValues += ((modes[i] >> 2) + 1) << 1;
}
// Then based on the number of values and the remaining number of bits,
// figure out the max value for each of them...
- uint32_t range = 256;
+ u32 range = 256;
while (--range > 0) {
- IntegerEncodedValue val = IntegerEncodedValue::CreateEncoding(range);
- uint32_t bitLength = val.GetBitLength(nValues);
+ IntegerEncodedValue val = EncodingsValues[range];
+ u32 bitLength = val.GetBitLength(nValues);
if (bitLength <= nBitsForColorData) {
// Find the smallest possible range that matches the given encoding
while (--range > 0) {
- IntegerEncodedValue newval = IntegerEncodedValue::CreateEncoding(range);
+ IntegerEncodedValue newval = EncodingsValues[range];
if (!newval.MatchesEncoding(val)) {
break;
}
@@ -835,12 +817,14 @@ static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* mode
// We now have enough to decode our integer sequence.
std::vector<IntegerEncodedValue> decodedColorValues;
+ decodedColorValues.reserve(32);
+
InputBitStream colorStream(data);
- IntegerEncodedValue::DecodeIntegerSequence(decodedColorValues, colorStream, range, nValues);
+ DecodeIntegerSequence(decodedColorValues, colorStream, range, nValues);
// Once we have the decoded values, we need to dequantize them to the 0-255 range
// This procedure is outlined in ASTC spec C.2.13
- uint32_t outIdx = 0;
+ u32 outIdx = 0;
for (auto itr = decodedColorValues.begin(); itr != decodedColorValues.end(); ++itr) {
// Have we already decoded all that we need?
if (outIdx >= nValues) {
@@ -848,25 +832,25 @@ static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* mode
}
const IntegerEncodedValue& val = *itr;
- uint32_t bitlen = val.BaseBitLength();
- uint32_t bitval = val.GetBitValue();
+ u32 bitlen = val.num_bits;
+ u32 bitval = val.bit_value;
assert(bitlen >= 1);
- uint32_t A = 0, B = 0, C = 0, D = 0;
+ u32 A = 0, B = 0, C = 0, D = 0;
// A is just the lsb replicated 9 times.
A = Replicate(bitval & 1, 1, 9);
- switch (val.GetEncoding()) {
+ switch (val.encoding) {
// Replicate bits
- case eIntegerEncoding_JustBits:
+ case IntegerEncoding::JustBits:
out[outIdx++] = Replicate(bitval, bitlen, 8);
break;
// Use algorithm in C.2.13
- case eIntegerEncoding_Trit: {
+ case IntegerEncoding::Trit: {
- D = val.GetTritValue();
+ D = val.trit_value;
switch (bitlen) {
case 1: {
@@ -876,48 +860,48 @@ static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* mode
case 2: {
C = 93;
// B = b000b0bb0
- uint32_t b = (bitval >> 1) & 1;
+ u32 b = (bitval >> 1) & 1;
B = (b << 8) | (b << 4) | (b << 2) | (b << 1);
} break;
case 3: {
C = 44;
// B = cb000cbcb
- uint32_t cb = (bitval >> 1) & 3;
+ u32 cb = (bitval >> 1) & 3;
B = (cb << 7) | (cb << 2) | cb;
} break;
case 4: {
C = 22;
// B = dcb000dcb
- uint32_t dcb = (bitval >> 1) & 7;
+ u32 dcb = (bitval >> 1) & 7;
B = (dcb << 6) | dcb;
} break;
case 5: {
C = 11;
// B = edcb000ed
- uint32_t edcb = (bitval >> 1) & 0xF;
+ u32 edcb = (bitval >> 1) & 0xF;
B = (edcb << 5) | (edcb >> 2);
} break;
case 6: {
C = 5;
// B = fedcb000f
- uint32_t fedcb = (bitval >> 1) & 0x1F;
+ u32 fedcb = (bitval >> 1) & 0x1F;
B = (fedcb << 4) | (fedcb >> 4);
} break;
default:
- assert(!"Unsupported trit encoding for color values!");
+ assert(false && "Unsupported trit encoding for color values!");
break;
} // switch(bitlen)
- } // case eIntegerEncoding_Trit
+ } // case IntegerEncoding::Trit
break;
- case eIntegerEncoding_Quint: {
+ case IntegerEncoding::Qus32: {
- D = val.GetQuintValue();
+ D = val.qus32_value;
switch (bitlen) {
case 1: {
@@ -927,41 +911,41 @@ static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* mode
case 2: {
C = 54;
// B = b0000bb00
- uint32_t b = (bitval >> 1) & 1;
+ u32 b = (bitval >> 1) & 1;
B = (b << 8) | (b << 3) | (b << 2);
} break;
case 3: {
C = 26;
// B = cb0000cbc
- uint32_t cb = (bitval >> 1) & 3;
+ u32 cb = (bitval >> 1) & 3;
B = (cb << 7) | (cb << 1) | (cb >> 1);
} break;
case 4: {
C = 13;
// B = dcb0000dc
- uint32_t dcb = (bitval >> 1) & 7;
+ u32 dcb = (bitval >> 1) & 7;
B = (dcb << 6) | (dcb >> 1);
} break;
case 5: {
C = 6;
// B = edcb0000e
- uint32_t edcb = (bitval >> 1) & 0xF;
+ u32 edcb = (bitval >> 1) & 0xF;
B = (edcb << 5) | (edcb >> 3);
} break;
default:
- assert(!"Unsupported quint encoding for color values!");
+ assert(false && "Unsupported quint encoding for color values!");
break;
} // switch(bitlen)
- } // case eIntegerEncoding_Quint
+ } // case IntegerEncoding::Qus32
break;
- } // switch(val.GetEncoding())
+ } // switch(val.encoding)
- if (val.GetEncoding() != eIntegerEncoding_JustBits) {
- uint32_t T = D * C + B;
+ if (val.encoding != IntegerEncoding::JustBits) {
+ u32 T = D * C + B;
T ^= A;
T = (A & 0x80) | (T >> 2);
out[outIdx++] = T;
@@ -969,31 +953,31 @@ static void DecodeColorValues(uint32_t* out, uint8_t* data, const uint32_t* mode
}
// Make sure that each of our values is in the proper range...
- for (uint32_t i = 0; i < nValues; i++) {
+ for (u32 i = 0; i < nValues; i++) {
assert(out[i] <= 255);
}
}
-static uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
- uint32_t bitval = val.GetBitValue();
- uint32_t bitlen = val.BaseBitLength();
+static u32 UnquantizeTexelWeight(const IntegerEncodedValue& val) {
+ u32 bitval = val.bit_value;
+ u32 bitlen = val.num_bits;
- uint32_t A = Replicate(bitval & 1, 1, 7);
- uint32_t B = 0, C = 0, D = 0;
+ u32 A = Replicate(bitval & 1, 1, 7);
+ u32 B = 0, C = 0, D = 0;
- uint32_t result = 0;
- switch (val.GetEncoding()) {
- case eIntegerEncoding_JustBits:
+ u32 result = 0;
+ switch (val.encoding) {
+ case IntegerEncoding::JustBits:
result = Replicate(bitval, bitlen, 6);
break;
- case eIntegerEncoding_Trit: {
- D = val.GetTritValue();
+ case IntegerEncoding::Trit: {
+ D = val.trit_value;
assert(D < 3);
switch (bitlen) {
case 0: {
- uint32_t results[3] = {0, 32, 63};
+ u32 results[3] = {0, 32, 63};
result = results[D];
} break;
@@ -1003,29 +987,29 @@ static uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
case 2: {
C = 23;
- uint32_t b = (bitval >> 1) & 1;
+ u32 b = (bitval >> 1) & 1;
B = (b << 6) | (b << 2) | b;
} break;
case 3: {
C = 11;
- uint32_t cb = (bitval >> 1) & 3;
+ u32 cb = (bitval >> 1) & 3;
B = (cb << 5) | cb;
} break;
default:
- assert(!"Invalid trit encoding for texel weight");
+ assert(false && "Invalid trit encoding for texel weight");
break;
}
} break;
- case eIntegerEncoding_Quint: {
- D = val.GetQuintValue();
+ case IntegerEncoding::Qus32: {
+ D = val.qus32_value;
assert(D < 5);
switch (bitlen) {
case 0: {
- uint32_t results[5] = {0, 16, 32, 47, 63};
+ u32 results[5] = {0, 16, 32, 47, 63};
result = results[D];
} break;
@@ -1035,18 +1019,18 @@ static uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
case 2: {
C = 13;
- uint32_t b = (bitval >> 1) & 1;
+ u32 b = (bitval >> 1) & 1;
B = (b << 6) | (b << 1);
} break;
default:
- assert(!"Invalid quint encoding for texel weight");
+ assert(false && "Invalid quint encoding for texel weight");
break;
}
} break;
}
- if (val.GetEncoding() != eIntegerEncoding_JustBits && bitlen > 0) {
+ if (val.encoding != IntegerEncoding::JustBits && bitlen > 0) {
// Decode the value...
result = D * C + B;
result ^= A;
@@ -1063,12 +1047,11 @@ static uint32_t UnquantizeTexelWeight(const IntegerEncodedValue& val) {
return result;
}
-static void UnquantizeTexelWeights(uint32_t out[2][144],
- const std::vector<IntegerEncodedValue>& weights,
- const TexelWeightParams& params, const uint32_t blockWidth,
- const uint32_t blockHeight) {
- uint32_t weightIdx = 0;
- uint32_t unquantized[2][144];
+static void UnquantizeTexelWeights(u32 out[2][144], const std::vector<IntegerEncodedValue>& weights,
+ const TexelWeightParams& params, const u32 blockWidth,
+ const u32 blockHeight) {
+ u32 weightIdx = 0;
+ u32 unquantized[2][144];
for (auto itr = weights.begin(); itr != weights.end(); ++itr) {
unquantized[0][weightIdx] = UnquantizeTexelWeight(*itr);
@@ -1086,34 +1069,34 @@ static void UnquantizeTexelWeights(uint32_t out[2][144],
}
// Do infill if necessary (Section C.2.18) ...
- uint32_t Ds = (1024 + (blockWidth / 2)) / (blockWidth - 1);
- uint32_t Dt = (1024 + (blockHeight / 2)) / (blockHeight - 1);
+ u32 Ds = (1024 + (blockWidth / 2)) / (blockWidth - 1);
+ u32 Dt = (1024 + (blockHeight / 2)) / (blockHeight - 1);
- const uint32_t kPlaneScale = params.m_bDualPlane ? 2U : 1U;
- for (uint32_t plane = 0; plane < kPlaneScale; plane++)
- for (uint32_t t = 0; t < blockHeight; t++)
- for (uint32_t s = 0; s < blockWidth; s++) {
- uint32_t cs = Ds * s;
- uint32_t ct = Dt * t;
+ const u32 kPlaneScale = params.m_bDualPlane ? 2U : 1U;
+ for (u32 plane = 0; plane < kPlaneScale; plane++)
+ for (u32 t = 0; t < blockHeight; t++)
+ for (u32 s = 0; s < blockWidth; s++) {
+ u32 cs = Ds * s;
+ u32 ct = Dt * t;
- uint32_t gs = (cs * (params.m_Width - 1) + 32) >> 6;
- uint32_t gt = (ct * (params.m_Height - 1) + 32) >> 6;
+ u32 gs = (cs * (params.m_Width - 1) + 32) >> 6;
+ u32 gt = (ct * (params.m_Height - 1) + 32) >> 6;
- uint32_t js = gs >> 4;
- uint32_t fs = gs & 0xF;
+ u32 js = gs >> 4;
+ u32 fs = gs & 0xF;
- uint32_t jt = gt >> 4;
- uint32_t ft = gt & 0x0F;
+ u32 jt = gt >> 4;
+ u32 ft = gt & 0x0F;
- uint32_t w11 = (fs * ft + 8) >> 4;
- uint32_t w10 = ft - w11;
- uint32_t w01 = fs - w11;
- uint32_t w00 = 16 - fs - ft + w11;
+ u32 w11 = (fs * ft + 8) >> 4;
+ u32 w10 = ft - w11;
+ u32 w01 = fs - w11;
+ u32 w00 = 16 - fs - ft + w11;
- uint32_t v0 = js + jt * params.m_Width;
+ u32 v0 = js + jt * params.m_Width;
#define FIND_TEXEL(tidx, bidx) \
- uint32_t p##bidx = 0; \
+ u32 p##bidx = 0; \
do { \
if ((tidx) < (params.m_Width * params.m_Height)) { \
p##bidx = unquantized[plane][(tidx)]; \
@@ -1133,7 +1116,7 @@ static void UnquantizeTexelWeights(uint32_t out[2][144],
}
// Transfers a bit as described in C.2.14
-static inline void BitTransferSigned(int32_t& a, int32_t& b) {
+static inline void BitTransferSigned(s32& a, s32& b) {
b >>= 1;
b |= a & 0x80;
a >>= 1;
@@ -1144,14 +1127,14 @@ static inline void BitTransferSigned(int32_t& a, int32_t& b) {
// Adds more precision to the blue channel as described
// in C.2.14
-static inline Pixel BlueContract(int32_t a, int32_t r, int32_t g, int32_t b) {
- return Pixel(static_cast<int16_t>(a), static_cast<int16_t>((r + b) >> 1),
- static_cast<int16_t>((g + b) >> 1), static_cast<int16_t>(b));
+static inline Pixel BlueContract(s32 a, s32 r, s32 g, s32 b) {
+ return Pixel(static_cast<s16>(a), static_cast<s16>((r + b) >> 1),
+ static_cast<s16>((g + b) >> 1), static_cast<s16>(b));
}
// Partition selection functions as specified in
// C.2.21
-static inline uint32_t hash52(uint32_t p) {
+static inline u32 hash52(u32 p) {
p ^= p >> 15;
p -= p << 17;
p += p << 7;
@@ -1165,8 +1148,7 @@ static inline uint32_t hash52(uint32_t p) {
return p;
}
-static uint32_t SelectPartition(int32_t seed, int32_t x, int32_t y, int32_t z,
- int32_t partitionCount, int32_t smallBlock) {
+static u32 SelectPartition(s32 seed, s32 x, s32 y, s32 z, s32 partitionCount, s32 smallBlock) {
if (1 == partitionCount)
return 0;
@@ -1178,34 +1160,34 @@ static uint32_t SelectPartition(int32_t seed, int32_t x, int32_t y, int32_t z,
seed += (partitionCount - 1) * 1024;
- uint32_t rnum = hash52(static_cast<uint32_t>(seed));
- uint8_t seed1 = static_cast<uint8_t>(rnum & 0xF);
- uint8_t seed2 = static_cast<uint8_t>((rnum >> 4) & 0xF);
- uint8_t seed3 = static_cast<uint8_t>((rnum >> 8) & 0xF);
- uint8_t seed4 = static_cast<uint8_t>((rnum >> 12) & 0xF);
- uint8_t seed5 = static_cast<uint8_t>((rnum >> 16) & 0xF);
- uint8_t seed6 = static_cast<uint8_t>((rnum >> 20) & 0xF);
- uint8_t seed7 = static_cast<uint8_t>((rnum >> 24) & 0xF);
- uint8_t seed8 = static_cast<uint8_t>((rnum >> 28) & 0xF);
- uint8_t seed9 = static_cast<uint8_t>((rnum >> 18) & 0xF);
- uint8_t seed10 = static_cast<uint8_t>((rnum >> 22) & 0xF);
- uint8_t seed11 = static_cast<uint8_t>((rnum >> 26) & 0xF);
- uint8_t seed12 = static_cast<uint8_t>(((rnum >> 30) | (rnum << 2)) & 0xF);
-
- seed1 = static_cast<uint8_t>(seed1 * seed1);
- seed2 = static_cast<uint8_t>(seed2 * seed2);
- seed3 = static_cast<uint8_t>(seed3 * seed3);
- seed4 = static_cast<uint8_t>(seed4 * seed4);
- seed5 = static_cast<uint8_t>(seed5 * seed5);
- seed6 = static_cast<uint8_t>(seed6 * seed6);
- seed7 = static_cast<uint8_t>(seed7 * seed7);
- seed8 = static_cast<uint8_t>(seed8 * seed8);
- seed9 = static_cast<uint8_t>(seed9 * seed9);
- seed10 = static_cast<uint8_t>(seed10 * seed10);
- seed11 = static_cast<uint8_t>(seed11 * seed11);
- seed12 = static_cast<uint8_t>(seed12 * seed12);
-
- int32_t sh1, sh2, sh3;
+ u32 rnum = hash52(static_cast<u32>(seed));
+ u8 seed1 = static_cast<u8>(rnum & 0xF);
+ u8 seed2 = static_cast<u8>((rnum >> 4) & 0xF);
+ u8 seed3 = static_cast<u8>((rnum >> 8) & 0xF);
+ u8 seed4 = static_cast<u8>((rnum >> 12) & 0xF);
+ u8 seed5 = static_cast<u8>((rnum >> 16) & 0xF);
+ u8 seed6 = static_cast<u8>((rnum >> 20) & 0xF);
+ u8 seed7 = static_cast<u8>((rnum >> 24) & 0xF);
+ u8 seed8 = static_cast<u8>((rnum >> 28) & 0xF);
+ u8 seed9 = static_cast<u8>((rnum >> 18) & 0xF);
+ u8 seed10 = static_cast<u8>((rnum >> 22) & 0xF);
+ u8 seed11 = static_cast<u8>((rnum >> 26) & 0xF);
+ u8 seed12 = static_cast<u8>(((rnum >> 30) | (rnum << 2)) & 0xF);
+
+ seed1 = static_cast<u8>(seed1 * seed1);
+ seed2 = static_cast<u8>(seed2 * seed2);
+ seed3 = static_cast<u8>(seed3 * seed3);
+ seed4 = static_cast<u8>(seed4 * seed4);
+ seed5 = static_cast<u8>(seed5 * seed5);
+ seed6 = static_cast<u8>(seed6 * seed6);
+ seed7 = static_cast<u8>(seed7 * seed7);
+ seed8 = static_cast<u8>(seed8 * seed8);
+ seed9 = static_cast<u8>(seed9 * seed9);
+ seed10 = static_cast<u8>(seed10 * seed10);
+ seed11 = static_cast<u8>(seed11 * seed11);
+ seed12 = static_cast<u8>(seed12 * seed12);
+
+ s32 sh1, sh2, sh3;
if (seed & 1) {
sh1 = (seed & 2) ? 4 : 5;
sh2 = (partitionCount == 3) ? 6 : 5;
@@ -1215,23 +1197,23 @@ static uint32_t SelectPartition(int32_t seed, int32_t x, int32_t y, int32_t z,
}
sh3 = (seed & 0x10) ? sh1 : sh2;
- seed1 = static_cast<uint8_t>(seed1 >> sh1);
- seed2 = static_cast<uint8_t>(seed2 >> sh2);
- seed3 = static_cast<uint8_t>(seed3 >> sh1);
- seed4 = static_cast<uint8_t>(seed4 >> sh2);
- seed5 = static_cast<uint8_t>(seed5 >> sh1);
- seed6 = static_cast<uint8_t>(seed6 >> sh2);
- seed7 = static_cast<uint8_t>(seed7 >> sh1);
- seed8 = static_cast<uint8_t>(seed8 >> sh2);
- seed9 = static_cast<uint8_t>(seed9 >> sh3);
- seed10 = static_cast<uint8_t>(seed10 >> sh3);
- seed11 = static_cast<uint8_t>(seed11 >> sh3);
- seed12 = static_cast<uint8_t>(seed12 >> sh3);
-
- int32_t a = seed1 * x + seed2 * y + seed11 * z + (rnum >> 14);
- int32_t b = seed3 * x + seed4 * y + seed12 * z + (rnum >> 10);
- int32_t c = seed5 * x + seed6 * y + seed9 * z + (rnum >> 6);
- int32_t d = seed7 * x + seed8 * y + seed10 * z + (rnum >> 2);
+ seed1 = static_cast<u8>(seed1 >> sh1);
+ seed2 = static_cast<u8>(seed2 >> sh2);
+ seed3 = static_cast<u8>(seed3 >> sh1);
+ seed4 = static_cast<u8>(seed4 >> sh2);
+ seed5 = static_cast<u8>(seed5 >> sh1);
+ seed6 = static_cast<u8>(seed6 >> sh2);
+ seed7 = static_cast<u8>(seed7 >> sh1);
+ seed8 = static_cast<u8>(seed8 >> sh2);
+ seed9 = static_cast<u8>(seed9 >> sh3);
+ seed10 = static_cast<u8>(seed10 >> sh3);
+ seed11 = static_cast<u8>(seed11 >> sh3);
+ seed12 = static_cast<u8>(seed12 >> sh3);
+
+ s32 a = seed1 * x + seed2 * y + seed11 * z + (rnum >> 14);
+ s32 b = seed3 * x + seed4 * y + seed12 * z + (rnum >> 10);
+ s32 c = seed5 * x + seed6 * y + seed9 * z + (rnum >> 6);
+ s32 d = seed7 * x + seed8 * y + seed10 * z + (rnum >> 2);
a &= 0x3F;
b &= 0x3F;
@@ -1252,27 +1234,26 @@ static uint32_t SelectPartition(int32_t seed, int32_t x, int32_t y, int32_t z,
return 3;
}
-static inline uint32_t Select2DPartition(int32_t seed, int32_t x, int32_t y, int32_t partitionCount,
- int32_t smallBlock) {
+static inline u32 Select2DPartition(s32 seed, s32 x, s32 y, s32 partitionCount, s32 smallBlock) {
return SelectPartition(seed, x, y, 0, partitionCount, smallBlock);
}
// Section C.2.14
-static void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValues,
- uint32_t colorEndpointMode) {
+static void ComputeEndpos32s(Pixel& ep1, Pixel& ep2, const u32*& colorValues,
+ u32 colorEndpos32Mode) {
#define READ_UINT_VALUES(N) \
- uint32_t v[N]; \
- for (uint32_t i = 0; i < N; i++) { \
+ u32 v[N]; \
+ for (u32 i = 0; i < N; i++) { \
v[i] = *(colorValues++); \
}
#define READ_INT_VALUES(N) \
- int32_t v[N]; \
- for (uint32_t i = 0; i < N; i++) { \
- v[i] = static_cast<int32_t>(*(colorValues++)); \
+ s32 v[N]; \
+ for (u32 i = 0; i < N; i++) { \
+ v[i] = static_cast<s32>(*(colorValues++)); \
}
- switch (colorEndpointMode) {
+ switch (colorEndpos32Mode) {
case 0: {
READ_UINT_VALUES(2)
ep1 = Pixel(0xFF, v[0], v[0], v[0]);
@@ -1281,8 +1262,8 @@ static void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValue
case 1: {
READ_UINT_VALUES(2)
- uint32_t L0 = (v[0] >> 2) | (v[1] & 0xC0);
- uint32_t L1 = std::max(L0 + (v[1] & 0x3F), 0xFFU);
+ u32 L0 = (v[0] >> 2) | (v[1] & 0xC0);
+ u32 L1 = std::max(L0 + (v[1] & 0x3F), 0xFFU);
ep1 = Pixel(0xFF, L0, L0, L0);
ep2 = Pixel(0xFF, L1, L1, L1);
} break;
@@ -1371,7 +1352,7 @@ static void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValue
} break;
default:
- assert(!"Unsupported color endpoint mode (is it HDR?)");
+ assert(false && "Unsupported color endpoint mode (is it HDR?)");
break;
}
@@ -1379,14 +1360,14 @@ static void ComputeEndpoints(Pixel& ep1, Pixel& ep2, const uint32_t*& colorValue
#undef READ_INT_VALUES
}
-static void DecompressBlock(const uint8_t inBuf[16], const uint32_t blockWidth,
- const uint32_t blockHeight, uint32_t* outBuf) {
+static void DecompressBlock(const u8 inBuf[16], const u32 blockWidth, const u32 blockHeight,
+ u32* outBuf) {
InputBitStream strm(inBuf);
TexelWeightParams weightParams = DecodeBlockInfo(strm);
// Was there an error?
if (weightParams.m_bError) {
- assert(!"Invalid block mode");
+ assert(false && "Invalid block mode");
FillError(outBuf, blockWidth, blockHeight);
return;
}
@@ -1397,63 +1378,63 @@ static void DecompressBlock(const uint8_t inBuf[16], const uint32_t blockWidth,
}
if (weightParams.m_bVoidExtentHDR) {
- assert(!"HDR void extent blocks are unsupported!");
+ assert(false && "HDR void extent blocks are unsupported!");
FillError(outBuf, blockWidth, blockHeight);
return;
}
if (weightParams.m_Width > blockWidth) {
- assert(!"Texel weight grid width should be smaller than block width");
+ assert(false && "Texel weight grid width should be smaller than block width");
FillError(outBuf, blockWidth, blockHeight);
return;
}
if (weightParams.m_Height > blockHeight) {
- assert(!"Texel weight grid height should be smaller than block height");
+ assert(false && "Texel weight grid height should be smaller than block height");
FillError(outBuf, blockWidth, blockHeight);
return;
}
// Read num partitions
- uint32_t nPartitions = strm.ReadBits(2) + 1;
+ u32 nPartitions = strm.ReadBits<2>() + 1;
assert(nPartitions <= 4);
if (nPartitions == 4 && weightParams.m_bDualPlane) {
- assert(!"Dual plane mode is incompatible with four partition blocks");
+ assert(false && "Dual plane mode is incompatible with four partition blocks");
FillError(outBuf, blockWidth, blockHeight);
return;
}
- // Based on the number of partitions, read the color endpoint mode for
+ // Based on the number of partitions, read the color endpos32 mode for
// each partition.
- // Determine partitions, partition index, and color endpoint modes
- int32_t planeIdx = -1;
- uint32_t partitionIndex;
- uint32_t colorEndpointMode[4] = {0, 0, 0, 0};
+ // Determine partitions, partition index, and color endpos32 modes
+ s32 planeIdx = -1;
+ u32 partitionIndex;
+ u32 colorEndpos32Mode[4] = {0, 0, 0, 0};
// Define color data.
- uint8_t colorEndpointData[16];
- memset(colorEndpointData, 0, sizeof(colorEndpointData));
- OutputBitStream colorEndpointStream(colorEndpointData, 16 * 8, 0);
+ u8 colorEndpos32Data[16];
+ memset(colorEndpos32Data, 0, sizeof(colorEndpos32Data));
+ OutputBitStream colorEndpos32Stream(colorEndpos32Data, 16 * 8, 0);
// Read extra config data...
- uint32_t baseCEM = 0;
+ u32 baseCEM = 0;
if (nPartitions == 1) {
- colorEndpointMode[0] = strm.ReadBits(4);
+ colorEndpos32Mode[0] = strm.ReadBits<4>();
partitionIndex = 0;
} else {
- partitionIndex = strm.ReadBits(10);
- baseCEM = strm.ReadBits(6);
+ partitionIndex = strm.ReadBits<10>();
+ baseCEM = strm.ReadBits<6>();
}
- uint32_t baseMode = (baseCEM & 3);
+ u32 baseMode = (baseCEM & 3);
- // Remaining bits are color endpoint data...
- uint32_t nWeightBits = weightParams.GetPackedBitSize();
- int32_t remainingBits = 128 - nWeightBits - strm.GetBitsRead();
+ // Remaining bits are color endpos32 data...
+ u32 nWeightBits = weightParams.GetPackedBitSize();
+ s32 remainingBits = 128 - nWeightBits - static_cast<s32>(strm.GetBitsRead());
// Consider extra bits prior to texel data...
- uint32_t extraCEMbits = 0;
+ u32 extraCEMbits = 0;
if (baseMode) {
switch (nPartitions) {
case 2:
@@ -1473,18 +1454,18 @@ static void DecompressBlock(const uint8_t inBuf[16], const uint32_t blockWidth,
remainingBits -= extraCEMbits;
// Do we have a dual plane situation?
- uint32_t planeSelectorBits = 0;
+ u32 planeSelectorBits = 0;
if (weightParams.m_bDualPlane) {
planeSelectorBits = 2;
}
remainingBits -= planeSelectorBits;
// Read color data...
- uint32_t colorDataBits = remainingBits;
+ u32 colorDataBits = remainingBits;
while (remainingBits > 0) {
- uint32_t nb = std::min(remainingBits, 8);
- uint32_t b = strm.ReadBits(nb);
- colorEndpointStream.WriteBits(b, nb);
+ u32 nb = std::min(remainingBits, 8);
+ u32 b = strm.ReadBits(nb);
+ colorEndpos32Stream.WriteBits(b, nb);
remainingBits -= 8;
}
@@ -1493,64 +1474,64 @@ static void DecompressBlock(const uint8_t inBuf[16], const uint32_t blockWidth,
// Read the rest of the CEM
if (baseMode) {
- uint32_t extraCEM = strm.ReadBits(extraCEMbits);
- uint32_t CEM = (extraCEM << 6) | baseCEM;
+ u32 extraCEM = strm.ReadBits(extraCEMbits);
+ u32 CEM = (extraCEM << 6) | baseCEM;
CEM >>= 2;
bool C[4] = {0};
- for (uint32_t i = 0; i < nPartitions; i++) {
+ for (u32 i = 0; i < nPartitions; i++) {
C[i] = CEM & 1;
CEM >>= 1;
}
- uint8_t M[4] = {0};
- for (uint32_t i = 0; i < nPartitions; i++) {
+ u8 M[4] = {0};
+ for (u32 i = 0; i < nPartitions; i++) {
M[i] = CEM & 3;
CEM >>= 2;
assert(M[i] <= 3);
}
- for (uint32_t i = 0; i < nPartitions; i++) {
- colorEndpointMode[i] = baseMode;
+ for (u32 i = 0; i < nPartitions; i++) {
+ colorEndpos32Mode[i] = baseMode;
if (!(C[i]))
- colorEndpointMode[i] -= 1;
- colorEndpointMode[i] <<= 2;
- colorEndpointMode[i] |= M[i];
+ colorEndpos32Mode[i] -= 1;
+ colorEndpos32Mode[i] <<= 2;
+ colorEndpos32Mode[i] |= M[i];
}
} else if (nPartitions > 1) {
- uint32_t CEM = baseCEM >> 2;
- for (uint32_t i = 0; i < nPartitions; i++) {
- colorEndpointMode[i] = CEM;
+ u32 CEM = baseCEM >> 2;
+ for (u32 i = 0; i < nPartitions; i++) {
+ colorEndpos32Mode[i] = CEM;
}
}
// Make sure everything up till here is sane.
- for (uint32_t i = 0; i < nPartitions; i++) {
- assert(colorEndpointMode[i] < 16);
+ for (u32 i = 0; i < nPartitions; i++) {
+ assert(colorEndpos32Mode[i] < 16);
}
assert(strm.GetBitsRead() + weightParams.GetPackedBitSize() == 128);
// Decode both color data and texel weight data
- uint32_t colorValues[32]; // Four values, two endpoints, four maximum paritions
- DecodeColorValues(colorValues, colorEndpointData, colorEndpointMode, nPartitions,
+ u32 colorValues[32]; // Four values, two endpos32s, four maximum paritions
+ DecodeColorValues(colorValues, colorEndpos32Data, colorEndpos32Mode, nPartitions,
colorDataBits);
- Pixel endpoints[4][2];
- const uint32_t* colorValuesPtr = colorValues;
- for (uint32_t i = 0; i < nPartitions; i++) {
- ComputeEndpoints(endpoints[i][0], endpoints[i][1], colorValuesPtr, colorEndpointMode[i]);
+ Pixel endpos32s[4][2];
+ const u32* colorValuesPtr = colorValues;
+ for (u32 i = 0; i < nPartitions; i++) {
+ ComputeEndpos32s(endpos32s[i][0], endpos32s[i][1], colorValuesPtr, colorEndpos32Mode[i]);
}
// Read the texel weight data..
- uint8_t texelWeightData[16];
+ u8 texelWeightData[16];
memcpy(texelWeightData, inBuf, sizeof(texelWeightData));
// Reverse everything
- for (uint32_t i = 0; i < 8; i++) {
+ for (u32 i = 0; i < 8; i++) {
// Taken from http://graphics.stanford.edu/~seander/bithacks.html#ReverseByteWith64Bits
#define REVERSE_BYTE(b) (((b)*0x80200802ULL) & 0x0884422110ULL) * 0x0101010101ULL >> 32
- unsigned char a = static_cast<unsigned char>(REVERSE_BYTE(texelWeightData[i]));
- unsigned char b = static_cast<unsigned char>(REVERSE_BYTE(texelWeightData[15 - i]));
+ u8 a = static_cast<u8>(REVERSE_BYTE(texelWeightData[i]));
+ u8 b = static_cast<u8>(REVERSE_BYTE(texelWeightData[15 - i]));
#undef REVERSE_BYTE
texelWeightData[i] = b;
@@ -1558,50 +1539,51 @@ static void DecompressBlock(const uint8_t inBuf[16], const uint32_t blockWidth,
}
// Make sure that higher non-texel bits are set to zero
- const uint32_t clearByteStart = (weightParams.GetPackedBitSize() >> 3) + 1;
+ const u32 clearByteStart = (weightParams.GetPackedBitSize() >> 3) + 1;
texelWeightData[clearByteStart - 1] =
texelWeightData[clearByteStart - 1] &
- static_cast<uint8_t>((1 << (weightParams.GetPackedBitSize() % 8)) - 1);
+ static_cast<u8>((1 << (weightParams.GetPackedBitSize() % 8)) - 1);
memset(texelWeightData + clearByteStart, 0, 16 - clearByteStart);
std::vector<IntegerEncodedValue> texelWeightValues;
+ texelWeightValues.reserve(64);
+
InputBitStream weightStream(texelWeightData);
- IntegerEncodedValue::DecodeIntegerSequence(texelWeightValues, weightStream,
- weightParams.m_MaxWeight,
- weightParams.GetNumWeightValues());
+ DecodeIntegerSequence(texelWeightValues, weightStream, weightParams.m_MaxWeight,
+ weightParams.GetNumWeightValues());
// Blocks can be at most 12x12, so we can have as many as 144 weights
- uint32_t weights[2][144];
+ u32 weights[2][144];
UnquantizeTexelWeights(weights, texelWeightValues, weightParams, blockWidth, blockHeight);
- // Now that we have endpoints and weights, we can interpolate and generate
+ // Now that we have endpos32s and weights, we can s32erpolate and generate
// the proper decoding...
- for (uint32_t j = 0; j < blockHeight; j++)
- for (uint32_t i = 0; i < blockWidth; i++) {
- uint32_t partition = Select2DPartition(partitionIndex, i, j, nPartitions,
- (blockHeight * blockWidth) < 32);
+ for (u32 j = 0; j < blockHeight; j++)
+ for (u32 i = 0; i < blockWidth; i++) {
+ u32 partition = Select2DPartition(partitionIndex, i, j, nPartitions,
+ (blockHeight * blockWidth) < 32);
assert(partition < nPartitions);
Pixel p;
- for (uint32_t c = 0; c < 4; c++) {
- uint32_t C0 = endpoints[partition][0].Component(c);
+ for (u32 c = 0; c < 4; c++) {
+ u32 C0 = endpos32s[partition][0].Component(c);
C0 = Replicate(C0, 8, 16);
- uint32_t C1 = endpoints[partition][1].Component(c);
+ u32 C1 = endpos32s[partition][1].Component(c);
C1 = Replicate(C1, 8, 16);
- uint32_t plane = 0;
+ u32 plane = 0;
if (weightParams.m_bDualPlane && (((planeIdx + 1) & 3) == c)) {
plane = 1;
}
- uint32_t weight = weights[plane][j * blockWidth + i];
- uint32_t C = (C0 * (64 - weight) + C1 * weight + 32) / 64;
+ u32 weight = weights[plane][j * blockWidth + i];
+ u32 C = (C0 * (64 - weight) + C1 * weight + 32) / 64;
if (C == 65535) {
p.Component(c) = 255;
} else {
double Cf = static_cast<double>(C);
- p.Component(c) = static_cast<uint16_t>(255.0 * (Cf / 65536.0) + 0.5);
+ p.Component(c) = static_cast<u16>(255.0 * (Cf / 65536.0) + 0.5);
}
}
@@ -1613,26 +1595,26 @@ static void DecompressBlock(const uint8_t inBuf[16], const uint32_t blockWidth,
namespace Tegra::Texture::ASTC {
-std::vector<uint8_t> Decompress(const uint8_t* data, uint32_t width, uint32_t height,
- uint32_t depth, uint32_t block_width, uint32_t block_height) {
- uint32_t blockIdx = 0;
+std::vector<u8> Decompress(const u8* data, u32 width, u32 height, u32 depth, u32 block_width,
+ u32 block_height) {
+ u32 blockIdx = 0;
std::size_t depth_offset = 0;
- std::vector<uint8_t> outData(height * width * depth * 4);
- for (uint32_t k = 0; k < depth; k++) {
- for (uint32_t j = 0; j < height; j += block_height) {
- for (uint32_t i = 0; i < width; i += block_width) {
+ std::vector<u8> outData(height * width * depth * 4);
+ for (u32 k = 0; k < depth; k++) {
+ for (u32 j = 0; j < height; j += block_height) {
+ for (u32 i = 0; i < width; i += block_width) {
- const uint8_t* blockPtr = data + blockIdx * 16;
+ const u8* blockPtr = data + blockIdx * 16;
// Blocks can be at most 12x12
- uint32_t uncompData[144];
+ u32 uncompData[144];
ASTCC::DecompressBlock(blockPtr, block_width, block_height, uncompData);
- uint32_t decompWidth = std::min(block_width, width - i);
- uint32_t decompHeight = std::min(block_height, height - j);
+ u32 decompWidth = std::min(block_width, width - i);
+ u32 decompHeight = std::min(block_height, height - j);
- uint8_t* outRow = depth_offset + outData.data() + (j * width + i) * 4;
- for (uint32_t jj = 0; jj < decompHeight; jj++) {
+ u8* outRow = depth_offset + outData.data() + (j * width + i) * 4;
+ for (u32 jj = 0; jj < decompHeight; jj++) {
memcpy(outRow + jj * width * 4, uncompData + jj * block_width, decompWidth * 4);
}
diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp
index c38860628..3b9ab38dd 100644
--- a/src/yuzu/configuration/config.cpp
+++ b/src/yuzu/configuration/config.cpp
@@ -539,7 +539,7 @@ void Config::ReadDebuggingValues() {
void Config::ReadServiceValues() {
qt_config->beginGroup(QStringLiteral("Services"));
Settings::values.bcat_backend =
- ReadSetting(QStringLiteral("bcat_backend"), QStringLiteral("boxcat"))
+ ReadSetting(QStringLiteral("bcat_backend"), QStringLiteral("null"))
.toString()
.toStdString();
Settings::values.bcat_boxcat_local =
@@ -682,6 +682,8 @@ void Config::ReadSystemValues() {
Settings::values.language_index = ReadSetting(QStringLiteral("language_index"), 1).toInt();
+ Settings::values.region_index = ReadSetting(QStringLiteral("region_index"), 1).toInt();
+
const auto rng_seed_enabled = ReadSetting(QStringLiteral("rng_seed_enabled"), false).toBool();
if (rng_seed_enabled) {
Settings::values.rng_seed = ReadSetting(QStringLiteral("rng_seed"), 0).toULongLong();
@@ -698,6 +700,8 @@ void Config::ReadSystemValues() {
Settings::values.custom_rtc = std::nullopt;
}
+ Settings::values.sound_index = ReadSetting(QStringLiteral("sound_index"), 1).toInt();
+
qt_config->endGroup();
}
@@ -1114,6 +1118,7 @@ void Config::SaveSystemValues() {
WriteSetting(QStringLiteral("use_docked_mode"), Settings::values.use_docked_mode, false);
WriteSetting(QStringLiteral("current_user"), Settings::values.current_user, 0);
WriteSetting(QStringLiteral("language_index"), Settings::values.language_index, 1);
+ WriteSetting(QStringLiteral("region_index"), Settings::values.region_index, 1);
WriteSetting(QStringLiteral("rng_seed_enabled"), Settings::values.rng_seed.has_value(), false);
WriteSetting(QStringLiteral("rng_seed"), Settings::values.rng_seed.value_or(0), 0);
@@ -1125,6 +1130,8 @@ void Config::SaveSystemValues() {
Settings::values.custom_rtc.value_or(std::chrono::seconds{}).count()),
0);
+ WriteSetting(QStringLiteral("sound_index"), Settings::values.sound_index, 1);
+
qt_config->endGroup();
}
diff --git a/src/yuzu/configuration/configure_system.cpp b/src/yuzu/configuration/configure_system.cpp
index e1b52f8d9..f49cd4c8f 100644
--- a/src/yuzu/configuration/configure_system.cpp
+++ b/src/yuzu/configuration/configure_system.cpp
@@ -56,6 +56,8 @@ void ConfigureSystem::SetConfiguration() {
enabled = !Core::System::GetInstance().IsPoweredOn();
ui->combo_language->setCurrentIndex(Settings::values.language_index);
+ ui->combo_region->setCurrentIndex(Settings::values.region_index);
+ ui->combo_sound->setCurrentIndex(Settings::values.sound_index);
ui->rng_seed_checkbox->setChecked(Settings::values.rng_seed.has_value());
ui->rng_seed_edit->setEnabled(Settings::values.rng_seed.has_value());
@@ -81,6 +83,8 @@ void ConfigureSystem::ApplyConfiguration() {
}
Settings::values.language_index = ui->combo_language->currentIndex();
+ Settings::values.region_index = ui->combo_region->currentIndex();
+ Settings::values.sound_index = ui->combo_sound->currentIndex();
if (ui->rng_seed_checkbox->isChecked()) {
Settings::values.rng_seed = ui->rng_seed_edit->text().toULongLong(nullptr, 16);
diff --git a/src/yuzu/configuration/configure_system.h b/src/yuzu/configuration/configure_system.h
index 1eab3781d..d8fa2d2cc 100644
--- a/src/yuzu/configuration/configure_system.h
+++ b/src/yuzu/configuration/configure_system.h
@@ -36,5 +36,6 @@ private:
bool enabled = false;
int language_index = 0;
+ int region_index = 0;
int sound_index = 0;
};
diff --git a/src/yuzu/configuration/configure_system.ui b/src/yuzu/configuration/configure_system.ui
index 65745a2f8..4e2c7e76e 100644
--- a/src/yuzu/configuration/configure_system.ui
+++ b/src/yuzu/configuration/configure_system.ui
@@ -22,14 +22,14 @@
<string>System Settings</string>
</property>
<layout class="QGridLayout" name="gridLayout">
- <item row="1" column="0">
+ <item row="2" column="0">
<widget class="QLabel" name="label_sound">
<property name="text">
<string>Sound output mode</string>
</property>
</widget>
</item>
- <item row="2" column="0">
+ <item row="3" column="0">
<widget class="QLabel" name="label_console_id">
<property name="text">
<string>Console ID:</string>
@@ -128,14 +128,60 @@
</item>
</widget>
</item>
- <item row="4" column="0">
+ <item row="1" column="0">
+ <widget class="QLabel" name="label_region">
+ <property name="text">
+ <string>Region:</string>
+ </property>
+ </widget>
+ </item>
+ <item row="1" column="1">
+ <widget class="QComboBox" name="combo_region">
+ <item>
+ <property name="text">
+ <string>Japan</string>
+ </property>
+ </item>
+ <item>
+ <property name="text">
+ <string>USA</string>
+ </property>
+ </item>
+ <item>
+ <property name="text">
+ <string>Europe</string>
+ </property>
+ </item>
+ <item>
+ <property name="text">
+ <string>Australia</string>
+ </property>
+ </item>
+ <item>
+ <property name="text">
+ <string>China</string>
+ </property>
+ </item>
+ <item>
+ <property name="text">
+ <string>Korea</string>
+ </property>
+ </item>
+ <item>
+ <property name="text">
+ <string>Taiwan</string>
+ </property>
+ </item>
+ </widget>
+ </item>
+ <item row="5" column="0">
<widget class="QCheckBox" name="rng_seed_checkbox">
<property name="text">
<string>RNG Seed</string>
</property>
</widget>
</item>
- <item row="1" column="1">
+ <item row="2" column="1">
<widget class="QComboBox" name="combo_sound">
<item>
<property name="text">
@@ -161,7 +207,7 @@
</property>
</widget>
</item>
- <item row="2" column="1">
+ <item row="3" column="1">
<widget class="QPushButton" name="button_regenerate_console_id">
<property name="sizePolicy">
<sizepolicy hsizetype="Fixed" vsizetype="Fixed">
@@ -177,14 +223,14 @@
</property>
</widget>
</item>
- <item row="3" column="0">
+ <item row="4" column="0">
<widget class="QCheckBox" name="custom_rtc_checkbox">
<property name="text">
<string>Custom RTC</string>
</property>
</widget>
</item>
- <item row="3" column="1">
+ <item row="4" column="1">
<widget class="QDateTimeEdit" name="custom_rtc_edit">
<property name="minimumDate">
<date>
@@ -198,7 +244,7 @@
</property>
</widget>
</item>
- <item row="4" column="1">
+ <item row="5" column="1">
<widget class="QLineEdit" name="rng_seed_edit">
<property name="sizePolicy">
<sizepolicy hsizetype="Minimum" vsizetype="Fixed">
diff --git a/src/yuzu_cmd/config.cpp b/src/yuzu_cmd/config.cpp
index 907abaa51..f4cd905c9 100644
--- a/src/yuzu_cmd/config.cpp
+++ b/src/yuzu_cmd/config.cpp
@@ -452,7 +452,7 @@ void Config::ReadValues() {
Settings::values.yuzu_token = sdl2_config->Get("WebService", "yuzu_token", "");
// Services
- Settings::values.bcat_backend = sdl2_config->Get("Services", "bcat_backend", "boxcat");
+ Settings::values.bcat_backend = sdl2_config->Get("Services", "bcat_backend", "null");
Settings::values.bcat_boxcat_local =
sdl2_config->GetBoolean("Services", "bcat_boxcat_local", false);
}