27 files changed, 543 insertions, 362 deletions

diff --git a/.gitmodules b/.gitmodules
index 059512902..db0905b3d 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -9,4 +9,4 @@
         url = https://github.com/neobrain/nihstro.git
 [submodule "soundtouch"]
         path = externals/soundtouch
-        url = https://github.com/citra-emu/soundtouch.git
+        url = https://github.com/citra-emu/ext-soundtouch.git
diff --git a/src/audio_core/hle/source.cpp b/src/audio_core/hle/source.cpp
index daaf6e3f3..30552fe26 100644
--- a/src/audio_core/hle/source.cpp
+++ b/src/audio_core/hle/source.cpp
@@ -126,13 +126,13 @@ void Source::ParseConfig(SourceConfiguration::Configuration& config, const s16_l
     if (config.simple_filter_dirty) {
         config.simple_filter_dirty.Assign(0);
         state.filters.Configure(config.simple_filter);
-        LOG_TRACE(Audio_DSP, "source_id=%zu simple filter update");
+        LOG_TRACE(Audio_DSP, "source_id=%zu simple filter update", source_id);
     }
 
     if (config.biquad_filter_dirty) {
         config.biquad_filter_dirty.Assign(0);
         state.filters.Configure(config.biquad_filter);
-        LOG_TRACE(Audio_DSP, "source_id=%zu biquad filter update");
+        LOG_TRACE(Audio_DSP, "source_id=%zu biquad filter update", source_id);
     }
 
     if (config.interpolation_dirty) {
diff --git a/src/citra_qt/debugger/graphics_tracing.cpp b/src/citra_qt/debugger/graphics_tracing.cpp
index 1402f8e79..9c80f7ec9 100644
--- a/src/citra_qt/debugger/graphics_tracing.cpp
+++ b/src/citra_qt/debugger/graphics_tracing.cpp
@@ -74,7 +74,7 @@ void GraphicsTracingWidget::StartRecording() {
     std::array<u32, 4 * 16> default_attributes;
     for (unsigned i = 0; i < 16; ++i) {
         for (unsigned comp = 0; comp < 3; ++comp) {
-            default_attributes[4 * i + comp] = nihstro::to_float24(Pica::g_state.vs.default_attributes[i][comp].ToFloat32());
+            default_attributes[4 * i + comp] = nihstro::to_float24(Pica::g_state.vs_default_attributes[i][comp].ToFloat32());
         }
     }
 
diff --git a/src/citra_qt/debugger/graphics_vertex_shader.cpp b/src/citra_qt/debugger/graphics_vertex_shader.cpp
index 854f6ff16..391666d35 100644
--- a/src/citra_qt/debugger/graphics_vertex_shader.cpp
+++ b/src/citra_qt/debugger/graphics_vertex_shader.cpp
@@ -501,7 +501,7 @@ void GraphicsVertexShaderWidget::Reload(bool replace_vertex_data, void* vertex_d
     info.labels.insert({ entry_point, "main" });
 
     // Generate debug information
-    debug_data = Pica::Shader::ProduceDebugInfo(input_vertex, num_attributes, shader_config, shader_setup);
+    debug_data = Pica::g_state.vs.ProduceDebugInfo(input_vertex, num_attributes, shader_config, shader_setup);
 
     // Reload widget state
     for (int attr = 0; attr < num_attributes; ++attr) {
diff --git a/src/common/swap.h b/src/common/swap.h
index a7c37bc44..1749bd7a4 100644
--- a/src/common/swap.h
+++ b/src/common/swap.h
@@ -25,6 +25,8 @@
     #include <sys/endian.h>
 #endif
 
+#include <cstring>
+
 #include "common/common_types.h"
 
 // GCC 4.6+
@@ -58,9 +60,6 @@
 
 namespace Common {
 
-inline u8 swap8(u8 _data) {return _data;}
-inline u32 swap24(const u8* _data) {return (_data[0] << 16) | (_data[1] << 8) | _data[2];}
-
 #ifdef _MSC_VER
 inline u16 swap16(u16 _data) {return _byteswap_ushort(_data);}
 inline u32 swap32(u32 _data) {return _byteswap_ulong (_data);}
@@ -92,52 +91,29 @@ inline u64 swap64(u64 data) {return ((u64)swap32(data) << 32) | swap32(data >> 3
 #endif
 
 inline float swapf(float f) {
-    union {
-        float f;
-        unsigned int u32;
-    } dat1, dat2;
-
-    dat1.f = f;
-    dat2.u32 = swap32(dat1.u32);
+    static_assert(sizeof(u32) == sizeof(float),
+                  "float must be the same size as uint32_t.");
 
-    return dat2.f;
-}
-
-inline double swapd(double f) {
-    union  {
-        double f;
-        unsigned long long u64;
-    } dat1, dat2;
+    u32 value;
+    std::memcpy(&value, &f, sizeof(u32));
 
-    dat1.f = f;
-    dat2.u64 = swap64(dat1.u64);
+    value = swap32(value);
+    std::memcpy(&f, &value, sizeof(u32));
 
-    return dat2.f;
+    return f;
 }
 
-inline u16 swap16(const u8* _pData) {return swap16(*(const u16*)_pData);}
-inline u32 swap32(const u8* _pData) {return swap32(*(const u32*)_pData);}
-inline u64 swap64(const u8* _pData) {return swap64(*(const u64*)_pData);}
-
-template <int count>
-void swap(u8*);
+inline double swapd(double f) {
+    static_assert(sizeof(u64) == sizeof(double),
+                  "double must be the same size as uint64_t.");
 
-template <>
-inline void swap<1>(u8* data) { }
+    u64 value;
+    std::memcpy(&value, &f, sizeof(u64));
 
-template <>
-inline void swap<2>(u8* data) {
-    *reinterpret_cast<u16*>(data) = swap16(data);
-}
-
-template <>
-inline void swap<4>(u8* data) {
-    *reinterpret_cast<u32*>(data) = swap32(data);
-}
+    value = swap64(value);
+    std::memcpy(&f, &value, sizeof(u64));
 
-template <>
-inline void swap<8>(u8* data) {
-    *reinterpret_cast<u64*>(data) = swap64(data);
+    return f;
 }
 
 }  // Namespace Common
@@ -534,35 +510,35 @@ bool operator==(const S &p, const swap_struct_t<T, F> v) {
 template <typename T>
 struct swap_64_t {
     static T swap(T x) {
-        return (T)Common::swap64(*(u64 *)&x);
+        return static_cast<T>(Common::swap64(x));
     }
 };
 
 template <typename T>
 struct swap_32_t {
     static T swap(T x) {
-        return (T)Common::swap32(*(u32 *)&x);
+        return static_cast<T>(Common::swap32(x));
     }
 };
 
 template <typename T>
 struct swap_16_t {
     static T swap(T x) {
-        return (T)Common::swap16(*(u16 *)&x);
+        return static_cast<T>(Common::swap16(x));
     }
 };
 
 template <typename T>
 struct swap_float_t {
     static T swap(T x) {
-        return (T)Common::swapf(*(float *)&x);
+        return static_cast<T>(Common::swapf(x));
     }
 };
 
 template <typename T>
 struct swap_double_t {
     static T swap(T x) {
-        return (T)Common::swapd(*(double *)&x);
+        return static_cast<T>(Common::swapd(x));
     }
 };
 
diff --git a/src/core/arm/dyncom/arm_dyncom.cpp b/src/core/arm/dyncom/arm_dyncom.cpp
index a3581132c..13492a08b 100644
--- a/src/core/arm/dyncom/arm_dyncom.cpp
+++ b/src/core/arm/dyncom/arm_dyncom.cpp
@@ -93,7 +93,7 @@ void ARM_DynCom::ResetContext(Core::ThreadContext& context, u32 stack_top, u32 e
     context.cpu_registers[0] = arg;
     context.pc = entry_point;
     context.sp = stack_top;
-    context.cpsr = 0x1F | ((entry_point & 1) << 5); // Usermode and THUMB mode
+    context.cpsr = USER32MODE | ((entry_point & 1) << 5); // Usermode and THUMB mode
 }
 
 void ARM_DynCom::SaveContext(Core::ThreadContext& ctx) {
diff --git a/src/core/gdbstub/gdbstub.cpp b/src/core/gdbstub/gdbstub.cpp
index 1360ee845..820b19e1a 100644
--- a/src/core/gdbstub/gdbstub.cpp
+++ b/src/core/gdbstub/gdbstub.cpp
@@ -437,7 +437,7 @@ static void HandleSetThread() {
  *
  * @param signal Signal to be sent to client.
  */
-void SendSignal(u32 signal) {
+static void SendSignal(u32 signal) {
     if (gdbserver_socket == -1) {
         return;
     }
@@ -713,7 +713,7 @@ static void Continue() {
  * @param addr Address of breakpoint.
  * @param len Length of breakpoint.
  */
-bool CommitBreakpoint(BreakpointType type, PAddr addr, u32 len) {
+static bool CommitBreakpoint(BreakpointType type, PAddr addr, u32 len) {
     std::map<u32, Breakpoint>& p = GetBreakpointList(type);
 
     Breakpoint breakpoint;
@@ -907,7 +907,7 @@ void ToggleServer(bool status) {
     }
 }
 
-void Init(u16 port) {
+static void Init(u16 port) {
     if (!g_server_enabled) {
         // Set the halt loop to false in case the user enabled the gdbstub mid-execution.
         // This way the CPU can still execute normally.
diff --git a/src/core/hle/kernel/memory.cpp b/src/core/hle/kernel/memory.cpp
index 6f1f0856c..4be20db22 100644
--- a/src/core/hle/kernel/memory.cpp
+++ b/src/core/hle/kernel/memory.cpp
@@ -108,7 +108,6 @@ struct MemoryArea {
 // We don't declare the IO regions in here since its handled by other means.
 static MemoryArea memory_areas[] = {
     {VRAM_VADDR,          VRAM_SIZE,              "VRAM"},          // Video memory (VRAM)
-    {TLS_AREA_VADDR,      TLS_AREA_SIZE,          "TLS Area"},      // TLS memory
 };
 
 }
diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h
index a06afef2b..d781ef32c 100644
--- a/src/core/hle/kernel/process.h
+++ b/src/core/hle/kernel/process.h
@@ -142,8 +142,11 @@ public:
 
     MemoryRegionInfo* memory_region = nullptr;
 
-    /// Bitmask of the used TLS slots
-    std::bitset<300> used_tls_slots;
+    /// The Thread Local Storage area is allocated as processes create threads,
+    /// each TLS area is 0x200 bytes, so one page (0x1000) is split up in 8 parts, and each part
+    /// holds the TLS for a specific thread. This vector contains which parts are in use for each page as a bitmask.
+    /// This vector will grow as more pages are allocated for new threads.
+    std::vector<std::bitset<8>> tls_slots;
 
     VAddr GetLinearHeapAreaAddress() const;
     VAddr GetLinearHeapBase() const;
diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp
index 6dc95d0f1..68f026918 100644
--- a/src/core/hle/kernel/thread.cpp
+++ b/src/core/hle/kernel/thread.cpp
@@ -117,9 +117,10 @@ void Thread::Stop() {
     }
     wait_objects.clear();
 
-    Kernel::g_current_process->used_tls_slots[tls_index] = false;
-    g_current_process->misc_memory_used -= Memory::TLS_ENTRY_SIZE;
-    g_current_process->memory_region->used -= Memory::TLS_ENTRY_SIZE;
+    // Mark the TLS slot in the thread's page as free.
+    u32 tls_page = (tls_address - Memory::TLS_AREA_VADDR) / Memory::PAGE_SIZE;
+    u32 tls_slot = ((tls_address - Memory::TLS_AREA_VADDR) % Memory::PAGE_SIZE) / Memory::TLS_ENTRY_SIZE;
+    Kernel::g_current_process->tls_slots[tls_page].reset(tls_slot);
 
     HLE::Reschedule(__func__);
 }
@@ -366,6 +367,31 @@ static void DebugThreadQueue() {
     }
 }
 
+/**
+ * Finds a free location for the TLS section of a thread.
+ * @param tls_slots The TLS page array of the thread's owner process.
+ * Returns a tuple of (page, slot, alloc_needed) where:
+ * page: The index of the first allocated TLS page that has free slots.
+ * slot: The index of the first free slot in the indicated page.
+ * alloc_needed: Whether there's a need to allocate a new TLS page (All pages are full).
+ */
+std::tuple<u32, u32, bool> GetFreeThreadLocalSlot(std::vector<std::bitset<8>>& tls_slots) {
+    // Iterate over all the allocated pages, and try to find one where not all slots are used.
+    for (unsigned page = 0; page < tls_slots.size(); ++page) {
+        const auto& page_tls_slots = tls_slots[page];
+        if (!page_tls_slots.all()) {
+            // We found a page with at least one free slot, find which slot it is
+            for (unsigned slot = 0; slot < page_tls_slots.size(); ++slot) {
+                if (!page_tls_slots.test(slot)) {
+                    return std::make_tuple(page, slot, false);
+                }
+            }
+        }
+    }
+
+    return std::make_tuple(0, 0, true);
+}
+
 ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point, s32 priority,
         u32 arg, s32 processor_id, VAddr stack_top) {
     if (priority < THREADPRIO_HIGHEST || priority > THREADPRIO_LOWEST) {
@@ -403,22 +429,50 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
     thread->name = std::move(name);
     thread->callback_handle = wakeup_callback_handle_table.Create(thread).MoveFrom();
     thread->owner_process = g_current_process;
-    thread->tls_index = -1;
     thread->waitsynch_waited = false;
 
     // Find the next available TLS index, and mark it as used
-    auto& used_tls_slots = Kernel::g_current_process->used_tls_slots;
-    for (unsigned int i = 0; i < used_tls_slots.size(); ++i) {
-        if (used_tls_slots[i] == false) {
-            thread->tls_index = i;
-            used_tls_slots[i] = true;
-            break;
+    auto& tls_slots = Kernel::g_current_process->tls_slots;
+    bool needs_allocation = true;
+    u32 available_page; // Which allocated page has free space
+    u32 available_slot; // Which slot within the page is free
+
+    std::tie(available_page, available_slot, needs_allocation) = GetFreeThreadLocalSlot(tls_slots);
+
+    if (needs_allocation) {
+        // There are no already-allocated pages with free slots, lets allocate a new one.
+        // TLS pages are allocated from the BASE region in the linear heap.
+        MemoryRegionInfo* memory_region = GetMemoryRegion(MemoryRegion::BASE);
+        auto& linheap_memory = memory_region->linear_heap_memory;
+
+        if (linheap_memory->size() + Memory::PAGE_SIZE > memory_region->size) {
+            LOG_ERROR(Kernel_SVC, "Not enough space in region to allocate a new TLS page for thread");
+            return ResultCode(ErrorDescription::OutOfMemory, ErrorModule::Kernel, ErrorSummary::OutOfResource, ErrorLevel::Permanent);
         }
+
+        u32 offset = linheap_memory->size();
+
+        // Allocate some memory from the end of the linear heap for this region.
+        linheap_memory->insert(linheap_memory->end(), Memory::PAGE_SIZE, 0);
+        memory_region->used += Memory::PAGE_SIZE;
+        Kernel::g_current_process->linear_heap_used += Memory::PAGE_SIZE;
+
+        tls_slots.emplace_back(0); // The page is completely available at the start
+        available_page = tls_slots.size() - 1;
+        available_slot = 0; // Use the first slot in the new page
+
+        auto& vm_manager = Kernel::g_current_process->vm_manager;
+        vm_manager.RefreshMemoryBlockMappings(linheap_memory.get());
+
+        // Map the page to the current process' address space.
+        // TODO(Subv): Find the correct MemoryState for this region.
+        vm_manager.MapMemoryBlock(Memory::TLS_AREA_VADDR + available_page * Memory::PAGE_SIZE,
+                                  linheap_memory, offset, Memory::PAGE_SIZE, MemoryState::Private);
     }
 
-    ASSERT_MSG(thread->tls_index != -1, "Out of TLS space");
-    g_current_process->misc_memory_used += Memory::TLS_ENTRY_SIZE;
-    g_current_process->memory_region->used += Memory::TLS_ENTRY_SIZE;
+    // Mark the slot as used
+    tls_slots[available_page].set(available_slot);
+    thread->tls_address = Memory::TLS_AREA_VADDR + available_page * Memory::PAGE_SIZE + available_slot * Memory::TLS_ENTRY_SIZE;
 
     // TODO(peachum): move to ScheduleThread() when scheduler is added so selected core is used
     // to initialize the context
@@ -509,10 +563,6 @@ void Thread::SetWaitSynchronizationOutput(s32 output) {
     context.cpu_registers[1] = output;
 }
 
-VAddr Thread::GetTLSAddress() const {
-    return Memory::TLS_AREA_VADDR + tls_index * Memory::TLS_ENTRY_SIZE;
-}
-
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
 void ThreadingInit() {
diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h
index 97ba57fc5..deab5d5a6 100644
--- a/src/core/hle/kernel/thread.h
+++ b/src/core/hle/kernel/thread.h
@@ -127,7 +127,7 @@ public:
      * Returns the Thread Local Storage address of the current thread
      * @returns VAddr of the thread's TLS
      */
-    VAddr GetTLSAddress() const;
+    VAddr GetTLSAddress() const { return tls_address; }
 
     Core::ThreadContext context;
 
@@ -144,7 +144,7 @@ public:
 
     s32 processor_id;
 
-    s32 tls_index; ///< Index of the Thread Local Storage of the thread
+    VAddr tls_address; ///< Virtual address of the Thread Local Storage of the thread
 
     bool waitsynch_waited; ///< Set to true if the last svcWaitSynch call caused the thread to wait
 
diff --git a/src/core/memory.h b/src/core/memory.h
index 9caa3c3f5..126d60471 100644
--- a/src/core/memory.h
+++ b/src/core/memory.h
@@ -100,15 +100,9 @@ enum : VAddr {
     SHARED_PAGE_SIZE      = 0x00001000,
     SHARED_PAGE_VADDR_END = SHARED_PAGE_VADDR + SHARED_PAGE_SIZE,
 
-    // TODO(yuriks): The size of this area is dynamic, the kernel grows
-    // it as more and more threads are created. For now we'll just use a
-    // hardcoded value.
     /// Area where TLS (Thread-Local Storage) buffers are allocated.
     TLS_AREA_VADDR     = 0x1FF82000,
     TLS_ENTRY_SIZE     = 0x200,
-    TLS_AREA_SIZE      = 300 * TLS_ENTRY_SIZE + 0x800, // Space for up to 300 threads + round to page size
-    TLS_AREA_VADDR_END = TLS_AREA_VADDR + TLS_AREA_SIZE,
-
 
     /// Equivalent to LINEAR_HEAP_VADDR, but expanded to cover the extra memory in the New 3DS.
     NEW_LINEAR_HEAP_VADDR     = 0x30000000,
diff --git a/src/video_core/clipper.cpp b/src/video_core/clipper.cpp
index 2bc747102..db99ce666 100644
--- a/src/video_core/clipper.cpp
+++ b/src/video_core/clipper.cpp
@@ -75,8 +75,6 @@ static void InitScreenCoordinates(OutputVertex& vtx)
     viewport.halfsize_y = float24::FromRaw(regs.viewport_size_y);
     viewport.offset_x   = float24::FromFloat32(static_cast<float>(regs.viewport_corner.x));
     viewport.offset_y   = float24::FromFloat32(static_cast<float>(regs.viewport_corner.y));
-    viewport.zscale     = float24::FromRaw(regs.viewport_depth_range);
-    viewport.offset_z   = float24::FromRaw(regs.viewport_depth_far_plane);
 
     float24 inv_w = float24::FromFloat32(1.f) / vtx.pos.w;
     vtx.color *= inv_w;
@@ -89,7 +87,7 @@ static void InitScreenCoordinates(OutputVertex& vtx)
 
     vtx.screenpos[0] = (vtx.pos.x * inv_w + float24::FromFloat32(1.0)) * viewport.halfsize_x + viewport.offset_x;
     vtx.screenpos[1] = (vtx.pos.y * inv_w + float24::FromFloat32(1.0)) * viewport.halfsize_y + viewport.offset_y;
-    vtx.screenpos[2] = viewport.offset_z + vtx.pos.z * inv_w * viewport.zscale;
+    vtx.screenpos[2] = vtx.pos.z * inv_w;
 }
 
 void ProcessTriangle(const OutputVertex &v0, const OutputVertex &v1, const OutputVertex &v2) {
diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp
index dd1379503..ad0da796e 100644
--- a/src/video_core/command_processor.cpp
+++ b/src/video_core/command_processor.cpp
@@ -128,7 +128,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
 
                 // TODO: Verify that this actually modifies the register!
                 if (setup.index < 15) {
-                    g_state.vs.default_attributes[setup.index] = attribute;
+                    g_state.vs_default_attributes[setup.index] = attribute;
                     setup.index++;
                 } else {
                     // Put each attribute into an immediate input buffer.
@@ -144,12 +144,12 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
                         immediate_attribute_id = 0;
 
                         Shader::UnitState<false> shader_unit;
-                        Shader::Setup();
+                        g_state.vs.Setup();
 
                         // Send to vertex shader
                         if (g_debug_context)
                             g_debug_context->OnEvent(DebugContext::Event::VertexShaderInvocation, static_cast<void*>(&immediate_input));
-                        Shader::OutputVertex output = Shader::Run(shader_unit, immediate_input, regs.vs.num_input_attributes+1);
+                        Shader::OutputVertex output = g_state.vs.Run(shader_unit, immediate_input, regs.vs.num_input_attributes+1);
 
                         // Send to renderer
                         using Pica::Shader::OutputVertex;
@@ -237,7 +237,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
             vertex_cache_ids.fill(-1);
 
             Shader::UnitState<false> shader_unit;
-            Shader::Setup();
+            g_state.vs.Setup();
 
             for (unsigned int index = 0; index < regs.num_vertices; ++index)
             {
@@ -274,7 +274,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
                     // Send to vertex shader
                     if (g_debug_context)
                         g_debug_context->OnEvent(DebugContext::Event::VertexShaderInvocation, (void*)&input);
-                    output = Shader::Run(shader_unit, input, loader.GetNumTotalAttributes());
+                    output = g_state.vs.Run(shader_unit, input, loader.GetNumTotalAttributes());
 
                     if (is_indexed) {
                         vertex_cache[vertex_cache_pos] = output;
diff --git a/src/video_core/pica.cpp b/src/video_core/pica.cpp
index be82cf4b5..ec78f9593 100644
--- a/src/video_core/pica.cpp
+++ b/src/video_core/pica.cpp
@@ -500,7 +500,7 @@ void Init() {
 }
 
 void Shutdown() {
-    Shader::Shutdown();
+    Shader::ClearCache();
 }
 
 template <typename T>
diff --git a/src/video_core/pica.h b/src/video_core/pica.h
index 5891fb72a..86c0a0096 100644
--- a/src/video_core/pica.h
+++ b/src/video_core/pica.h
@@ -70,7 +70,7 @@ struct Regs {
     INSERT_PADDING_WORDS(0x9);
 
     BitField<0, 24, u32> viewport_depth_range; // float24
-    BitField<0, 24, u32> viewport_depth_far_plane; // float24
+    BitField<0, 24, u32> viewport_depth_near_plane; // float24
 
     BitField<0, 3, u32> vs_output_total;
 
@@ -122,9 +122,31 @@ struct Regs {
         BitField<16, 10, s32> y;
     } viewport_corner;
 
-    INSERT_PADDING_WORDS(0x17);
+    INSERT_PADDING_WORDS(0x1);
+
+    //TODO: early depth
+    INSERT_PADDING_WORDS(0x1);
+
+    INSERT_PADDING_WORDS(0x2);
+
+    enum DepthBuffering : u32 {
+        WBuffering  = 0,
+        ZBuffering  = 1,
+    };
+    BitField< 0, 1, DepthBuffering> depthmap_enable;
+
+    INSERT_PADDING_WORDS(0x12);
 
     struct TextureConfig {
+        enum TextureType : u32 {
+            Texture2D    = 0,
+            TextureCube  = 1,
+            Shadow2D     = 2,
+            Projection2D = 3,
+            ShadowCube   = 4,
+            Disabled     = 5,
+        };
+
         enum WrapMode : u32 {
             ClampToEdge    = 0,
             ClampToBorder  = 1,
@@ -155,6 +177,7 @@ struct Regs {
             BitField< 2, 1, TextureFilter> min_filter;
             BitField< 8, 2, WrapMode> wrap_t;
             BitField<12, 2, WrapMode> wrap_s;
+            BitField<28, 2, TextureType> type; ///< @note Only valid for texture 0 according to 3DBrew.
         };
 
         INSERT_PADDING_WORDS(0x1);
@@ -1279,10 +1302,11 @@ ASSERT_REG_POSITION(cull_mode, 0x40);
 ASSERT_REG_POSITION(viewport_size_x, 0x41);
 ASSERT_REG_POSITION(viewport_size_y, 0x43);
 ASSERT_REG_POSITION(viewport_depth_range, 0x4d);
-ASSERT_REG_POSITION(viewport_depth_far_plane, 0x4e);
+ASSERT_REG_POSITION(viewport_depth_near_plane, 0x4e);
 ASSERT_REG_POSITION(vs_output_attributes[0], 0x50);
 ASSERT_REG_POSITION(vs_output_attributes[1], 0x51);
 ASSERT_REG_POSITION(viewport_corner, 0x68);
+ASSERT_REG_POSITION(depthmap_enable, 0x6D);
 ASSERT_REG_POSITION(texture0_enable, 0x80);
 ASSERT_REG_POSITION(texture0, 0x81);
 ASSERT_REG_POSITION(texture0_format, 0x8e);
diff --git a/src/video_core/pica_state.h b/src/video_core/pica_state.h
index 1059c6ae4..495174c25 100644
--- a/src/video_core/pica_state.h
+++ b/src/video_core/pica_state.h
@@ -25,6 +25,8 @@ struct State {
     Shader::ShaderSetup vs;
     Shader::ShaderSetup gs;
 
+    std::array<Math::Vec4<float24>, 16> vs_default_attributes;
+
     struct {
         union LutEntry {
             // Used for raw access
diff --git a/src/video_core/rasterizer.cpp b/src/video_core/rasterizer.cpp
index df67b9081..65168f05a 100644
--- a/src/video_core/rasterizer.cpp
+++ b/src/video_core/rasterizer.cpp
@@ -442,8 +442,33 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0,
 
                 DEBUG_ASSERT(0 != texture.config.address);
 
-                int s = (int)(uv[i].u() * float24::FromFloat32(static_cast<float>(texture.config.width))).ToFloat32();
-                int t = (int)(uv[i].v() * float24::FromFloat32(static_cast<float>(texture.config.height))).ToFloat32();
+                float24 u = uv[i].u();
+                float24 v = uv[i].v();
+
+                // Only unit 0 respects the texturing type (according to 3DBrew)
+                // TODO: Refactor so cubemaps and shadowmaps can be handled
+                if (i == 0) {
+                    switch(texture.config.type) {
+                    case Regs::TextureConfig::Texture2D:
+                        break;
+                    case Regs::TextureConfig::Projection2D: {
+                        auto tc0_w = GetInterpolatedAttribute(v0.tc0_w, v1.tc0_w, v2.tc0_w);
+                        u /= tc0_w;
+                        v /= tc0_w;
+                        break;
+                    }
+                    default:
+                        // TODO: Change to LOG_ERROR when more types are handled.
+                        LOG_DEBUG(HW_GPU, "Unhandled texture type %x", (int)texture.config.type);
+                        UNIMPLEMENTED();
+                        break;
+                    }
+                }
+
+                int s = (int)(u * float24::FromFloat32(static_cast<float>(texture.config.width))).ToFloat32();
+                int t = (int)(v * float24::FromFloat32(static_cast<float>(texture.config.height))).ToFloat32();
+
+
                 static auto GetWrappedTexCoord = [](Regs::TextureConfig::WrapMode mode, int val, unsigned size) {
                     switch (mode) {
                         case Regs::TextureConfig::ClampToEdge:
@@ -862,10 +887,30 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0,
                 }
             }
 
+            // interpolated_z = z / w
+            float interpolated_z_over_w = (v0.screenpos[2].ToFloat32() * w0 +
+                                           v1.screenpos[2].ToFloat32() * w1 +
+                                           v2.screenpos[2].ToFloat32() * w2) / wsum;
+
+            // Not fully accurate. About 3 bits in precision are missing.
+            // Z-Buffer (z / w * scale + offset)
+            float depth_scale = float24::FromRaw(regs.viewport_depth_range).ToFloat32();
+            float depth_offset = float24::FromRaw(regs.viewport_depth_near_plane).ToFloat32();
+            float depth = interpolated_z_over_w * depth_scale + depth_offset;
+
+            // Potentially switch to W-Buffer
+            if (regs.depthmap_enable == Pica::Regs::DepthBuffering::WBuffering) {
+
+                // W-Buffer (z * scale + w * offset = (z / w * scale + offset) * w)
+                depth *= interpolated_w_inverse.ToFloat32() * wsum;
+            }
+
+            // Clamp the result
+            depth = MathUtil::Clamp(depth, 0.0f, 1.0f);
+
+            // Convert float to integer
             unsigned num_bits = Regs::DepthBitsPerPixel(regs.framebuffer.depth_format);
-            u32 z = (u32)((v0.screenpos[2].ToFloat32() * w0 +
-                           v1.screenpos[2].ToFloat32() * w1 +
-                           v2.screenpos[2].ToFloat32() * w2) * ((1 << num_bits) - 1) / wsum);
+            u32 z = (u32)(depth * ((1 << num_bits) - 1));
 
             if (output_merger.depth_test_enable) {
                 u32 ref_z = GetDepth(x >> 4, y >> 4);
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 0b471dfd2..ed2e2f3ae 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -76,6 +76,9 @@ RasterizerOpenGL::RasterizerOpenGL() : shader_dirty(true) {
     glEnableVertexAttribArray(GLShader::ATTRIBUTE_TEXCOORD1);
     glEnableVertexAttribArray(GLShader::ATTRIBUTE_TEXCOORD2);
 
+    glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD0_W, 1, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord0_w));
+    glEnableVertexAttribArray(GLShader::ATTRIBUTE_TEXCOORD0_W);
+
     glVertexAttribPointer(GLShader::ATTRIBUTE_NORMQUAT, 4, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, normquat));
     glEnableVertexAttribArray(GLShader::ATTRIBUTE_NORMQUAT);
 
@@ -256,10 +259,15 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) {
 
     // Depth modifiers
     case PICA_REG_INDEX(viewport_depth_range):
-    case PICA_REG_INDEX(viewport_depth_far_plane):
+    case PICA_REG_INDEX(viewport_depth_near_plane):
         SyncDepthModifiers();
         break;
 
+    // Depth buffering
+    case PICA_REG_INDEX(depthmap_enable):
+        shader_dirty = true;
+        break;
+
     // Blending
     case PICA_REG_INDEX(output_merger.alphablend_enable):
         SyncBlendEnabled();
@@ -314,6 +322,11 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) {
         SyncLogicOp();
         break;
 
+    // Texture 0 type
+    case PICA_REG_INDEX(texture0.type):
+        shader_dirty = true;
+        break;
+
     // TEV stages
     case PICA_REG_INDEX(tev_stage0.color_source1):
     case PICA_REG_INDEX(tev_stage0.color_modifier1):
@@ -910,10 +923,10 @@ void RasterizerOpenGL::SyncCullMode() {
 }
 
 void RasterizerOpenGL::SyncDepthModifiers() {
-    float depth_scale = -Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_range).ToFloat32();
-    float depth_offset = Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_far_plane).ToFloat32() / 2.0f;
+    float depth_scale = Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_range).ToFloat32();
+    float depth_offset = Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_near_plane).ToFloat32();
 
-    // TODO: Implement scale modifier
+    uniform_block_data.data.depth_scale = depth_scale;
     uniform_block_data.data.depth_offset = depth_offset;
     uniform_block_data.dirty = true;
 }
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index 82fa61742..eed00011a 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -39,140 +39,185 @@ struct ScreenInfo;
  * directly accessing Pica registers. This should reduce the risk of bugs in shader generation where
  * Pica state is not being captured in the shader cache key, thereby resulting in (what should be)
  * two separate shaders sharing the same key.
+ *
+ * We use a union because "implicitly-defined copy/move constructor for a union X copies the object representation of X."
+ * and "implicitly-defined copy assignment operator for a union X copies the object representation (3.9) of X."
+ * = Bytewise copy instead of memberwise copy.
+ * This is important because the padding bytes are included in the hash and comparison between objects.
  */
-struct PicaShaderConfig {
+union PicaShaderConfig {
+
     /// Construct a PicaShaderConfig with the current Pica register configuration.
     static PicaShaderConfig CurrentConfig() {
         PicaShaderConfig res;
+
+        auto& state = res.state;
+        std::memset(&state, 0, sizeof(PicaShaderConfig::State));
+
         const auto& regs = Pica::g_state.regs;
 
-        res.alpha_test_func = regs.output_merger.alpha_test.enable ?
+        state.depthmap_enable = regs.depthmap_enable;
+
+        state.alpha_test_func = regs.output_merger.alpha_test.enable ?
             regs.output_merger.alpha_test.func.Value() : Pica::Regs::CompareFunc::Always;
 
+        state.texture0_type = regs.texture0.type;
+
         // Copy relevant tev stages fields.
         // We don't sync const_color here because of the high variance, it is a
         // shader uniform instead.
         const auto& tev_stages = regs.GetTevStages();
-        DEBUG_ASSERT(res.tev_stages.size() == tev_stages.size());
+        DEBUG_ASSERT(state.tev_stages.size() == tev_stages.size());
         for (size_t i = 0; i < tev_stages.size(); i++) {
             const auto& tev_stage = tev_stages[i];
-            res.tev_stages[i].sources_raw = tev_stage.sources_raw;
-            res.tev_stages[i].modifiers_raw = tev_stage.modifiers_raw;
-            res.tev_stages[i].ops_raw = tev_stage.ops_raw;
-            res.tev_stages[i].scales_raw = tev_stage.scales_raw;
+            state.tev_stages[i].sources_raw = tev_stage.sources_raw;
+            state.tev_stages[i].modifiers_raw = tev_stage.modifiers_raw;
+            state.tev_stages[i].ops_raw = tev_stage.ops_raw;
+            state.tev_stages[i].scales_raw = tev_stage.scales_raw;
         }
 
-        res.combiner_buffer_input =
+        state.combiner_buffer_input =
             regs.tev_combiner_buffer_input.update_mask_rgb.Value() |
             regs.tev_combiner_buffer_input.update_mask_a.Value() << 4;
 
         // Fragment lighting
 
-        res.lighting.enable = !regs.lighting.disable;
-        res.lighting.src_num = regs.lighting.num_lights + 1;
+        state.lighting.enable = !regs.lighting.disable;
+        state.lighting.src_num = regs.lighting.num_lights + 1;
 
-        for (unsigned light_index = 0; light_index < res.lighting.src_num; ++light_index) {
+        for (unsigned light_index = 0; light_index < state.lighting.src_num; ++light_index) {
             unsigned num = regs.lighting.light_enable.GetNum(light_index);
             const auto& light = regs.lighting.light[num];
-            res.lighting.light[light_index].num = num;
-            res.lighting.light[light_index].directional = light.directional != 0;
-            res.lighting.light[light_index].two_sided_diffuse = light.two_sided_diffuse != 0;
-            res.lighting.light[light_index].dist_atten_enable = !regs.lighting.IsDistAttenDisabled(num);
-            res.lighting.light[light_index].dist_atten_bias = Pica::float20::FromRaw(light.dist_atten_bias).ToFloat32();
-            res.lighting.light[light_index].dist_atten_scale = Pica::float20::FromRaw(light.dist_atten_scale).ToFloat32();
+            state.lighting.light[light_index].num = num;
+            state.lighting.light[light_index].directional = light.directional != 0;
+            state.lighting.light[light_index].two_sided_diffuse = light.two_sided_diffuse != 0;
+            state.lighting.light[light_index].dist_atten_enable = !regs.lighting.IsDistAttenDisabled(num);
+            state.lighting.light[light_index].dist_atten_bias = Pica::float20::FromRaw(light.dist_atten_bias).ToFloat32();
+            state.lighting.light[light_index].dist_atten_scale = Pica::float20::FromRaw(light.dist_atten_scale).ToFloat32();
         }
 
-        res.lighting.lut_d0.enable = regs.lighting.disable_lut_d0 == 0;
-        res.lighting.lut_d0.abs_input = regs.lighting.abs_lut_input.disable_d0 == 0;
-        res.lighting.lut_d0.type = regs.lighting.lut_input.d0.Value();
-        res.lighting.lut_d0.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.d0);
-
-        res.lighting.lut_d1.enable = regs.lighting.disable_lut_d1 == 0;
-        res.lighting.lut_d1.abs_input = regs.lighting.abs_lut_input.disable_d1 == 0;
-        res.lighting.lut_d1.type = regs.lighting.lut_input.d1.Value();
-        res.lighting.lut_d1.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.d1);
-
-        res.lighting.lut_fr.enable = regs.lighting.disable_lut_fr == 0;
-        res.lighting.lut_fr.abs_input = regs.lighting.abs_lut_input.disable_fr == 0;
-        res.lighting.lut_fr.type = regs.lighting.lut_input.fr.Value();
-        res.lighting.lut_fr.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.fr);
-
-        res.lighting.lut_rr.enable = regs.lighting.disable_lut_rr == 0;
-        res.lighting.lut_rr.abs_input = regs.lighting.abs_lut_input.disable_rr == 0;
-        res.lighting.lut_rr.type = regs.lighting.lut_input.rr.Value();
-        res.lighting.lut_rr.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.rr);
-
-        res.lighting.lut_rg.enable = regs.lighting.disable_lut_rg == 0;
-        res.lighting.lut_rg.abs_input = regs.lighting.abs_lut_input.disable_rg == 0;
-        res.lighting.lut_rg.type = regs.lighting.lut_input.rg.Value();
-        res.lighting.lut_rg.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.rg);
-
-        res.lighting.lut_rb.enable = regs.lighting.disable_lut_rb == 0;
-        res.lighting.lut_rb.abs_input = regs.lighting.abs_lut_input.disable_rb == 0;
-        res.lighting.lut_rb.type = regs.lighting.lut_input.rb.Value();
-        res.lighting.lut_rb.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.rb);
-
-        res.lighting.config = regs.lighting.config;
-        res.lighting.fresnel_selector = regs.lighting.fresnel_selector;
-        res.lighting.bump_mode = regs.lighting.bump_mode;
-        res.lighting.bump_selector = regs.lighting.bump_selector;
-        res.lighting.bump_renorm = regs.lighting.disable_bump_renorm == 0;
-        res.lighting.clamp_highlights = regs.lighting.clamp_highlights != 0;
+        state.lighting.lut_d0.enable = regs.lighting.disable_lut_d0 == 0;
+        state.lighting.lut_d0.abs_input = regs.lighting.abs_lut_input.disable_d0 == 0;
+        state.lighting.lut_d0.type = regs.lighting.lut_input.d0.Value();
+        state.lighting.lut_d0.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.d0);
+
+        state.lighting.lut_d1.enable = regs.lighting.disable_lut_d1 == 0;
+        state.lighting.lut_d1.abs_input = regs.lighting.abs_lut_input.disable_d1 == 0;
+        state.lighting.lut_d1.type = regs.lighting.lut_input.d1.Value();
+        state.lighting.lut_d1.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.d1);
+
+        state.lighting.lut_fr.enable = regs.lighting.disable_lut_fr == 0;
+        state.lighting.lut_fr.abs_input = regs.lighting.abs_lut_input.disable_fr == 0;
+        state.lighting.lut_fr.type = regs.lighting.lut_input.fr.Value();
+        state.lighting.lut_fr.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.fr);
+
+        state.lighting.lut_rr.enable = regs.lighting.disable_lut_rr == 0;
+        state.lighting.lut_rr.abs_input = regs.lighting.abs_lut_input.disable_rr == 0;
+        state.lighting.lut_rr.type = regs.lighting.lut_input.rr.Value();
+        state.lighting.lut_rr.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.rr);
+
+        state.lighting.lut_rg.enable = regs.lighting.disable_lut_rg == 0;
+        state.lighting.lut_rg.abs_input = regs.lighting.abs_lut_input.disable_rg == 0;
+        state.lighting.lut_rg.type = regs.lighting.lut_input.rg.Value();
+        state.lighting.lut_rg.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.rg);
+
+        state.lighting.lut_rb.enable = regs.lighting.disable_lut_rb == 0;
+        state.lighting.lut_rb.abs_input = regs.lighting.abs_lut_input.disable_rb == 0;
+        state.lighting.lut_rb.type = regs.lighting.lut_input.rb.Value();
+        state.lighting.lut_rb.scale = regs.lighting.lut_scale.GetScale(regs.lighting.lut_scale.rb);
+
+        state.lighting.config = regs.lighting.config;
+        state.lighting.fresnel_selector = regs.lighting.fresnel_selector;
+        state.lighting.bump_mode = regs.lighting.bump_mode;
+        state.lighting.bump_selector = regs.lighting.bump_selector;
+        state.lighting.bump_renorm = regs.lighting.disable_bump_renorm == 0;
+        state.lighting.clamp_highlights = regs.lighting.clamp_highlights != 0;
 
         return res;
     }
 
     bool TevStageUpdatesCombinerBufferColor(unsigned stage_index) const {
-        return (stage_index < 4) && (combiner_buffer_input & (1 << stage_index));
+        return (stage_index < 4) && (state.combiner_buffer_input & (1 << stage_index));
     }
 
     bool TevStageUpdatesCombinerBufferAlpha(unsigned stage_index) const {
-        return (stage_index < 4) && ((combiner_buffer_input >> 4) & (1 << stage_index));
+        return (stage_index < 4) && ((state.combiner_buffer_input >> 4) & (1 << stage_index));
     }
 
     bool operator ==(const PicaShaderConfig& o) const {
-        return std::memcmp(this, &o, sizeof(PicaShaderConfig)) == 0;
+        return std::memcmp(&state, &o.state, sizeof(PicaShaderConfig::State)) == 0;
     };
 
-    Pica::Regs::CompareFunc alpha_test_func = Pica::Regs::CompareFunc::Never;
-    std::array<Pica::Regs::TevStageConfig, 6> tev_stages = {};
-    u8 combiner_buffer_input = 0;
+    // NOTE: MSVC15 (Update 2) doesn't think `delete`'d constructors and operators are TC.
+    //       This makes BitField not TC when used in a union or struct so we have to resort
+    //       to this ugly hack.
+    //       Once that bug is fixed we can use Pica::Regs::TevStageConfig here.
+    //       Doesn't include const_color because we don't sync it, see comment in CurrentConfig()
+    struct TevStageConfigRaw {
+        u32 sources_raw;
+        u32 modifiers_raw;
+        u32 ops_raw;
+        u32 scales_raw;
+        explicit operator Pica::Regs::TevStageConfig() const noexcept {
+            Pica::Regs::TevStageConfig stage;
+            stage.sources_raw = sources_raw;
+            stage.modifiers_raw = modifiers_raw;
+            stage.ops_raw = ops_raw;
+            stage.const_color = 0;
+            stage.scales_raw = scales_raw;
+            return stage;
+        }
+    };
 
-    struct {
-        struct {
-            unsigned num = 0;
-            bool directional = false;
-            bool two_sided_diffuse = false;
-            bool dist_atten_enable = false;
-            GLfloat dist_atten_scale = 0.0f;
-            GLfloat dist_atten_bias = 0.0f;
-        } light[8];
-
-        bool enable = false;
-        unsigned src_num = 0;
-        Pica::Regs::LightingBumpMode bump_mode = Pica::Regs::LightingBumpMode::None;
-        unsigned bump_selector = 0;
-        bool bump_renorm = false;
-        bool clamp_highlights = false;
-
-        Pica::Regs::LightingConfig config = Pica::Regs::LightingConfig::Config0;
-        Pica::Regs::LightingFresnelSelector fresnel_selector = Pica::Regs::LightingFresnelSelector::None;
+    struct State {
+
+        Pica::Regs::CompareFunc alpha_test_func;
+        Pica::Regs::TextureConfig::TextureType texture0_type;
+        std::array<TevStageConfigRaw, 6> tev_stages;
+        u8 combiner_buffer_input;
+
+        Pica::Regs::DepthBuffering depthmap_enable;
 
         struct {
-            bool enable = false;
-            bool abs_input = false;
-            Pica::Regs::LightingLutInput type = Pica::Regs::LightingLutInput::NH;
-            float scale = 1.0f;
-        } lut_d0, lut_d1, lut_fr, lut_rr, lut_rg, lut_rb;
-    } lighting;
+            struct {
+                unsigned num;
+                bool directional;
+                bool two_sided_diffuse;
+                bool dist_atten_enable;
+                GLfloat dist_atten_scale;
+                GLfloat dist_atten_bias;
+            } light[8];
+
+            bool enable;
+            unsigned src_num;
+            Pica::Regs::LightingBumpMode bump_mode;
+            unsigned bump_selector;
+            bool bump_renorm;
+            bool clamp_highlights;
+
+            Pica::Regs::LightingConfig config;
+            Pica::Regs::LightingFresnelSelector fresnel_selector;
+
+            struct {
+                bool enable;
+                bool abs_input;
+                Pica::Regs::LightingLutInput type;
+                float scale;
+            } lut_d0, lut_d1, lut_fr, lut_rr, lut_rg, lut_rb;
+        } lighting;
+
+    } state;
 };
+#if (__GNUC__ >= 5) || defined(__clang__) || defined(_MSC_VER)
+static_assert(std::is_trivially_copyable<PicaShaderConfig::State>::value, "PicaShaderConfig::State must be trivially copyable");
+#endif
 
 namespace std {
 
 template <>
 struct hash<PicaShaderConfig> {
     size_t operator()(const PicaShaderConfig& k) const {
-        return Common::ComputeHash64(&k, sizeof(PicaShaderConfig));
+        return Common::ComputeHash64(&k.state, sizeof(PicaShaderConfig::State));
     }
 };
 
@@ -239,6 +284,7 @@ private:
             tex_coord1[1] = v.tc1.y.ToFloat32();
             tex_coord2[0] = v.tc2.x.ToFloat32();
             tex_coord2[1] = v.tc2.y.ToFloat32();
+            tex_coord0_w = v.tc0_w.ToFloat32();
             normquat[0] = v.quat.x.ToFloat32();
             normquat[1] = v.quat.y.ToFloat32();
             normquat[2] = v.quat.z.ToFloat32();
@@ -259,6 +305,7 @@ private:
         GLfloat tex_coord0[2];
         GLfloat tex_coord1[2];
         GLfloat tex_coord2[2];
+        GLfloat tex_coord0_w;
         GLfloat normquat[4];
         GLfloat view[3];
     };
@@ -277,6 +324,7 @@ private:
         GLvec4 const_color[6];
         GLvec4 tev_combiner_buffer_color;
         GLint alphatest_ref;
+        GLfloat depth_scale;
         GLfloat depth_offset;
         alignas(16) GLvec3 lighting_global_ambient;
         LightSrc light_src[8];
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index 9011caa39..71d60e69c 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp
@@ -32,8 +32,9 @@ static bool IsPassThroughTevStage(const TevStageConfig& stage) {
 }
 
 /// Writes the specified TEV stage source component(s)
-static void AppendSource(std::string& out, TevStageConfig::Source source,
+static void AppendSource(std::string& out, const PicaShaderConfig& config, TevStageConfig::Source source,
         const std::string& index_name) {
+    const auto& state = config.state;
     using Source = TevStageConfig::Source;
     switch (source) {
     case Source::PrimaryColor:
@@ -46,7 +47,20 @@ static void AppendSource(std::string& out, TevStageConfig::Source source,
         out += "secondary_fragment_color";
         break;
     case Source::Texture0:
-        out += "texture(tex[0], texcoord[0])";
+        // Only unit 0 respects the texturing type (according to 3DBrew)
+        switch(state.texture0_type) {
+        case Pica::Regs::TextureConfig::Texture2D:
+            out += "texture(tex[0], texcoord[0])";
+            break;
+        case Pica::Regs::TextureConfig::Projection2D:
+            out += "textureProj(tex[0], vec3(texcoord[0], texcoord0_w))";
+            break;
+        default:
+            out += "texture(tex[0], texcoord[0])";
+            LOG_CRITICAL(HW_GPU, "Unhandled texture type %x", static_cast<int>(state.texture0_type));
+            UNIMPLEMENTED();
+            break;
+        }
         break;
     case Source::Texture1:
         out += "texture(tex[1], texcoord[1])";
@@ -71,53 +85,53 @@ static void AppendSource(std::string& out, TevStageConfig::Source source,
 }
 
 /// Writes the color components to use for the specified TEV stage color modifier
-static void AppendColorModifier(std::string& out, TevStageConfig::ColorModifier modifier,
+static void AppendColorModifier(std::string& out, const PicaShaderConfig& config, TevStageConfig::ColorModifier modifier,
         TevStageConfig::Source source, const std::string& index_name) {
     using ColorModifier = TevStageConfig::ColorModifier;
     switch (modifier) {
     case ColorModifier::SourceColor:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".rgb";
         break;
     case ColorModifier::OneMinusSourceColor:
         out += "vec3(1.0) - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".rgb";
         break;
     case ColorModifier::SourceAlpha:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".aaa";
         break;
     case ColorModifier::OneMinusSourceAlpha:
         out += "vec3(1.0) - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".aaa";
         break;
     case ColorModifier::SourceRed:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".rrr";
         break;
     case ColorModifier::OneMinusSourceRed:
         out += "vec3(1.0) - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".rrr";
         break;
     case ColorModifier::SourceGreen:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".ggg";
         break;
     case ColorModifier::OneMinusSourceGreen:
         out += "vec3(1.0) - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".ggg";
         break;
     case ColorModifier::SourceBlue:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".bbb";
         break;
     case ColorModifier::OneMinusSourceBlue:
         out += "vec3(1.0) - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".bbb";
         break;
     default:
@@ -128,44 +142,44 @@ static void AppendColorModifier(std::string& out, TevStageConfig::ColorModifier
 }
 
 /// Writes the alpha component to use for the specified TEV stage alpha modifier
-static void AppendAlphaModifier(std::string& out, TevStageConfig::AlphaModifier modifier,
+static void AppendAlphaModifier(std::string& out, const PicaShaderConfig& config, TevStageConfig::AlphaModifier modifier,
         TevStageConfig::Source source, const std::string& index_name) {
     using AlphaModifier = TevStageConfig::AlphaModifier;
     switch (modifier) {
     case AlphaModifier::SourceAlpha:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".a";
         break;
     case AlphaModifier::OneMinusSourceAlpha:
         out += "1.0 - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".a";
         break;
     case AlphaModifier::SourceRed:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".r";
         break;
     case AlphaModifier::OneMinusSourceRed:
         out += "1.0 - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".r";
         break;
     case AlphaModifier::SourceGreen:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".g";
         break;
     case AlphaModifier::OneMinusSourceGreen:
         out += "1.0 - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".g";
         break;
     case AlphaModifier::SourceBlue:
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".b";
         break;
     case AlphaModifier::OneMinusSourceBlue:
         out += "1.0 - ";
-        AppendSource(out, source, index_name);
+        AppendSource(out, config, source, index_name);
         out += ".b";
         break;
     default:
@@ -287,16 +301,16 @@ static void AppendAlphaTestCondition(std::string& out, Regs::CompareFunc func) {
 
 /// Writes the code to emulate the specified TEV stage
 static void WriteTevStage(std::string& out, const PicaShaderConfig& config, unsigned index) {
-    auto& stage = config.tev_stages[index];
+    const auto stage = static_cast<const Pica::Regs::TevStageConfig>(config.state.tev_stages[index]);
     if (!IsPassThroughTevStage(stage)) {
         std::string index_name = std::to_string(index);
 
         out += "vec3 color_results_" + index_name + "[3] = vec3[3](";
-        AppendColorModifier(out, stage.color_modifier1, stage.color_source1, index_name);
+        AppendColorModifier(out, config, stage.color_modifier1, stage.color_source1, index_name);
         out += ", ";
-        AppendColorModifier(out, stage.color_modifier2, stage.color_source2, index_name);
+        AppendColorModifier(out, config, stage.color_modifier2, stage.color_source2, index_name);
         out += ", ";
-        AppendColorModifier(out, stage.color_modifier3, stage.color_source3, index_name);
+        AppendColorModifier(out, config, stage.color_modifier3, stage.color_source3, index_name);
         out += ");\n";
 
         out += "vec3 color_output_" + index_name + " = ";
@@ -304,11 +318,11 @@ static void WriteTevStage(std::string& out, const PicaShaderConfig& config, unsi
         out += ";\n";
 
         out += "float alpha_results_" + index_name + "[3] = float[3](";
-        AppendAlphaModifier(out, stage.alpha_modifier1, stage.alpha_source1, index_name);
+        AppendAlphaModifier(out, config, stage.alpha_modifier1, stage.alpha_source1, index_name);
         out += ", ";
-        AppendAlphaModifier(out, stage.alpha_modifier2, stage.alpha_source2, index_name);
+        AppendAlphaModifier(out, config, stage.alpha_modifier2, stage.alpha_source2, index_name);
         out += ", ";
-        AppendAlphaModifier(out, stage.alpha_modifier3, stage.alpha_source3, index_name);
+        AppendAlphaModifier(out, config, stage.alpha_modifier3, stage.alpha_source3, index_name);
         out += ");\n";
 
         out += "float alpha_output_" + index_name + " = ";
@@ -331,6 +345,8 @@ static void WriteTevStage(std::string& out, const PicaShaderConfig& config, unsi
 
 /// Writes the code to emulate fragment lighting
 static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
+    const auto& lighting = config.state.lighting;
+
     // Define lighting globals
     out += "vec4 diffuse_sum = vec4(0.0, 0.0, 0.0, 1.0);\n"
            "vec4 specular_sum = vec4(0.0, 0.0, 0.0, 1.0);\n"
@@ -338,17 +354,17 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
            "vec3 refl_value = vec3(0.0);\n";
 
     // Compute fragment normals
-    if (config.lighting.bump_mode == Pica::Regs::LightingBumpMode::NormalMap) {
+    if (lighting.bump_mode == Pica::Regs::LightingBumpMode::NormalMap) {
         // Bump mapping is enabled using a normal map, read perturbation vector from the selected texture
-        std::string bump_selector = std::to_string(config.lighting.bump_selector);
+        std::string bump_selector = std::to_string(lighting.bump_selector);
         out += "vec3 surface_normal = 2.0 * texture(tex[" + bump_selector + "], texcoord[" + bump_selector + "]).rgb - 1.0;\n";
 
         // Recompute Z-component of perturbation if 'renorm' is enabled, this provides a higher precision result
-        if (config.lighting.bump_renorm) {
+        if (lighting.bump_renorm) {
             std::string val = "(1.0 - (surface_normal.x*surface_normal.x + surface_normal.y*surface_normal.y))";
             out += "surface_normal.z = sqrt(max(" + val + ", 0.0));\n";
         }
-    } else if (config.lighting.bump_mode == Pica::Regs::LightingBumpMode::TangentMap) {
+    } else if (lighting.bump_mode == Pica::Regs::LightingBumpMode::TangentMap) {
         // Bump mapping is enabled using a tangent map
         LOG_CRITICAL(HW_GPU, "unimplemented bump mapping mode (tangent mapping)");
         UNIMPLEMENTED();
@@ -361,7 +377,7 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
     out += "vec3 normal = normalize(quaternion_rotate(normquat, surface_normal));\n";
 
     // Gets the index into the specified lookup table for specular lighting
-    auto GetLutIndex = [config](unsigned light_num, Regs::LightingLutInput input, bool abs) {
+    auto GetLutIndex = [&lighting](unsigned light_num, Regs::LightingLutInput input, bool abs) {
         const std::string half_angle = "normalize(normalize(view) + light_vector)";
         std::string index;
         switch (input) {
@@ -389,7 +405,7 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
 
         if (abs) {
             // LUT index is in the range of (0.0, 1.0)
-            index = config.lighting.light[light_num].two_sided_diffuse ? "abs(" + index + ")" : "max(" + index + ", 0.f)";
+            index = lighting.light[light_num].two_sided_diffuse ? "abs(" + index + ")" : "max(" + index + ", 0.f)";
             return "(FLOAT_255 * clamp(" + index + ", 0.0, 1.0))";
         } else {
             // LUT index is in the range of (-1.0, 1.0)
@@ -407,8 +423,8 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
     };
 
     // Write the code to emulate each enabled light
-    for (unsigned light_index = 0; light_index < config.lighting.src_num; ++light_index) {
-        const auto& light_config = config.lighting.light[light_index];
+    for (unsigned light_index = 0; light_index < lighting.src_num; ++light_index) {
+        const auto& light_config = lighting.light[light_index];
         std::string light_src = "light_src[" + std::to_string(light_config.num) + "]";
 
         // Compute light vector (directional or positional)
@@ -432,39 +448,39 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
         }
 
         // If enabled, clamp specular component if lighting result is negative
-        std::string clamp_highlights = config.lighting.clamp_highlights ? "(dot(light_vector, normal) <= 0.0 ? 0.0 : 1.0)" : "1.0";
+        std::string clamp_highlights = lighting.clamp_highlights ? "(dot(light_vector, normal) <= 0.0 ? 0.0 : 1.0)" : "1.0";
 
         // Specular 0 component
         std::string d0_lut_value = "1.0";
-        if (config.lighting.lut_d0.enable && Pica::Regs::IsLightingSamplerSupported(config.lighting.config, Pica::Regs::LightingSampler::Distribution0)) {
+        if (lighting.lut_d0.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::Distribution0)) {
             // Lookup specular "distribution 0" LUT value
-            std::string index = GetLutIndex(light_config.num, config.lighting.lut_d0.type, config.lighting.lut_d0.abs_input);
-            d0_lut_value = "(" + std::to_string(config.lighting.lut_d0.scale) + " * " + GetLutValue(Regs::LightingSampler::Distribution0, index) + ")";
+            std::string index = GetLutIndex(light_config.num, lighting.lut_d0.type, lighting.lut_d0.abs_input);
+            d0_lut_value = "(" + std::to_string(lighting.lut_d0.scale) + " * " + GetLutValue(Regs::LightingSampler::Distribution0, index) + ")";
         }
         std::string specular_0 = "(" + d0_lut_value + " * " + light_src + ".specular_0)";
 
         // If enabled, lookup ReflectRed value, otherwise, 1.0 is used
-        if (config.lighting.lut_rr.enable && Pica::Regs::IsLightingSamplerSupported(config.lighting.config, Pica::Regs::LightingSampler::ReflectRed)) {
-            std::string index = GetLutIndex(light_config.num, config.lighting.lut_rr.type, config.lighting.lut_rr.abs_input);
-            std::string value = "(" + std::to_string(config.lighting.lut_rr.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectRed, index) + ")";
+        if (lighting.lut_rr.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::ReflectRed)) {
+            std::string index = GetLutIndex(light_config.num, lighting.lut_rr.type, lighting.lut_rr.abs_input);
+            std::string value = "(" + std::to_string(lighting.lut_rr.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectRed, index) + ")";
             out += "refl_value.r = " + value + ";\n";
         } else {
             out += "refl_value.r = 1.0;\n";
         }
 
         // If enabled, lookup ReflectGreen value, otherwise, ReflectRed value is used
-        if (config.lighting.lut_rg.enable && Pica::Regs::IsLightingSamplerSupported(config.lighting.config, Pica::Regs::LightingSampler::ReflectGreen)) {
-            std::string index = GetLutIndex(light_config.num, config.lighting.lut_rg.type, config.lighting.lut_rg.abs_input);
-            std::string value = "(" + std::to_string(config.lighting.lut_rg.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectGreen, index) + ")";
+        if (lighting.lut_rg.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::ReflectGreen)) {
+            std::string index = GetLutIndex(light_config.num, lighting.lut_rg.type, lighting.lut_rg.abs_input);
+            std::string value = "(" + std::to_string(lighting.lut_rg.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectGreen, index) + ")";
             out += "refl_value.g = " + value + ";\n";
         } else {
             out += "refl_value.g = refl_value.r;\n";
         }
 
         // If enabled, lookup ReflectBlue value, otherwise, ReflectRed value is used
-        if (config.lighting.lut_rb.enable && Pica::Regs::IsLightingSamplerSupported(config.lighting.config, Pica::Regs::LightingSampler::ReflectBlue)) {
-            std::string index = GetLutIndex(light_config.num, config.lighting.lut_rb.type, config.lighting.lut_rb.abs_input);
-            std::string value = "(" + std::to_string(config.lighting.lut_rb.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectBlue, index) + ")";
+        if (lighting.lut_rb.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::ReflectBlue)) {
+            std::string index = GetLutIndex(light_config.num, lighting.lut_rb.type, lighting.lut_rb.abs_input);
+            std::string value = "(" + std::to_string(lighting.lut_rb.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectBlue, index) + ")";
             out += "refl_value.b = " + value + ";\n";
         } else {
             out += "refl_value.b = refl_value.r;\n";
@@ -472,27 +488,27 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
 
         // Specular 1 component
         std::string d1_lut_value = "1.0";
-        if (config.lighting.lut_d1.enable && Pica::Regs::IsLightingSamplerSupported(config.lighting.config, Pica::Regs::LightingSampler::Distribution1)) {
+        if (lighting.lut_d1.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::Distribution1)) {
             // Lookup specular "distribution 1" LUT value
-            std::string index = GetLutIndex(light_config.num, config.lighting.lut_d1.type, config.lighting.lut_d1.abs_input);
-            d1_lut_value = "(" + std::to_string(config.lighting.lut_d1.scale) + " * " + GetLutValue(Regs::LightingSampler::Distribution1, index) + ")";
+            std::string index = GetLutIndex(light_config.num, lighting.lut_d1.type, lighting.lut_d1.abs_input);
+            d1_lut_value = "(" + std::to_string(lighting.lut_d1.scale) + " * " + GetLutValue(Regs::LightingSampler::Distribution1, index) + ")";
         }
         std::string specular_1 = "(" + d1_lut_value + " * refl_value * " + light_src + ".specular_1)";
 
         // Fresnel
-        if (config.lighting.lut_fr.enable && Pica::Regs::IsLightingSamplerSupported(config.lighting.config, Pica::Regs::LightingSampler::Fresnel)) {
+        if (lighting.lut_fr.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::Fresnel)) {
             // Lookup fresnel LUT value
-            std::string index = GetLutIndex(light_config.num, config.lighting.lut_fr.type, config.lighting.lut_fr.abs_input);
-            std::string value = "(" + std::to_string(config.lighting.lut_fr.scale) + " * " + GetLutValue(Regs::LightingSampler::Fresnel, index) + ")";
+            std::string index = GetLutIndex(light_config.num, lighting.lut_fr.type, lighting.lut_fr.abs_input);
+            std::string value = "(" + std::to_string(lighting.lut_fr.scale) + " * " + GetLutValue(Regs::LightingSampler::Fresnel, index) + ")";
 
             // Enabled for difffuse lighting alpha component
-            if (config.lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::PrimaryAlpha ||
-                config.lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::Both)
+            if (lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::PrimaryAlpha ||
+                lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::Both)
                 out += "diffuse_sum.a  *= " + value + ";\n";
 
             // Enabled for the specular lighting alpha component
-            if (config.lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::SecondaryAlpha ||
-                config.lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::Both)
+            if (lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::SecondaryAlpha ||
+                lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::Both)
                 out += "specular_sum.a *= " + value + ";\n";
         }
 
@@ -510,6 +526,8 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
 }
 
 std::string GenerateFragmentShader(const PicaShaderConfig& config) {
+    const auto& state = config.state;
+
     std::string out = R"(
 #version 330 core
 #define NUM_TEV_STAGES 6
@@ -519,6 +537,7 @@ std::string GenerateFragmentShader(const PicaShaderConfig& config) {
 
 in vec4 primary_color;
 in vec2 texcoord[3];
+in float texcoord0_w;
 in vec4 normquat;
 in vec3 view;
 
@@ -536,6 +555,7 @@ layout (std140) uniform shader_data {
     vec4 const_color[NUM_TEV_STAGES];
     vec4 tev_combiner_buffer_color;
     int alphatest_ref;
+    float depth_scale;
     float depth_offset;
     vec3 lighting_global_ambient;
     LightSrc light_src[NUM_LIGHTS];
@@ -555,29 +575,37 @@ vec4 secondary_fragment_color = vec4(0.0);
 )";
 
     // Do not do any sort of processing if it's obvious we're not going to pass the alpha test
-    if (config.alpha_test_func == Regs::CompareFunc::Never) {
+    if (state.alpha_test_func == Regs::CompareFunc::Never) {
         out += "discard; }";
         return out;
     }
 
-    if (config.lighting.enable)
+    if (state.lighting.enable)
         WriteLighting(out, config);
 
     out += "vec4 combiner_buffer = vec4(0.0);\n";
     out += "vec4 next_combiner_buffer = tev_combiner_buffer_color;\n";
     out += "vec4 last_tex_env_out = vec4(0.0);\n";
 
-    for (size_t index = 0; index < config.tev_stages.size(); ++index)
+    for (size_t index = 0; index < state.tev_stages.size(); ++index)
         WriteTevStage(out, config, (unsigned)index);
 
-    if (config.alpha_test_func != Regs::CompareFunc::Always) {
+    if (state.alpha_test_func != Regs::CompareFunc::Always) {
         out += "if (";
-        AppendAlphaTestCondition(out, config.alpha_test_func);
+        AppendAlphaTestCondition(out, state.alpha_test_func);
         out += ") discard;\n";
     }
 
     out += "color = last_tex_env_out;\n";
-    out += "gl_FragDepth = gl_FragCoord.z + depth_offset;\n}";
+
+    out += "float z_over_w = 1.0 - gl_FragCoord.z * 2.0;\n";
+    out += "float depth = z_over_w * depth_scale + depth_offset;\n";
+    if (state.depthmap_enable == Pica::Regs::DepthBuffering::WBuffering) {
+        out += "depth /= gl_FragCoord.w;\n";
+    }
+    out += "gl_FragDepth = depth;\n";
+
+    out += "}";
 
     return out;
 }
@@ -585,17 +613,19 @@ vec4 secondary_fragment_color = vec4(0.0);
 std::string GenerateVertexShader() {
     std::string out = "#version 330 core\n";
 
-    out += "layout(location = " + std::to_string((int)ATTRIBUTE_POSITION)  + ") in vec4 vert_position;\n";
-    out += "layout(location = " + std::to_string((int)ATTRIBUTE_COLOR)     + ") in vec4 vert_color;\n";
-    out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD0) + ") in vec2 vert_texcoord0;\n";
-    out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD1) + ") in vec2 vert_texcoord1;\n";
-    out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD2) + ") in vec2 vert_texcoord2;\n";
-    out += "layout(location = " + std::to_string((int)ATTRIBUTE_NORMQUAT)  + ") in vec4 vert_normquat;\n";
-    out += "layout(location = " + std::to_string((int)ATTRIBUTE_VIEW)      + ") in vec3 vert_view;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_POSITION)    + ") in vec4 vert_position;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_COLOR)       + ") in vec4 vert_color;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD0)   + ") in vec2 vert_texcoord0;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD1)   + ") in vec2 vert_texcoord1;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD2)   + ") in vec2 vert_texcoord2;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD0_W) + ") in float vert_texcoord0_w;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_NORMQUAT)    + ") in vec4 vert_normquat;\n";
+    out += "layout(location = " + std::to_string((int)ATTRIBUTE_VIEW)        + ") in vec3 vert_view;\n";
 
     out += R"(
 out vec4 primary_color;
 out vec2 texcoord[3];
+out float texcoord0_w;
 out vec4 normquat;
 out vec3 view;
 
@@ -604,6 +634,7 @@ void main() {
     texcoord[0] = vert_texcoord0;
     texcoord[1] = vert_texcoord1;
     texcoord[2] = vert_texcoord2;
+    texcoord0_w = vert_texcoord0_w;
     normquat = vert_normquat;
     view = vert_view;
     gl_Position = vec4(vert_position.x, vert_position.y, -vert_position.z, vert_position.w);
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.h b/src/video_core/renderer_opengl/gl_shader_gen.h
index 3eb07d57a..bef3249cf 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.h
+++ b/src/video_core/renderer_opengl/gl_shader_gen.h
@@ -6,7 +6,7 @@
 
 #include <string>
 
-struct PicaShaderConfig;
+union PicaShaderConfig;
 
 namespace GLShader {
 
diff --git a/src/video_core/renderer_opengl/gl_shader_util.h b/src/video_core/renderer_opengl/gl_shader_util.h
index 097242f6f..f59912f79 100644
--- a/src/video_core/renderer_opengl/gl_shader_util.h
+++ b/src/video_core/renderer_opengl/gl_shader_util.h
@@ -14,6 +14,7 @@ enum Attributes {
     ATTRIBUTE_TEXCOORD0,
     ATTRIBUTE_TEXCOORD1,
     ATTRIBUTE_TEXCOORD2,
+    ATTRIBUTE_TEXCOORD0_W,
     ATTRIBUTE_NORMQUAT,
     ATTRIBUTE_VIEW,
 };
diff --git a/src/video_core/shader/shader.cpp b/src/video_core/shader/shader.cpp
index 65dcc9156..e93a9d92a 100644
--- a/src/video_core/shader/shader.cpp
+++ b/src/video_core/shader/shader.cpp
@@ -35,7 +35,13 @@ static std::unordered_map<u64, std::unique_ptr<JitShader>> shader_map;
 static const JitShader* jit_shader;
 #endif // ARCHITECTURE_x86_64
 
-void Setup() {
+void ClearCache() {
+#ifdef ARCHITECTURE_x86_64
+    shader_map.clear();
+#endif // ARCHITECTURE_x86_64
+}
+
+void ShaderSetup::Setup() {
 #ifdef ARCHITECTURE_x86_64
     if (VideoCore::g_shader_jit_enabled) {
         u64 cache_key = (Common::ComputeHash64(&g_state.vs.program_code, sizeof(g_state.vs.program_code)) ^
@@ -54,20 +60,13 @@ void Setup() {
 #endif // ARCHITECTURE_x86_64
 }
 
-void Shutdown() {
-#ifdef ARCHITECTURE_x86_64
-    shader_map.clear();
-#endif // ARCHITECTURE_x86_64
-}
-
-MICROPROFILE_DEFINE(GPU_VertexShader, "GPU", "Vertex Shader", MP_RGB(50, 50, 240));
+MICROPROFILE_DEFINE(GPU_Shader, "GPU", "Shader", MP_RGB(50, 50, 240));
 
-OutputVertex Run(UnitState<false>& state, const InputVertex& input, int num_attributes) {
+OutputVertex ShaderSetup::Run(UnitState<false>& state, const InputVertex& input, int num_attributes) {
     auto& config = g_state.regs.vs;
 
-    MICROPROFILE_SCOPE(GPU_VertexShader);
+    MICROPROFILE_SCOPE(GPU_Shader);
 
-    state.program_counter = config.main_offset;
     state.debug.max_offset = 0;
     state.debug.max_opdesc_id = 0;
 
@@ -140,10 +139,9 @@ OutputVertex Run(UnitState<false>& state, const InputVertex& input, int num_attr
     return ret;
 }
 
-DebugData<true> ProduceDebugInfo(const InputVertex& input, int num_attributes, const Regs::ShaderConfig& config, const ShaderSetup& setup) {
+DebugData<true> ShaderSetup::ProduceDebugInfo(const InputVertex& input, int num_attributes, const Regs::ShaderConfig& config, const ShaderSetup& setup) {
     UnitState<true> state;
 
-    state.program_counter = config.main_offset;
     state.debug.max_offset = 0;
     state.debug.max_opdesc_id = 0;
 
diff --git a/src/video_core/shader/shader.h b/src/video_core/shader/shader.h
index 56b83bfeb..983e4a967 100644
--- a/src/video_core/shader/shader.h
+++ b/src/video_core/shader/shader.h
@@ -43,7 +43,8 @@ struct OutputVertex {
     Math::Vec4<float24> color;
     Math::Vec2<float24> tc0;
     Math::Vec2<float24> tc1;
-    INSERT_PADDING_WORDS(2);
+    float24 tc0_w;
+    INSERT_PADDING_WORDS(1);
     Math::Vec3<float24> view;
     INSERT_PADDING_WORDS(1);
     Math::Vec2<float24> tc2;
@@ -83,23 +84,6 @@ struct OutputVertex {
 static_assert(std::is_pod<OutputVertex>::value, "Structure is not POD");
 static_assert(sizeof(OutputVertex) == 32 * sizeof(float), "OutputVertex has invalid size");
 
-/// Vertex shader memory
-struct ShaderSetup {
-    struct {
-        // The float uniforms are accessed by the shader JIT using SSE instructions, and are
-        // therefore required to be 16-byte aligned.
-        alignas(16) Math::Vec4<float24> f[96];
-
-        std::array<bool, 16> b;
-        std::array<Math::Vec4<u8>, 4> i;
-    } uniforms;
-
-    Math::Vec4<float24> default_attributes[16];
-
-    std::array<u32, 1024> program_code;
-    std::array<u32, 1024> swizzle_data;
-};
-
 // Helper structure used to keep track of data useful for inspection of shader emulation
 template<bool full_debugging>
 struct DebugData;
@@ -288,29 +272,12 @@ struct UnitState {
     } registers;
     static_assert(std::is_pod<Registers>::value, "Structure is not POD");
 
-    u32 program_counter;
     bool conditional_code[2];
 
     // Two Address registers and one loop counter
     // TODO: How many bits do these actually have?
     s32 address_registers[3];
 
-    enum {
-        INVALID_ADDRESS = 0xFFFFFFFF
-    };
-
-    struct CallStackElement {
-        u32 final_address;  // Address upon which we jump to return_address
-        u32 return_address; // Where to jump when leaving scope
-        u8 repeat_counter;  // How often to repeat until this call stack element is removed
-        u8 loop_increment;  // Which value to add to the loop counter after an iteration
-                            // TODO: Should this be a signed value? Does it even matter?
-        u32 loop_address;   // The address where we'll return to after each loop iteration
-    };
-
-    // TODO: Is there a maximal size for this?
-    boost::container::static_vector<CallStackElement, 16> call_stack;
-
     DebugData<Debug> debug;
 
     static size_t InputOffset(const SourceRegister& reg) {
@@ -342,33 +309,49 @@ struct UnitState {
     }
 };
 
-/**
- * Performs any shader unit setup that only needs to happen once per shader (as opposed to once per
- * vertex, which would happen within the `Run` function).
- */
-void Setup();
+/// Clears the shader cache
+void ClearCache();
 
-/// Performs any cleanup when the emulator is shutdown
-void Shutdown();
+struct ShaderSetup {
 
-/**
- * Runs the currently setup shader
- * @param state Shader unit state, must be setup per shader and per shader unit
- * @param input Input vertex into the shader
- * @param num_attributes The number of vertex shader attributes
- * @return The output vertex, after having been processed by the vertex shader
- */
-OutputVertex Run(UnitState<false>& state, const InputVertex& input, int num_attributes);
+    struct {
+        // The float uniforms are accessed by the shader JIT using SSE instructions, and are
+        // therefore required to be 16-byte aligned.
+        alignas(16) Math::Vec4<float24> f[96];
 
-/**
- * Produce debug information based on the given shader and input vertex
- * @param input Input vertex into the shader
- * @param num_attributes The number of vertex shader attributes
- * @param config Configuration object for the shader pipeline
- * @param setup Setup object for the shader pipeline
- * @return Debug information for this shader with regards to the given vertex
- */
-DebugData<true> ProduceDebugInfo(const InputVertex& input, int num_attributes, const Regs::ShaderConfig& config, const ShaderSetup& setup);
+        std::array<bool, 16> b;
+        std::array<Math::Vec4<u8>, 4> i;
+    } uniforms;
+
+    std::array<u32, 1024> program_code;
+    std::array<u32, 1024> swizzle_data;
+
+    /**
+     * Performs any shader unit setup that only needs to happen once per shader (as opposed to once per
+     * vertex, which would happen within the `Run` function).
+     */
+    void Setup();
+
+    /**
+     * Runs the currently setup shader
+     * @param state Shader unit state, must be setup per shader and per shader unit
+     * @param input Input vertex into the shader
+     * @param num_attributes The number of vertex shader attributes
+     * @return The output vertex, after having been processed by the vertex shader
+     */
+    OutputVertex Run(UnitState<false>& state, const InputVertex& input, int num_attributes);
+
+    /**
+     * Produce debug information based on the given shader and input vertex
+     * @param input Input vertex into the shader
+     * @param num_attributes The number of vertex shader attributes
+     * @param config Configuration object for the shader pipeline
+     * @param setup Setup object for the shader pipeline
+     * @return Debug information for this shader with regards to the given vertex
+     */
+    DebugData<true> ProduceDebugInfo(const InputVertex& input, int num_attributes, const Regs::ShaderConfig& config, const ShaderSetup& setup);
+
+};
 
 } // namespace Shader
 
diff --git a/src/video_core/shader/shader_interpreter.cpp b/src/video_core/shader/shader_interpreter.cpp
index 7710f7fbc..3a827d11f 100644
--- a/src/video_core/shader/shader_interpreter.cpp
+++ b/src/video_core/shader/shader_interpreter.cpp
@@ -29,8 +29,24 @@ namespace Pica {
 
 namespace Shader {
 
+constexpr u32 INVALID_ADDRESS = 0xFFFFFFFF;
+
+struct CallStackElement {
+    u32 final_address;  // Address upon which we jump to return_address
+    u32 return_address; // Where to jump when leaving scope
+    u8 repeat_counter;  // How often to repeat until this call stack element is removed
+    u8 loop_increment;  // Which value to add to the loop counter after an iteration
+                        // TODO: Should this be a signed value? Does it even matter?
+    u32 loop_address;   // The address where we'll return to after each loop iteration
+};
+
 template<bool Debug>
 void RunInterpreter(UnitState<Debug>& state) {
+    // TODO: Is there a maximal size for this?
+    boost::container::static_vector<CallStackElement, 16> call_stack;
+
+    u32 program_counter = g_state.regs.vs.main_offset;
+
     const auto& uniforms = g_state.vs.uniforms;
     const auto& swizzle_data = g_state.vs.swizzle_data;
     const auto& program_code = g_state.vs.program_code;
@@ -41,16 +57,16 @@ void RunInterpreter(UnitState<Debug>& state) {
     unsigned iteration = 0;
     bool exit_loop = false;
     while (!exit_loop) {
-        if (!state.call_stack.empty()) {
-            auto& top = state.call_stack.back();
-            if (state.program_counter == top.final_address) {
+        if (!call_stack.empty()) {
+            auto& top = call_stack.back();
+            if (program_counter == top.final_address) {
                 state.address_registers[2] += top.loop_increment;
 
                 if (top.repeat_counter-- == 0) {
-                    state.program_counter = top.return_address;
-                    state.call_stack.pop_back();
+                    program_counter = top.return_address;
+                    call_stack.pop_back();
                 } else {
-                    state.program_counter = top.loop_address;
+                    program_counter = top.loop_address;
                 }
 
                 // TODO: Is "trying again" accurate to hardware?
@@ -58,20 +74,20 @@ void RunInterpreter(UnitState<Debug>& state) {
             }
         }
 
-        const Instruction instr = { program_code[state.program_counter] };
+        const Instruction instr = { program_code[program_counter] };
         const SwizzlePattern swizzle = { swizzle_data[instr.common.operand_desc_id] };
 
-        static auto call = [](UnitState<Debug>& state, u32 offset, u32 num_instructions,
+        static auto call = [&program_counter, &call_stack](UnitState<Debug>& state, u32 offset, u32 num_instructions,
                               u32 return_offset, u8 repeat_count, u8 loop_increment) {
-            state.program_counter = offset - 1; // -1 to make sure when incrementing the PC we end up at the correct offset
-            ASSERT(state.call_stack.size() < state.call_stack.capacity());
-            state.call_stack.push_back({ offset + num_instructions, return_offset, repeat_count, loop_increment, offset });
+            program_counter = offset - 1; // -1 to make sure when incrementing the PC we end up at the correct offset
+            ASSERT(call_stack.size() < call_stack.capacity());
+            call_stack.push_back({ offset + num_instructions, return_offset, repeat_count, loop_increment, offset });
         };
-        Record<DebugDataRecord::CUR_INSTR>(state.debug, iteration, state.program_counter);
+        Record<DebugDataRecord::CUR_INSTR>(state.debug, iteration, program_counter);
         if (iteration > 0)
-            Record<DebugDataRecord::NEXT_INSTR>(state.debug, iteration - 1, state.program_counter);
+            Record<DebugDataRecord::NEXT_INSTR>(state.debug, iteration - 1, program_counter);
 
-        state.debug.max_offset = std::max<u32>(state.debug.max_offset, 1 + state.program_counter);
+        state.debug.max_offset = std::max<u32>(state.debug.max_offset, 1 + program_counter);
 
         auto LookupSourceRegister = [&](const SourceRegister& source_reg) -> const float24* {
             switch (source_reg.GetRegisterType()) {
@@ -519,7 +535,7 @@ void RunInterpreter(UnitState<Debug>& state) {
             case OpCode::Id::JMPC:
                 Record<DebugDataRecord::COND_CMP_IN>(state.debug, iteration, state.conditional_code);
                 if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) {
-                    state.program_counter = instr.flow_control.dest_offset - 1;
+                    program_counter = instr.flow_control.dest_offset - 1;
                 }
                 break;
 
@@ -527,7 +543,7 @@ void RunInterpreter(UnitState<Debug>& state) {
                 Record<DebugDataRecord::COND_BOOL_IN>(state.debug, iteration, uniforms.b[instr.flow_control.bool_uniform_id]);
 
                 if (uniforms.b[instr.flow_control.bool_uniform_id] == !(instr.flow_control.num_instructions & 1)) {
-                    state.program_counter = instr.flow_control.dest_offset - 1;
+                    program_counter = instr.flow_control.dest_offset - 1;
                 }
                 break;
 
@@ -535,7 +551,7 @@ void RunInterpreter(UnitState<Debug>& state) {
                 call(state,
                      instr.flow_control.dest_offset,
                      instr.flow_control.num_instructions,
-                     state.program_counter + 1, 0, 0);
+                     program_counter + 1, 0, 0);
                 break;
 
             case OpCode::Id::CALLU:
@@ -544,7 +560,7 @@ void RunInterpreter(UnitState<Debug>& state) {
                     call(state,
                         instr.flow_control.dest_offset,
                         instr.flow_control.num_instructions,
-                        state.program_counter + 1, 0, 0);
+                        program_counter + 1, 0, 0);
                 }
                 break;
 
@@ -554,7 +570,7 @@ void RunInterpreter(UnitState<Debug>& state) {
                     call(state,
                         instr.flow_control.dest_offset,
                         instr.flow_control.num_instructions,
-                        state.program_counter + 1, 0, 0);
+                        program_counter + 1, 0, 0);
                 }
                 break;
 
@@ -565,8 +581,8 @@ void RunInterpreter(UnitState<Debug>& state) {
                 Record<DebugDataRecord::COND_BOOL_IN>(state.debug, iteration, uniforms.b[instr.flow_control.bool_uniform_id]);
                 if (uniforms.b[instr.flow_control.bool_uniform_id]) {
                     call(state,
-                         state.program_counter + 1,
-                         instr.flow_control.dest_offset - state.program_counter - 1,
+                         program_counter + 1,
+                         instr.flow_control.dest_offset - program_counter - 1,
                          instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0);
                 } else {
                     call(state,
@@ -584,8 +600,8 @@ void RunInterpreter(UnitState<Debug>& state) {
                 Record<DebugDataRecord::COND_CMP_IN>(state.debug, iteration, state.conditional_code);
                 if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) {
                     call(state,
-                         state.program_counter + 1,
-                         instr.flow_control.dest_offset - state.program_counter - 1,
+                         program_counter + 1,
+                         instr.flow_control.dest_offset - program_counter - 1,
                          instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0);
                 } else {
                     call(state,
@@ -607,8 +623,8 @@ void RunInterpreter(UnitState<Debug>& state) {
 
                 Record<DebugDataRecord::LOOP_INT_IN>(state.debug, iteration, loop_param);
                 call(state,
-                     state.program_counter + 1,
-                     instr.flow_control.dest_offset - state.program_counter + 1,
+                     program_counter + 1,
+                     instr.flow_control.dest_offset - program_counter + 1,
                      instr.flow_control.dest_offset + 1,
                      loop_param.x,
                      loop_param.z);
@@ -625,7 +641,7 @@ void RunInterpreter(UnitState<Debug>& state) {
         }
         }
 
-        ++state.program_counter;
+        ++program_counter;
         ++iteration;
     }
 }
diff --git a/src/video_core/vertex_loader.cpp b/src/video_core/vertex_loader.cpp
index 21ae52949..83896814f 100644
--- a/src/video_core/vertex_loader.cpp
+++ b/src/video_core/vertex_loader.cpp
@@ -124,7 +124,7 @@ void VertexLoader::LoadVertex(u32 base_address, int index, int vertex, Shader::I
                 input.attr[i][0].ToFloat32(), input.attr[i][1].ToFloat32(), input.attr[i][2].ToFloat32(), input.attr[i][3].ToFloat32());
         } else if (vertex_attribute_is_default[i]) {
             // Load the default attribute if we're configured to do so
-            input.attr[i] = g_state.vs.default_attributes[i];
+            input.attr[i] = g_state.vs_default_attributes[i];
             LOG_TRACE(HW_GPU, "Loaded default attribute %x for vertex %x (index %x): (%f, %f, %f, %f)",
                 i, vertex, index,
                 input.attr[i][0].ToFloat32(), input.attr[i][1].ToFloat32(),