20 files changed, 913 insertions, 446 deletions

diff --git a/src/core/memory.cpp b/src/core/memory.cpp
index d6469dd3d..291bf066f 100644
--- a/src/core/memory.cpp
+++ b/src/core/memory.cpp
@@ -15,6 +15,7 @@
 #include "core/core.h"
 #include "core/hle/kernel/memory.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/lock.h"
 #include "core/memory.h"
 #include "core/memory_setup.h"
 #include "video_core/renderer_base.h"
@@ -115,91 +116,120 @@ static std::set<MemoryHookPointer> GetSpecialHandlers(VAddr vaddr, u64 size) {
     return GetSpecialHandlers(page_table, vaddr, size);
 }
 
-template <typename T>
-boost::optional<T> ReadSpecial(VAddr addr);
+/**
+ * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
+ * using a VMA from the current process
+ */
+static u8* GetPointerFromVMA(const Kernel::Process& process, VAddr vaddr) {
+    u8* direct_pointer = nullptr;
+
+    auto& vm_manager = process.vm_manager;
+
+    auto it = vm_manager.FindVMA(vaddr);
+    ASSERT(it != vm_manager.vma_map.end());
+
+    auto& vma = it->second;
+    switch (vma.type) {
+    case Kernel::VMAType::AllocatedMemoryBlock:
+        direct_pointer = vma.backing_block->data() + vma.offset;
+        break;
+    case Kernel::VMAType::BackingMemory:
+        direct_pointer = vma.backing_memory;
+        break;
+    case Kernel::VMAType::Free:
+        return nullptr;
+    default:
+        UNREACHABLE();
+    }
+
+    return direct_pointer + (vaddr - vma.base);
+}
+
+/**
+ * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
+ * using a VMA from the current process.
+ */
+static u8* GetPointerFromVMA(VAddr vaddr) {
+    return GetPointerFromVMA(*Core::CurrentProcess(), vaddr);
+}
 
 template <typename T>
 T Read(const VAddr vaddr) {
-    if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES) {
-        LOG_ERROR(HW_Memory, "Read%lu after page table @ 0x%016" PRIX64, sizeof(T) * 8, vaddr);
-        return 0;
+    const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+    if (page_pointer) {
+        // NOTE: Avoid adding any extra logic to this fast-path block
+        T value;
+        std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T));
+        return value;
     }
 
-    const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+    // The memory access might do an MMIO or cached access, so we have to lock the HLE kernel state
+    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+
+    PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
     switch (type) {
     case PageType::Unmapped:
-        LOG_ERROR(HW_Memory, "unmapped Read%zu @ 0x%016" PRIX64, sizeof(T) * 8, vaddr);
+        LOG_ERROR(HW_Memory, "unmapped Read%lu @ 0x%08X", sizeof(T) * 8, vaddr);
         return 0;
-    case PageType::Special: {
-        if (auto result = ReadSpecial<T>(vaddr))
-            return *result;
-        [[fallthrough]];
-    }
-    case PageType::Memory: {
-        const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
-        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr);
+    case PageType::Memory:
+        ASSERT_MSG(false, "Mapped memory page without a pointer @ %08X", vaddr);
+        break;
+    case PageType::RasterizerCachedMemory: {
+        RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Flush);
 
         T value;
-        std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T));
+        std::memcpy(&value, GetPointerFromVMA(vaddr), sizeof(T));
         return value;
     }
+    default:
+        UNREACHABLE();
     }
-    UNREACHABLE();
-    return 0;
 }
 
 template <typename T>
-bool WriteSpecial(VAddr addr, const T data);
-
-template <typename T>
 void Write(const VAddr vaddr, const T data) {
-    if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES) {
-        LOG_ERROR(HW_Memory, "Write%lu after page table 0x%08X @ 0x%016" PRIX64, sizeof(data) * 8,
-                  (u32)data, vaddr);
+    u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+    if (page_pointer) {
+        // NOTE: Avoid adding any extra logic to this fast-path block
+        std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T));
         return;
     }
 
-    const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+    // The memory access might do an MMIO or cached access, so we have to lock the HLE kernel state
+    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+
+    PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
     switch (type) {
     case PageType::Unmapped:
-        LOG_ERROR(HW_Memory, "unmapped Write%zu 0x%08X @ 0x%016" PRIX64, sizeof(data) * 8,
-                  static_cast<u32>(data), vaddr);
-        return;
-    case PageType::Special: {
-        if (WriteSpecial<T>(vaddr, data))
-            return;
-        [[fallthrough]];
-    }
-    case PageType::Memory: {
-        u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
-        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr);
-        std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T));
+        LOG_ERROR(HW_Memory, "unmapped Write%lu 0x%08X @ 0x%08X", sizeof(data) * 8, (u32)data,
+                  vaddr);
         return;
+    case PageType::Memory:
+        ASSERT_MSG(false, "Mapped memory page without a pointer @ %08X", vaddr);
+        break;
+    case PageType::RasterizerCachedMemory: {
+        RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Invalidate);
+        std::memcpy(GetPointerFromVMA(vaddr), &data, sizeof(T));
+        break;
     }
+    default:
+        UNREACHABLE();
     }
-    UNREACHABLE();
 }
 
 bool IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr) {
     auto& page_table = process.vm_manager.page_table;
 
-    if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES)
-        return false;
+    const u8* page_pointer = page_table.pointers[vaddr >> PAGE_BITS];
+    if (page_pointer)
+        return true;
 
-    const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
-    switch (type) {
-    case PageType::Unmapped:
-        return false;
-    case PageType::Memory:
+    if (page_table.attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory)
         return true;
-    case PageType::Special: {
-        for (auto handler : GetSpecialHandlers(page_table, vaddr, 1))
-            if (auto result = handler->IsValidAddress(vaddr))
-                return *result;
-        return current_page_table->pointers[vaddr >> PAGE_BITS] != nullptr;
-    }
-    }
-    UNREACHABLE();
+
+    if (page_table.attributes[vaddr >> PAGE_BITS] != PageType::Special)
+        return false;
+
     return false;
 }
 
@@ -217,7 +247,11 @@ u8* GetPointer(const VAddr vaddr) {
         return page_pointer + (vaddr & PAGE_MASK);
     }
 
-    LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%016" PRIx64, vaddr);
+    if (current_page_table->attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory) {
+        return GetPointerFromVMA(vaddr);
+    }
+
+    LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%08x", vaddr);
     return nullptr;
 }
 
@@ -291,6 +325,58 @@ u8* GetPhysicalPointer(PAddr address) {
     return target_pointer;
 }
 
+void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached) {
+    if (start == 0) {
+        return;
+    }
+
+    u64 num_pages = ((start + size - 1) >> PAGE_BITS) - (start >> PAGE_BITS) + 1;
+    VAddr vaddr = start;
+
+    for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
+        PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
+
+        if (cached) {
+            // Switch page type to cached if now cached
+            switch (page_type) {
+            case PageType::Unmapped:
+                // It is not necessary for a process to have this region mapped into its address
+                // space, for example, a system module need not have a VRAM mapping.
+                break;
+            case PageType::Memory:
+                page_type = PageType::RasterizerCachedMemory;
+                current_page_table->pointers[vaddr >> PAGE_BITS] = nullptr;
+                break;
+            default:
+                UNREACHABLE();
+            }
+        } else {
+            // Switch page type to uncached if now uncached
+            switch (page_type) {
+            case PageType::Unmapped:
+                // It is not necessary for a process to have this region mapped into its address
+                // space, for example, a system module need not have a VRAM mapping.
+                break;
+            case PageType::RasterizerCachedMemory: {
+                u8* pointer = GetPointerFromVMA(vaddr & ~PAGE_MASK);
+                if (pointer == nullptr) {
+                    // It's possible that this function has been called while updating the pagetable
+                    // after unmapping a VMA. In that case the underlying VMA will no longer exist,
+                    // and we should just leave the pagetable entry blank.
+                    page_type = PageType::Unmapped;
+                } else {
+                    page_type = PageType::Memory;
+                    current_page_table->pointers[vaddr >> PAGE_BITS] = pointer;
+                }
+                break;
+            }
+            default:
+                UNREACHABLE();
+            }
+        }
+    }
+}
+
 void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode) {
     // Since pages are unmapped on shutdown after video core is shutdown, the renderer may be
     // null here
@@ -344,17 +430,6 @@ u64 Read64(const VAddr addr) {
     return Read<u64_le>(addr);
 }
 
-static bool ReadSpecialBlock(const Kernel::Process& process, const VAddr src_addr,
-                             void* dest_buffer, const size_t size) {
-    auto& page_table = process.vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, src_addr, size)) {
-        if (handler->ReadBlock(src_addr, dest_buffer, size)) {
-            return true;
-        }
-    }
-    return false;
-}
-
 void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_buffer,
                const size_t size) {
     auto& page_table = process.vm_manager.page_table;
@@ -364,21 +439,16 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_
     size_t page_offset = src_addr & PAGE_MASK;
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
         switch (page_table.attributes[page_index]) {
-        case PageType::Unmapped:
-            LOG_ERROR(HW_Memory,
-                      "unmapped ReadBlock @ 0x%016" PRIX64 " (start address = 0x%" PRIx64
-                      ", size = %zu)",
+        case PageType::Unmapped: {
+            LOG_ERROR(HW_Memory, "unmapped ReadBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, src_addr, size);
             std::memset(dest_buffer, 0, copy_amount);
             break;
-        case PageType::Special: {
-            if (ReadSpecialBlock(process, current_vaddr, dest_buffer, copy_amount))
-                break;
-            [[fallthrough]];
         }
         case PageType::Memory: {
             DEBUG_ASSERT(page_table.pointers[page_index]);
@@ -387,6 +457,12 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_
             std::memcpy(dest_buffer, src_ptr, copy_amount);
             break;
         }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Flush);
+            std::memcpy(dest_buffer, GetPointerFromVMA(process, current_vaddr), copy_amount);
+            break;
+        }
         default:
             UNREACHABLE();
         }
@@ -418,17 +494,6 @@ void Write64(const VAddr addr, const u64 data) {
     Write<u64_le>(addr, data);
 }
 
-static bool WriteSpecialBlock(const Kernel::Process& process, const VAddr dest_addr,
-                              const void* src_buffer, const size_t size) {
-    auto& page_table = process.vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, dest_addr, size)) {
-        if (handler->WriteBlock(dest_addr, src_buffer, size)) {
-            return true;
-        }
-    }
-    return false;
-}
-
 void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const void* src_buffer,
                 const size_t size) {
     auto& page_table = process.vm_manager.page_table;
@@ -437,20 +502,17 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi
     size_t page_offset = dest_addr & PAGE_MASK;
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
         switch (page_table.attributes[page_index]) {
-        case PageType::Unmapped:
+        case PageType::Unmapped: {
             LOG_ERROR(HW_Memory,
-                      "unmapped WriteBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
-                      ", size = %zu)",
+                      "unmapped WriteBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, dest_addr, size);
             break;
-        case PageType::Special:
-            if (WriteSpecialBlock(process, current_vaddr, src_buffer, copy_amount))
-                break;
-            [[fallthrough]];
+        }
         case PageType::Memory: {
             DEBUG_ASSERT(page_table.pointers[page_index]);
 
@@ -458,6 +520,12 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi
             std::memcpy(dest_ptr, src_buffer, copy_amount);
             break;
         }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Invalidate);
+            std::memcpy(GetPointerFromVMA(process, current_vaddr), src_buffer, copy_amount);
+            break;
+        }
         default:
             UNREACHABLE();
         }
@@ -473,9 +541,8 @@ void WriteBlock(const VAddr dest_addr, const void* src_buffer, const size_t size
     WriteBlock(*Core::CurrentProcess(), dest_addr, src_buffer, size);
 }
 
-void ZeroBlock(const VAddr dest_addr, const size_t size) {
-    const auto& process = *Core::CurrentProcess();
-
+void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const size_t size) {
+    auto& page_table = process.vm_manager.page_table;
     size_t remaining_size = size;
     size_t page_index = dest_addr >> PAGE_BITS;
     size_t page_offset = dest_addr & PAGE_MASK;
@@ -483,27 +550,29 @@ void ZeroBlock(const VAddr dest_addr, const size_t size) {
     static const std::array<u8, PAGE_SIZE> zeros = {};
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
-        switch (current_page_table->attributes[page_index]) {
-        case PageType::Unmapped:
-            LOG_ERROR(HW_Memory,
-                      "unmapped ZeroBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
-                      ", size = %zu)",
+        switch (page_table.attributes[page_index]) {
+        case PageType::Unmapped: {
+            LOG_ERROR(HW_Memory, "unmapped ZeroBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, dest_addr, size);
             break;
-        case PageType::Special:
-            if (WriteSpecialBlock(process, current_vaddr, zeros.data(), copy_amount))
-                break;
-            [[fallthrough]];
+        }
         case PageType::Memory: {
-            DEBUG_ASSERT(current_page_table->pointers[page_index]);
+            DEBUG_ASSERT(page_table.pointers[page_index]);
 
-            u8* dest_ptr = current_page_table->pointers[page_index] + page_offset;
+            u8* dest_ptr = page_table.pointers[page_index] + page_offset;
             std::memset(dest_ptr, 0, copy_amount);
             break;
         }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Invalidate);
+            std::memset(GetPointerFromVMA(process, current_vaddr), 0, copy_amount);
+            break;
+        }
         default:
             UNREACHABLE();
         }
@@ -514,37 +583,34 @@ void ZeroBlock(const VAddr dest_addr, const size_t size) {
     }
 }
 
-void CopyBlock(VAddr dest_addr, VAddr src_addr, const size_t size) {
-    const auto& process = *Core::CurrentProcess();
-
+void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr, const size_t size) {
+    auto& page_table = process.vm_manager.page_table;
     size_t remaining_size = size;
     size_t page_index = src_addr >> PAGE_BITS;
     size_t page_offset = src_addr & PAGE_MASK;
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
-        switch (current_page_table->attributes[page_index]) {
-        case PageType::Unmapped:
-            LOG_ERROR(HW_Memory,
-                      "unmapped CopyBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
-                      ", size = %zu)",
+        switch (page_table.attributes[page_index]) {
+        case PageType::Unmapped: {
+            LOG_ERROR(HW_Memory, "unmapped CopyBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, src_addr, size);
-            ZeroBlock(dest_addr, copy_amount);
+            ZeroBlock(process, dest_addr, copy_amount);
             break;
-        case PageType::Special: {
-            std::vector<u8> buffer(copy_amount);
-            if (ReadSpecialBlock(process, current_vaddr, buffer.data(), buffer.size())) {
-                WriteBlock(dest_addr, buffer.data(), buffer.size());
-                break;
-            }
-            [[fallthrough]];
         }
         case PageType::Memory: {
-            DEBUG_ASSERT(current_page_table->pointers[page_index]);
-            const u8* src_ptr = current_page_table->pointers[page_index] + page_offset;
-            WriteBlock(dest_addr, src_ptr, copy_amount);
+            DEBUG_ASSERT(page_table.pointers[page_index]);
+            const u8* src_ptr = page_table.pointers[page_index] + page_offset;
+            WriteBlock(process, dest_addr, src_ptr, copy_amount);
+            break;
+        }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Flush);
+            WriteBlock(process, dest_addr, GetPointerFromVMA(process, current_vaddr), copy_amount);
             break;
         }
         default:
@@ -559,78 +625,6 @@ void CopyBlock(VAddr dest_addr, VAddr src_addr, const size_t size) {
     }
 }
 
-template <>
-boost::optional<u8> ReadSpecial<u8>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u8)))
-        if (auto result = handler->Read8(addr))
-            return *result;
-    return {};
-}
-
-template <>
-boost::optional<u16> ReadSpecial<u16>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u16)))
-        if (auto result = handler->Read16(addr))
-            return *result;
-    return {};
-}
-
-template <>
-boost::optional<u32> ReadSpecial<u32>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u32)))
-        if (auto result = handler->Read32(addr))
-            return *result;
-    return {};
-}
-
-template <>
-boost::optional<u64> ReadSpecial<u64>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u64)))
-        if (auto result = handler->Read64(addr))
-            return *result;
-    return {};
-}
-
-template <>
-bool WriteSpecial<u8>(VAddr addr, const u8 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u8)))
-        if (handler->Write8(addr, data))
-            return true;
-    return false;
-}
-
-template <>
-bool WriteSpecial<u16>(VAddr addr, const u16 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u16)))
-        if (handler->Write16(addr, data))
-            return true;
-    return false;
-}
-
-template <>
-bool WriteSpecial<u32>(VAddr addr, const u32 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u32)))
-        if (handler->Write32(addr, data))
-            return true;
-    return false;
-}
-
-template <>
-bool WriteSpecial<u64>(VAddr addr, const u64 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u64)))
-        if (handler->Write64(addr, data))
-            return true;
-    return false;
-}
-
 boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) {
     if (addr == 0) {
         return 0;
diff --git a/src/core/memory.h b/src/core/memory.h
index 4b9c482fe..413a7b4e8 100644
--- a/src/core/memory.h
+++ b/src/core/memory.h
@@ -255,6 +255,11 @@ enum class FlushMode {
 };
 
 /**
+ * Mark each page touching the region as cached.
+ */
+void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached);
+
+/**
  * Flushes and invalidates any externally cached rasterizer resources touching the given virtual
  * address region.
  */
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index 3dab81769..841f27d7f 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -31,6 +31,7 @@ add_library(video_core STATIC
     renderer_opengl/gl_state.h
     renderer_opengl/gl_stream_buffer.cpp
     renderer_opengl/gl_stream_buffer.h
+    renderer_opengl/maxwell_to_gl.h
     renderer_opengl/renderer_opengl.cpp
     renderer_opengl/renderer_opengl.h
     textures/decoders.cpp
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 088d4357e..5359d21a2 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -7,8 +7,11 @@
 #include "core/core.h"
 #include "video_core/debug_utils/debug_utils.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_base.h"
 #include "video_core/textures/decoders.h"
 #include "video_core/textures/texture.h"
+#include "video_core/video_core.h"
 
 namespace Tegra {
 namespace Engines {
@@ -174,7 +177,9 @@ void Maxwell3D::ProcessQueryGet() {
 }
 
 void Maxwell3D::DrawArrays() {
-    LOG_WARNING(HW_GPU, "Game requested a DrawArrays, ignoring");
+    LOG_DEBUG(HW_GPU, "called, topology=%d, count=%d", regs.draw.topology.Value(),
+              regs.vertex_buffer.count);
+
     auto debug_context = Core::System::GetInstance().GetGPUDebugContext();
 
     if (debug_context) {
@@ -184,6 +189,8 @@ void Maxwell3D::DrawArrays() {
     if (debug_context) {
         debug_context->OnEvent(Tegra::DebugContext::Event::FinishedPrimitiveBatch, nullptr);
     }
+
+    VideoCore::g_renderer->Rasterizer()->AccelerateDrawBatch(false /*is_indexed*/);
 }
 
 void Maxwell3D::BindTextureInfoBuffer(const std::vector<u32>& parameters) {
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index 8e2d888e7..3066bc606 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -11,6 +11,8 @@
 #include "common/bit_field.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
+#include "common/math_util.h"
+#include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 #include "video_core/textures/texture.h"
 
@@ -59,88 +61,173 @@ public:
             Fragment = 4,
         };
 
-        enum class VertexSize : u32 {
-            Size_32_32_32_32 = 0x01,
-            Size_32_32_32 = 0x02,
-            Size_16_16_16_16 = 0x03,
-            Size_32_32 = 0x04,
-            Size_16_16_16 = 0x05,
-            Size_8_8_8_8 = 0x0a,
-            Size_16_16 = 0x0f,
-            Size_32 = 0x12,
-            Size_8_8_8 = 0x13,
-            Size_8_8 = 0x18,
-            Size_16 = 0x1b,
-            Size_8 = 0x1d,
-            Size_10_10_10_2 = 0x30,
-            Size_11_11_10 = 0x31,
-        };
+        struct VertexAttribute {
+            enum class Size : u32 {
+                Size_32_32_32_32 = 0x01,
+                Size_32_32_32 = 0x02,
+                Size_16_16_16_16 = 0x03,
+                Size_32_32 = 0x04,
+                Size_16_16_16 = 0x05,
+                Size_8_8_8_8 = 0x0a,
+                Size_16_16 = 0x0f,
+                Size_32 = 0x12,
+                Size_8_8_8 = 0x13,
+                Size_8_8 = 0x18,
+                Size_16 = 0x1b,
+                Size_8 = 0x1d,
+                Size_10_10_10_2 = 0x30,
+                Size_11_11_10 = 0x31,
+            };
 
-        static std::string VertexSizeToString(VertexSize vertex_size) {
-            switch (vertex_size) {
-            case VertexSize::Size_32_32_32_32:
-                return "32_32_32_32";
-            case VertexSize::Size_32_32_32:
-                return "32_32_32";
-            case VertexSize::Size_16_16_16_16:
-                return "16_16_16_16";
-            case VertexSize::Size_32_32:
-                return "32_32";
-            case VertexSize::Size_16_16_16:
-                return "16_16_16";
-            case VertexSize::Size_8_8_8_8:
-                return "8_8_8_8";
-            case VertexSize::Size_16_16:
-                return "16_16";
-            case VertexSize::Size_32:
-                return "32";
-            case VertexSize::Size_8_8_8:
-                return "8_8_8";
-            case VertexSize::Size_8_8:
-                return "8_8";
-            case VertexSize::Size_16:
-                return "16";
-            case VertexSize::Size_8:
-                return "8";
-            case VertexSize::Size_10_10_10_2:
-                return "10_10_10_2";
-            case VertexSize::Size_11_11_10:
-                return "11_11_10";
+            enum class Type : u32 {
+                SignedNorm = 1,
+                UnsignedNorm = 2,
+                SignedInt = 3,
+                UnsignedInt = 4,
+                UnsignedScaled = 5,
+                SignedScaled = 6,
+                Float = 7,
+            };
+
+            union {
+                BitField<0, 5, u32> buffer;
+                BitField<6, 1, u32> constant;
+                BitField<7, 14, u32> offset;
+                BitField<21, 6, Size> size;
+                BitField<27, 3, Type> type;
+                BitField<31, 1, u32> bgra;
+            };
+
+            u32 ComponentCount() const {
+                switch (size) {
+                case Size::Size_32_32_32_32:
+                    return 4;
+                case Size::Size_32_32_32:
+                    return 3;
+                case Size::Size_16_16_16_16:
+                    return 4;
+                case Size::Size_32_32:
+                    return 2;
+                case Size::Size_16_16_16:
+                    return 3;
+                case Size::Size_8_8_8_8:
+                    return 4;
+                case Size::Size_16_16:
+                    return 2;
+                case Size::Size_32:
+                    return 1;
+                case Size::Size_8_8_8:
+                    return 3;
+                case Size::Size_8_8:
+                    return 2;
+                case Size::Size_16:
+                    return 1;
+                case Size::Size_8:
+                    return 1;
+                case Size::Size_10_10_10_2:
+                    return 4;
+                case Size::Size_11_11_10:
+                    return 3;
+                default:
+                    UNREACHABLE();
+                }
+            }
+
+            u32 SizeInBytes() const {
+                switch (size) {
+                case Size::Size_32_32_32_32:
+                    return 16;
+                case Size::Size_32_32_32:
+                    return 12;
+                case Size::Size_16_16_16_16:
+                    return 8;
+                case Size::Size_32_32:
+                    return 8;
+                case Size::Size_16_16_16:
+                    return 6;
+                case Size::Size_8_8_8_8:
+                    return 4;
+                case Size::Size_16_16:
+                    return 4;
+                case Size::Size_32:
+                    return 4;
+                case Size::Size_8_8_8:
+                    return 3;
+                case Size::Size_8_8:
+                    return 2;
+                case Size::Size_16:
+                    return 2;
+                case Size::Size_8:
+                    return 1;
+                case Size::Size_10_10_10_2:
+                    return 4;
+                case Size::Size_11_11_10:
+                    return 4;
+                default:
+                    UNREACHABLE();
+                }
             }
-            UNIMPLEMENTED();
-            return {};
-        }
-
-        enum class VertexType : u32 {
-            SignedNorm = 1,
-            UnsignedNorm = 2,
-            SignedInt = 3,
-            UnsignedInt = 4,
-            UnsignedScaled = 5,
-            SignedScaled = 6,
-            Float = 7,
-        };
 
-        static std::string VertexTypeToString(VertexType vertex_type) {
-            switch (vertex_type) {
-            case VertexType::SignedNorm:
-                return "SignedNorm";
-            case VertexType::UnsignedNorm:
-                return "UnsignedNorm";
-            case VertexType::SignedInt:
-                return "SignedInt";
-            case VertexType::UnsignedInt:
-                return "UnsignedInt";
-            case VertexType::UnsignedScaled:
-                return "UnsignedScaled";
-            case VertexType::SignedScaled:
-                return "SignedScaled";
-            case VertexType::Float:
-                return "Float";
+            std::string SizeString() const {
+                switch (size) {
+                case Size::Size_32_32_32_32:
+                    return "32_32_32_32";
+                case Size::Size_32_32_32:
+                    return "32_32_32";
+                case Size::Size_16_16_16_16:
+                    return "16_16_16_16";
+                case Size::Size_32_32:
+                    return "32_32";
+                case Size::Size_16_16_16:
+                    return "16_16_16";
+                case Size::Size_8_8_8_8:
+                    return "8_8_8_8";
+                case Size::Size_16_16:
+                    return "16_16";
+                case Size::Size_32:
+                    return "32";
+                case Size::Size_8_8_8:
+                    return "8_8_8";
+                case Size::Size_8_8:
+                    return "8_8";
+                case Size::Size_16:
+                    return "16";
+                case Size::Size_8:
+                    return "8";
+                case Size::Size_10_10_10_2:
+                    return "10_10_10_2";
+                case Size::Size_11_11_10:
+                    return "11_11_10";
+                }
+                UNREACHABLE();
+                return {};
             }
-            UNIMPLEMENTED();
-            return {};
-        }
+
+            std::string TypeString() const {
+                switch (type) {
+                case Type::SignedNorm:
+                    return "SNORM";
+                case Type::UnsignedNorm:
+                    return "UNORM";
+                case Type::SignedInt:
+                    return "SINT";
+                case Type::UnsignedInt:
+                    return "UINT";
+                case Type::UnsignedScaled:
+                    return "USCALED";
+                case Type::SignedScaled:
+                    return "SSCALED";
+                case Type::Float:
+                    return "FLOAT";
+                }
+                UNREACHABLE();
+                return {};
+            }
+
+            bool IsNormalized() const {
+                return (type == Type::SignedNorm) || (type == Type::UnsignedNorm);
+            }
+        };
 
         enum class PrimitiveTopology : u32 {
             Points = 0x0,
@@ -167,9 +254,9 @@ public:
                 struct {
                     u32 address_high;
                     u32 address_low;
-                    u32 horiz;
-                    u32 vert;
-                    u32 format;
+                    u32 width;
+                    u32 height;
+                    Tegra::RenderTargetFormat format;
                     u32 block_dimensions;
                     u32 array_mode;
                     u32 layer_stride;
@@ -195,6 +282,15 @@ public:
                     };
                     float depth_range_near;
                     float depth_range_far;
+
+                    MathUtil::Rectangle<s32> GetRect() const {
+                        return {
+                            static_cast<s32>(x),          // left
+                            static_cast<s32>(y + height), // top
+                            static_cast<s32>(x + width),  // right
+                            static_cast<s32>(y)           // bottom
+                        };
+                    };
                 } viewport[NumViewports];
 
                 INSERT_PADDING_WORDS(0x1D);
@@ -221,14 +317,7 @@ public:
 
                 INSERT_PADDING_WORDS(0x5B);
 
-                union {
-                    BitField<0, 5, u32> buffer;
-                    BitField<6, 1, u32> constant;
-                    BitField<7, 14, u32> offset;
-                    BitField<21, 6, VertexSize> size;
-                    BitField<27, 3, VertexType> type;
-                    BitField<31, 1, u32> bgra;
-                } vertex_attrib_format[NumVertexAttributes];
+                VertexAttribute vertex_attrib_format[NumVertexAttributes];
 
                 INSERT_PADDING_WORDS(0xF);
 
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index 8183b12e9..71a8661b4 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -13,7 +13,8 @@
 
 namespace Tegra {
 
-enum class RenderTargetFormat {
+enum class RenderTargetFormat : u32 {
+    NONE = 0x0,
     RGBA8_UNORM = 0xD5,
 };
 
diff --git a/src/video_core/rasterizer_interface.h b/src/video_core/rasterizer_interface.h
index a493e1d60..8239f9aad 100644
--- a/src/video_core/rasterizer_interface.h
+++ b/src/video_core/rasterizer_interface.h
@@ -15,8 +15,8 @@ class RasterizerInterface {
 public:
     virtual ~RasterizerInterface() {}
 
-    /// Draw the current batch of triangles
-    virtual void DrawTriangles() = 0;
+    /// Draw the current batch of vertex arrays
+    virtual void DrawArrays() = 0;
 
     /// Notify rasterizer that the specified Maxwell register has been changed
     virtual void NotifyMaxwellRegisterChanged(u32 id) = 0;
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 286491b73..911890f16 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -14,11 +14,16 @@
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "common/vector_math.h"
+#include "core/core.h"
+#include "core/hle/kernel/process.h"
 #include "core/settings.h"
+#include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
+#include "video_core/renderer_opengl/maxwell_to_gl.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
 
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using PixelFormat = SurfaceParams::PixelFormat;
 using SurfaceType = SurfaceParams::SurfaceType;
 
@@ -120,14 +125,14 @@ RasterizerOpenGL::RasterizerOpenGL() {
         glBufferData(GL_UNIFORM_BUFFER, sizeof(VSUniformData), nullptr, GL_STREAM_COPY);
         glBindBufferBase(GL_UNIFORM_BUFFER, 1, vs_uniform_buffer.handle);
     } else {
-        ASSERT_MSG(false, "Unimplemented");
+        UNREACHABLE();
     }
 
     accelerate_draw = AccelDraw::Disabled;
 
     glEnable(GL_BLEND);
 
-    LOG_WARNING(HW_GPU, "Sync fixed function OpenGL state here when ready");
+    LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
 }
 
 RasterizerOpenGL::~RasterizerOpenGL() {
@@ -138,47 +143,235 @@ RasterizerOpenGL::~RasterizerOpenGL() {
     }
 }
 
-static constexpr std::array<GLenum, 4> vs_attrib_types{
-    GL_BYTE,          // VertexAttributeFormat::BYTE
-    GL_UNSIGNED_BYTE, // VertexAttributeFormat::UBYTE
-    GL_SHORT,         // VertexAttributeFormat::SHORT
-    GL_FLOAT          // VertexAttributeFormat::FLOAT
-};
-
 void RasterizerOpenGL::AnalyzeVertexArray(bool is_indexed) {
-    UNIMPLEMENTED();
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+
+    if (is_indexed) {
+        UNREACHABLE();
+    }
+
+    // TODO(bunnei): Add support for 1+ vertex arrays
+    vs_input_size = regs.vertex_buffer.count * regs.vertex_array[0].stride;
 }
 
 void RasterizerOpenGL::SetupVertexArray(u8* array_ptr, GLintptr buffer_offset) {
     MICROPROFILE_SCOPE(OpenGL_VAO);
-    UNIMPLEMENTED();
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
+
+    state.draw.vertex_array = hw_vao.handle;
+    state.draw.vertex_buffer = stream_buffer->GetHandle();
+    state.Apply();
+
+    // TODO(bunnei): Add support for 1+ vertex arrays
+    const auto& vertex_array{regs.vertex_array[0]};
+    ASSERT_MSG(vertex_array.enable, "vertex array 0 is disabled?");
+    ASSERT_MSG(!vertex_array.divisor, "vertex array 0 divisor is unimplemented!");
+    for (unsigned index = 1; index < Maxwell::NumVertexArrays; ++index) {
+        ASSERT_MSG(!regs.vertex_array[index].enable, "vertex array %d is unimplemented!", index);
+    }
+
+    // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
+    // Enables the first 16 vertex attributes always, as we don't know which ones are actually used
+    // until shader time. Note, Tegra technically supports 32, but we're cappinig this to 16 for now
+    // to avoid OpenGL errors.
+    for (unsigned index = 0; index < 16; ++index) {
+        auto& attrib = regs.vertex_attrib_format[index];
+        glVertexAttribPointer(index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib),
+                              attrib.IsNormalized() ? GL_TRUE : GL_FALSE, vertex_array.stride,
+                              reinterpret_cast<GLvoid*>(buffer_offset + attrib.offset));
+        glEnableVertexAttribArray(index);
+        hw_vao_enabled_attributes[index] = true;
+    }
+
+    // Copy vertex array data
+    const u32 data_size{vertex_array.stride * regs.vertex_buffer.count};
+    const VAddr data_addr{memory_manager->PhysicalToVirtualAddress(vertex_array.StartAddress())};
+    res_cache.FlushRegion(data_addr, data_size, nullptr);
+    Memory::ReadBlock(data_addr, array_ptr, data_size);
+
+    array_ptr += data_size;
+    buffer_offset += data_size;
 }
 
 void RasterizerOpenGL::SetupVertexShader(VSUniformData* ub_ptr, GLintptr buffer_offset) {
     MICROPROFILE_SCOPE(OpenGL_VS);
-    UNIMPLEMENTED();
+    LOG_CRITICAL(Render_OpenGL, "Emulated shaders are not supported! Using a passthrough shader.");
+    glUseProgramStages(pipeline.handle, GL_VERTEX_SHADER_BIT, current_shader->shader.handle);
 }
 
 void RasterizerOpenGL::SetupFragmentShader(FSUniformData* ub_ptr, GLintptr buffer_offset) {
     MICROPROFILE_SCOPE(OpenGL_FS);
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
     if (!has_ARB_separate_shader_objects) {
-        ASSERT_MSG(false, "Unimplemented");
+        UNREACHABLE();
         return false;
     }
 
     accelerate_draw = is_indexed ? AccelDraw::Indexed : AccelDraw::Arrays;
-    DrawTriangles();
+    DrawArrays();
 
     return true;
 }
 
-void RasterizerOpenGL::DrawTriangles() {
+void RasterizerOpenGL::DrawArrays() {
+    if (accelerate_draw == AccelDraw::Disabled)
+        return;
+
     MICROPROFILE_SCOPE(OpenGL_Drawing);
-    UNIMPLEMENTED();
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+
+    // TODO(bunnei): Implement these
+    const bool has_stencil = false;
+    const bool using_color_fb = true;
+    const bool using_depth_fb = false;
+    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()};
+
+    const bool write_color_fb =
+        state.color_mask.red_enabled == GL_TRUE || state.color_mask.green_enabled == GL_TRUE ||
+        state.color_mask.blue_enabled == GL_TRUE || state.color_mask.alpha_enabled == GL_TRUE;
+
+    const bool write_depth_fb =
+        (state.depth.test_enabled && state.depth.write_mask == GL_TRUE) ||
+        (has_stencil && state.stencil.test_enabled && state.stencil.write_mask != 0);
+
+    Surface color_surface;
+    Surface depth_surface;
+    MathUtil::Rectangle<u32> surfaces_rect;
+    std::tie(color_surface, depth_surface, surfaces_rect) =
+        res_cache.GetFramebufferSurfaces(using_color_fb, using_depth_fb, viewport_rect);
+
+    const u16 res_scale = color_surface != nullptr
+                              ? color_surface->res_scale
+                              : (depth_surface == nullptr ? 1u : depth_surface->res_scale);
+
+    MathUtil::Rectangle<u32> draw_rect{
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.left) +
+                                                  viewport_rect.left * res_scale,
+                                              surfaces_rect.left, surfaces_rect.right)), // Left
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.bottom) +
+                                                  viewport_rect.top * res_scale,
+                                              surfaces_rect.bottom, surfaces_rect.top)), // Top
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.left) +
+                                                  viewport_rect.right * res_scale,
+                                              surfaces_rect.left, surfaces_rect.right)), // Right
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.bottom) +
+                                                  viewport_rect.bottom * res_scale,
+                                              surfaces_rect.bottom, surfaces_rect.top))}; // Bottom
+
+    // Bind the framebuffer surfaces
+    BindFramebufferSurfaces(color_surface, depth_surface, has_stencil);
+
+    // Sync the viewport
+    SyncViewport(surfaces_rect, res_scale);
+
+    // TODO(bunnei): Sync framebuffer_scale uniform here
+    // TODO(bunnei): Sync scissorbox uniform(s) here
+    // TODO(bunnei): Sync and bind the texture surfaces
+
+    // Sync and bind the shader
+    if (shader_dirty) {
+        SetShader();
+        shader_dirty = false;
+    }
+
+    // Sync the uniform data
+    if (uniform_block_data.dirty) {
+        glBufferSubData(GL_UNIFORM_BUFFER, 0, sizeof(UniformData), &uniform_block_data.data);
+        uniform_block_data.dirty = false;
+    }
+
+    // Viewport can have negative offsets or larger dimensions than our framebuffer sub-rect. Enable
+    // scissor test to prevent drawing outside of the framebuffer region
+    state.scissor.enabled = true;
+    state.scissor.x = draw_rect.left;
+    state.scissor.y = draw_rect.bottom;
+    state.scissor.width = draw_rect.GetWidth();
+    state.scissor.height = draw_rect.GetHeight();
+    state.Apply();
+
+    // Draw the vertex batch
+    const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
+    AnalyzeVertexArray(is_indexed);
+    state.draw.vertex_buffer = stream_buffer->GetHandle();
+    state.Apply();
+
+    size_t buffer_size = static_cast<size_t>(vs_input_size);
+    if (is_indexed) {
+        UNREACHABLE();
+    }
+    buffer_size += sizeof(VSUniformData);
+
+    size_t ptr_pos = 0;
+    u8* buffer_ptr;
+    GLintptr buffer_offset;
+    std::tie(buffer_ptr, buffer_offset) =
+        stream_buffer->Map(static_cast<GLsizeiptr>(buffer_size), 4);
+
+    SetupVertexArray(buffer_ptr, buffer_offset);
+    ptr_pos += vs_input_size;
+
+    GLintptr index_buffer_offset = 0;
+    if (is_indexed) {
+        UNREACHABLE();
+    }
+
+    SetupVertexShader(reinterpret_cast<VSUniformData*>(&buffer_ptr[ptr_pos]),
+                      buffer_offset + static_cast<GLintptr>(ptr_pos));
+    const GLintptr vs_ubo_offset = buffer_offset + static_cast<GLintptr>(ptr_pos);
+    ptr_pos += sizeof(VSUniformData);
+
+    stream_buffer->Unmap();
+
+    const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) {
+        if (has_ARB_direct_state_access) {
+            glCopyNamedBufferSubData(stream_buffer->GetHandle(), handle, offset, 0, size);
+        } else {
+            glBindBuffer(GL_COPY_WRITE_BUFFER, handle);
+            glCopyBufferSubData(GL_ARRAY_BUFFER, GL_COPY_WRITE_BUFFER, offset, 0, size);
+        }
+    };
+
+    copy_buffer(vs_uniform_buffer.handle, vs_ubo_offset, sizeof(VSUniformData));
+
+    glUseProgramStages(pipeline.handle, GL_FRAGMENT_SHADER_BIT, current_shader->shader.handle);
+
+    if (is_indexed) {
+        UNREACHABLE();
+    } else {
+        glDrawArrays(MaxwellToGL::PrimitiveTopology(regs.draw.topology), 0,
+                     regs.vertex_buffer.count);
+    }
+
+    // Disable scissor test
+    state.scissor.enabled = false;
+
+    accelerate_draw = AccelDraw::Disabled;
+
+    // Unbind textures for potential future use as framebuffer attachments
+    for (auto& texture_unit : state.texture_units) {
+        texture_unit.texture_2d = 0;
+    }
+    state.Apply();
+
+    // Mark framebuffer surfaces as dirty
+    MathUtil::Rectangle<u32> draw_rect_unscaled{
+        draw_rect.left / res_scale, draw_rect.top / res_scale, draw_rect.right / res_scale,
+        draw_rect.bottom / res_scale};
+
+    if (color_surface != nullptr && write_color_fb) {
+        auto interval = color_surface->GetSubRectInterval(draw_rect_unscaled);
+        res_cache.InvalidateRegion(boost::icl::first(interval), boost::icl::length(interval),
+                                   color_surface);
+    }
+    if (depth_surface != nullptr && write_depth_fb) {
+        auto interval = depth_surface->GetSubRectInterval(draw_rect_unscaled);
+        res_cache.InvalidateRegion(boost::icl::first(interval), boost::icl::length(interval),
+                                   depth_surface);
+    }
 }
 
 void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 id) {}
@@ -206,17 +399,17 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
 
 bool RasterizerOpenGL::AccelerateDisplayTransfer(const void* config) {
     MICROPROFILE_SCOPE(OpenGL_Blits);
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return true;
 }
 
 bool RasterizerOpenGL::AccelerateTextureCopy(const void* config) {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return true;
 }
 
 bool RasterizerOpenGL::AccelerateFill(const void* config) {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return true;
 }
 
@@ -297,14 +490,14 @@ void main() {
         return;
     }
 
-    LOG_ERROR(HW_GPU, "Emulated shaders are not supported! Using a passthrough shader.");
+    LOG_CRITICAL(Render_OpenGL, "Emulated shaders are not supported! Using a passthrough shader.");
 
     current_shader = &test_shader;
     if (has_ARB_separate_shader_objects) {
         test_shader.shader.Create(vertex_shader, nullptr, fragment_shader, {}, true);
         glActiveShaderProgram(pipeline.handle, test_shader.shader.handle);
     } else {
-        ASSERT_MSG(false, "Unimplemented");
+        UNREACHABLE();
     }
 
     state.draw.shader_program = test_shader.shader.handle;
@@ -316,34 +509,70 @@ void main() {
     }
 }
 
+void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
+                                               const Surface& depth_surface, bool has_stencil) {
+    state.draw.draw_framebuffer = framebuffer.handle;
+    state.Apply();
+
+    glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
+                           color_surface != nullptr ? color_surface->texture.handle : 0, 0);
+    if (depth_surface != nullptr) {
+        if (has_stencil) {
+            // attach both depth and stencil
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
+                                   depth_surface->texture.handle, 0);
+        } else {
+            // attach depth
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
+                                   depth_surface->texture.handle, 0);
+            // clear stencil attachment
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
+        }
+    } else {
+        // clear both depth and stencil attachment
+        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
+                               0);
+    }
+}
+
+void RasterizerOpenGL::SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale) {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()};
+
+    state.viewport.x = static_cast<GLint>(surfaces_rect.left) + viewport_rect.left * res_scale;
+    state.viewport.y = static_cast<GLint>(surfaces_rect.bottom) + viewport_rect.bottom * res_scale;
+    state.viewport.width = static_cast<GLsizei>(viewport_rect.GetWidth() * res_scale);
+    state.viewport.height = static_cast<GLsizei>(viewport_rect.GetHeight() * res_scale);
+}
+
 void RasterizerOpenGL::SyncClipEnabled() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncClipCoef() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncCullMode() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncDepthScale() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncDepthOffset() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendEnabled() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendFuncs() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendColor() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index b387f383b..fd53e94cd 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -29,7 +29,7 @@ public:
     RasterizerOpenGL();
     ~RasterizerOpenGL() override;
 
-    void DrawTriangles() override;
+    void DrawArrays() override;
     void NotifyMaxwellRegisterChanged(u32 id) override;
     void FlushAll() override;
     void FlushRegion(VAddr addr, u64 size) override;
@@ -87,6 +87,13 @@ public:
 private:
     struct SamplerInfo {};
 
+    /// Binds the framebuffer color and depth surface
+    void BindFramebufferSurfaces(const Surface& color_surface, const Surface& depth_surface,
+                                 bool has_stencil);
+
+    /// Syncs the viewport to match the guest state
+    void SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale);
+
     /// Syncs the clip enabled status to match the guest state
     void SyncClipEnabled();
 
@@ -139,7 +146,7 @@ private:
     OGLVertexArray hw_vao;
     std::array<bool, 16> hw_vao_enabled_attributes;
 
-    std::array<SamplerInfo, 3> texture_samplers;
+    std::array<SamplerInfo, 32> texture_samplers;
     static constexpr size_t VERTEX_BUFFER_SIZE = 128 * 1024 * 1024;
     std::unique_ptr<OGLStreamBuffer> vertex_buffer;
     OGLBuffer uniform_buffer;
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
index 78fa7c051..2ffbd3bab 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -21,10 +21,13 @@
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "common/vector_math.h"
+#include "core/core.h"
 #include "core/frontend/emu_window.h"
+#include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
 #include "core/settings.h"
+#include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/utils.h"
@@ -110,65 +113,26 @@ static void MortonCopyTile(u32 stride, u8* tile_buffer, u8* gl_buffer) {
 template <bool morton_to_gl, PixelFormat format>
 static void MortonCopy(u32 stride, u32 height, u8* gl_buffer, VAddr base, VAddr start, VAddr end) {
     constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / 8;
-    constexpr u32 tile_size = bytes_per_pixel * 64;
-
     constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
-    static_assert(gl_bytes_per_pixel >= bytes_per_pixel, "");
-    gl_buffer += gl_bytes_per_pixel - bytes_per_pixel;
-
-    const VAddr aligned_down_start = base + Common::AlignDown(start - base, tile_size);
-    const VAddr aligned_start = base + Common::AlignUp(start - base, tile_size);
-    const VAddr aligned_end = base + Common::AlignDown(end - base, tile_size);
-
-    ASSERT(!morton_to_gl || (aligned_start == start && aligned_end == end));
-
-    const u64 begin_pixel_index = (aligned_down_start - base) / bytes_per_pixel;
-    u32 x = static_cast<u32>((begin_pixel_index % (stride * 8)) / 8);
-    u32 y = static_cast<u32>((begin_pixel_index / (stride * 8)) * 8);
-
-    gl_buffer += ((height - 8 - y) * stride + x) * gl_bytes_per_pixel;
-
-    auto glbuf_next_tile = [&] {
-        x = (x + 8) % stride;
-        gl_buffer += 8 * gl_bytes_per_pixel;
-        if (!x) {
-            y += 8;
-            gl_buffer -= stride * 9 * gl_bytes_per_pixel;
-        }
-    };
-
-    u8* tile_buffer = Memory::GetPointer(start);
-
-    if (start < aligned_start && !morton_to_gl) {
-        std::array<u8, tile_size> tmp_buf;
-        MortonCopyTile<morton_to_gl, format>(stride, &tmp_buf[0], gl_buffer);
-        std::memcpy(tile_buffer, &tmp_buf[start - aligned_down_start],
-                    std::min(aligned_start, end) - start);
-
-        tile_buffer += aligned_start - start;
-        glbuf_next_tile();
-    }
-
-    const u8* const buffer_end = tile_buffer + aligned_end - aligned_start;
-    while (tile_buffer < buffer_end) {
-        MortonCopyTile<morton_to_gl, format>(stride, tile_buffer, gl_buffer);
-        tile_buffer += tile_size;
-        glbuf_next_tile();
-    }
 
-    if (end > std::max(aligned_start, aligned_end) && !morton_to_gl) {
-        std::array<u8, tile_size> tmp_buf;
-        MortonCopyTile<morton_to_gl, format>(stride, &tmp_buf[0], gl_buffer);
-        std::memcpy(tile_buffer, &tmp_buf[0], end - aligned_end);
-    }
+    // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check the
+    // configuration for this and perform more generic un/swizzle
+    LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
+    VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel,
+                                   Memory::GetPointer(base), gl_buffer, morton_to_gl);
 }
 
 static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> morton_to_gl_fns = {
-    MortonCopy<true, PixelFormat::RGBA8>,  // 0
-    MortonCopy<true, PixelFormat::RGB8>,   // 1
-    MortonCopy<true, PixelFormat::RGB5A1>, // 2
-    MortonCopy<true, PixelFormat::RGB565>, // 3
-    MortonCopy<true, PixelFormat::RGBA4>,  // 4
+    MortonCopy<true, PixelFormat::RGBA8>,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
     nullptr,
     nullptr,
     nullptr,
@@ -177,19 +141,19 @@ static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> mo
     nullptr,
     nullptr,
     nullptr,
-    nullptr,                             // 5 - 13
-    MortonCopy<true, PixelFormat::D16>,  // 14
-    nullptr,                             // 15
-    MortonCopy<true, PixelFormat::D24>,  // 16
-    MortonCopy<true, PixelFormat::D24S8> // 17
 };
 
 static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> gl_to_morton_fns = {
-    MortonCopy<false, PixelFormat::RGBA8>,  // 0
-    MortonCopy<false, PixelFormat::RGB8>,   // 1
-    MortonCopy<false, PixelFormat::RGB5A1>, // 2
-    MortonCopy<false, PixelFormat::RGB565>, // 3
-    MortonCopy<false, PixelFormat::RGBA4>,  // 4
+    MortonCopy<false, PixelFormat::RGBA8>,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
     nullptr,
     nullptr,
     nullptr,
@@ -198,11 +162,6 @@ static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> gl
     nullptr,
     nullptr,
     nullptr,
-    nullptr,                              // 5 - 13
-    MortonCopy<false, PixelFormat::D16>,  // 14
-    nullptr,                              // 15
-    MortonCopy<false, PixelFormat::D24>,  // 16
-    MortonCopy<false, PixelFormat::D24S8> // 17
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
@@ -291,8 +250,8 @@ static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rec
 
 static bool FillSurface(const Surface& surface, const u8* fill_data,
                         const MathUtil::Rectangle<u32>& fill_rect, GLuint draw_fb_handle) {
-    ASSERT_MSG(false, "Unimplemented");
-    return true;
+    UNREACHABLE();
+    return {};
 }
 
 SurfaceParams SurfaceParams::FromInterval(SurfaceInterval interval) const {
@@ -531,7 +490,7 @@ MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64
 void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) {
     ASSERT(type != SurfaceType::Fill);
 
-    u8* texture_src_data = Memory::GetPointer(addr);
+    u8* const texture_src_data = Memory::GetPointer(addr);
     if (texture_src_data == nullptr)
         return;
 
@@ -548,11 +507,16 @@ void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) {
     if (!is_tiled) {
         ASSERT(type == SurfaceType::Color);
         const u32 bytes_per_pixel{GetFormatBpp() >> 3};
+
+        // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check
+        // the configuration for this and perform more generic un/swizzle
+        LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
         VideoCore::MortonCopyPixels128(width, height, bytes_per_pixel, 4,
                                        texture_src_data + start_offset, &gl_buffer[start_offset],
                                        true);
     } else {
-        ASSERT_MSG(false, "Unimplemented");
+        morton_to_gl_fns[static_cast<size_t>(pixel_format)](stride, height, &gl_buffer[0], addr,
+                                                            load_start, load_end);
     }
 }
 
@@ -1093,18 +1057,106 @@ SurfaceRect_Tuple RasterizerCacheOpenGL::GetSurfaceSubRect(const SurfaceParams&
 }
 
 Surface RasterizerCacheOpenGL::GetTextureSurface(const void* config) {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return {};
 }
 
 SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(
-    bool using_color_fb, bool using_depth_fb, const MathUtil::Rectangle<s32>& viewport_rect) {
-    UNIMPLEMENTED();
-    return {};
+    bool using_color_fb, bool using_depth_fb, const MathUtil::Rectangle<s32>& viewport) {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
+    const auto& config = regs.rt[0];
+
+    // TODO(bunnei): This is hard corded to use just the first render buffer
+    LOG_WARNING(Render_OpenGL, "hard-coded for render target 0!");
+
+    // update resolution_scale_factor and reset cache if changed
+    // TODO (bunnei): This code was ported as-is from Citra, and is technically not thread-safe. We
+    // need to fix this before making the renderer multi-threaded.
+    static u16 resolution_scale_factor = GetResolutionScaleFactor();
+    if (resolution_scale_factor != GetResolutionScaleFactor()) {
+        resolution_scale_factor = GetResolutionScaleFactor();
+        FlushAll();
+        while (!surface_cache.empty())
+            UnregisterSurface(*surface_cache.begin()->second.begin());
+    }
+
+    MathUtil::Rectangle<u32> viewport_clamped{
+        static_cast<u32>(MathUtil::Clamp(viewport.left, 0, static_cast<s32>(config.width))),
+        static_cast<u32>(MathUtil::Clamp(viewport.top, 0, static_cast<s32>(config.height))),
+        static_cast<u32>(MathUtil::Clamp(viewport.right, 0, static_cast<s32>(config.width))),
+        static_cast<u32>(MathUtil::Clamp(viewport.bottom, 0, static_cast<s32>(config.height)))};
+
+    // get color and depth surfaces
+    SurfaceParams color_params;
+    color_params.is_tiled = true;
+    color_params.res_scale = resolution_scale_factor;
+    color_params.width = config.width;
+    color_params.height = config.height;
+    SurfaceParams depth_params = color_params;
+
+    color_params.addr = memory_manager->PhysicalToVirtualAddress(config.Address());
+    color_params.pixel_format = SurfaceParams::PixelFormatFromRenderTargetFormat(config.format);
+    color_params.UpdateParams();
+
+    ASSERT_MSG(!using_depth_fb, "depth buffer is unimplemented");
+    // depth_params.addr = config.GetDepthBufferPhysicalAddress();
+    // depth_params.pixel_format = SurfaceParams::PixelFormatFromDepthFormat(config.depth_format);
+    // depth_params.UpdateParams();
+
+    auto color_vp_interval = color_params.GetSubRectInterval(viewport_clamped);
+    auto depth_vp_interval = depth_params.GetSubRectInterval(viewport_clamped);
+
+    // Make sure that framebuffers don't overlap if both color and depth are being used
+    if (using_color_fb && using_depth_fb &&
+        boost::icl::length(color_vp_interval & depth_vp_interval)) {
+        LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; "
+                                    "overlapping framebuffers not supported!");
+        using_depth_fb = false;
+    }
+
+    MathUtil::Rectangle<u32> color_rect{};
+    Surface color_surface = nullptr;
+    if (using_color_fb)
+        std::tie(color_surface, color_rect) =
+            GetSurfaceSubRect(color_params, ScaleMatch::Exact, false);
+
+    MathUtil::Rectangle<u32> depth_rect{};
+    Surface depth_surface = nullptr;
+    if (using_depth_fb)
+        std::tie(depth_surface, depth_rect) =
+            GetSurfaceSubRect(depth_params, ScaleMatch::Exact, false);
+
+    MathUtil::Rectangle<u32> fb_rect{};
+    if (color_surface != nullptr && depth_surface != nullptr) {
+        fb_rect = color_rect;
+        // Color and Depth surfaces must have the same dimensions and offsets
+        if (color_rect.bottom != depth_rect.bottom || color_rect.top != depth_rect.top ||
+            color_rect.left != depth_rect.left || color_rect.right != depth_rect.right) {
+            color_surface = GetSurface(color_params, ScaleMatch::Exact, false);
+            depth_surface = GetSurface(depth_params, ScaleMatch::Exact, false);
+            fb_rect = color_surface->GetScaledRect();
+        }
+    } else if (color_surface != nullptr) {
+        fb_rect = color_rect;
+    } else if (depth_surface != nullptr) {
+        fb_rect = depth_rect;
+    }
+
+    if (color_surface != nullptr) {
+        ValidateSurface(color_surface, boost::icl::first(color_vp_interval),
+                        boost::icl::length(color_vp_interval));
+    }
+    if (depth_surface != nullptr) {
+        ValidateSurface(depth_surface, boost::icl::first(depth_vp_interval),
+                        boost::icl::length(depth_vp_interval));
+    }
+
+    return std::make_tuple(color_surface, depth_surface, fb_rect);
 }
 
 Surface RasterizerCacheOpenGL::GetFillSurface(const void* config) {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return {};
 }
 
@@ -1348,5 +1400,33 @@ void RasterizerCacheOpenGL::UnregisterSurface(const Surface& surface) {
 }
 
 void RasterizerCacheOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
-    // ASSERT_MSG(false, "Unimplemented");
+    const u64 num_pages =
+        ((addr + size - 1) >> Memory::PAGE_BITS) - (addr >> Memory::PAGE_BITS) + 1;
+    const u64 page_start = addr >> Memory::PAGE_BITS;
+    const u64 page_end = page_start + num_pages;
+
+    // Interval maps will erase segments if count reaches 0, so if delta is negative we have to
+    // subtract after iterating
+    const auto pages_interval = PageMap::interval_type::right_open(page_start, page_end);
+    if (delta > 0)
+        cached_pages.add({pages_interval, delta});
+
+    for (const auto& pair : RangeFromInterval(cached_pages, pages_interval)) {
+        const auto interval = pair.first & pages_interval;
+        const int count = pair.second;
+
+        const VAddr interval_start_addr = boost::icl::first(interval) << Memory::PAGE_BITS;
+        const VAddr interval_end_addr = boost::icl::last_next(interval) << Memory::PAGE_BITS;
+        const u64 interval_size = interval_end_addr - interval_start_addr;
+
+        if (delta > 0 && count == delta)
+            Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, true);
+        else if (delta < 0 && count == -delta)
+            Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, false);
+        else
+            ASSERT(count >= 0);
+    }
+
+    if (delta < 0)
+        cached_pages.add({pages_interval, delta});
 }
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index 14f3cdc38..1f660d30c 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -41,7 +41,7 @@ static_assert(std::is_same<SurfaceRegions::interval_type, SurfaceCache::interval
 using SurfaceRect_Tuple = std::tuple<Surface, MathUtil::Rectangle<u32>>;
 using SurfaceSurfaceRect_Tuple = std::tuple<Surface, Surface, MathUtil::Rectangle<u32>>;
 
-using PageMap = boost::icl::interval_map<u32, int>;
+using PageMap = boost::icl::interval_map<u64, int>;
 
 enum class ScaleMatch {
     Exact,   // only accept same res scale
@@ -116,6 +116,15 @@ struct SurfaceParams {
         return GetFormatBpp(pixel_format);
     }
 
+    static PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) {
+        switch (format) {
+        case Tegra::RenderTargetFormat::RGBA8_UNORM:
+            return PixelFormat::RGBA8;
+        default:
+            UNREACHABLE();
+        }
+    }
+
     static PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
         switch (format) {
         case Tegra::FramebufferConfig::PixelFormat::ABGR8:
@@ -308,7 +317,7 @@ public:
 
     /// Get the color and depth surfaces based on the framebuffer configuration
     SurfaceSurfaceRect_Tuple GetFramebufferSurfaces(bool using_color_fb, bool using_depth_fb,
-                                                    const MathUtil::Rectangle<s32>& viewport_rect);
+                                                    const MathUtil::Rectangle<s32>& viewport);
 
     /// Get a surface that matches the fill config
     Surface GetFillSurface(const void* config);
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 0e0ef18cc..564ea8f9e 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -26,7 +26,7 @@ public:
           sanitize_mul(sanitize_mul), emit_cb(emit_cb), setemit_cb(setemit_cb) {}
 
     std::string Decompile() {
-        UNIMPLEMENTED();
+        UNREACHABLE();
         return {};
     }
 
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index f242bce1d..8f3c98800 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp
@@ -8,12 +8,12 @@
 namespace GLShader {
 
 std::string GenerateVertexShader(const MaxwellVSConfig& config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return {};
 }
 
 std::string GenerateFragmentShader(const MaxwellFSConfig& config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return {};
 }
 
diff --git a/src/video_core/renderer_opengl/gl_shader_util.cpp b/src/video_core/renderer_opengl/gl_shader_util.cpp
index a3ba16761..a6c6204d5 100644
--- a/src/video_core/renderer_opengl/gl_shader_util.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_util.cpp
@@ -38,8 +38,8 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
             if (result == GL_TRUE) {
                 LOG_DEBUG(Render_OpenGL, "%s", &vertex_shader_error[0]);
             } else {
-                LOG_ERROR(Render_OpenGL, "Error compiling vertex shader:\n%s",
-                          &vertex_shader_error[0]);
+                LOG_CRITICAL(Render_OpenGL, "Error compiling vertex shader:\n%s",
+                             &vertex_shader_error[0]);
             }
         }
     }
@@ -62,8 +62,8 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
             if (result == GL_TRUE) {
                 LOG_DEBUG(Render_OpenGL, "%s", &geometry_shader_error[0]);
             } else {
-                LOG_ERROR(Render_OpenGL, "Error compiling geometry shader:\n%s",
-                          &geometry_shader_error[0]);
+                LOG_CRITICAL(Render_OpenGL, "Error compiling geometry shader:\n%s",
+                             &geometry_shader_error[0]);
             }
         }
     }
@@ -86,8 +86,8 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
             if (result == GL_TRUE) {
                 LOG_DEBUG(Render_OpenGL, "%s", &fragment_shader_error[0]);
             } else {
-                LOG_ERROR(Render_OpenGL, "Error compiling fragment shader:\n%s",
-                          &fragment_shader_error[0]);
+                LOG_CRITICAL(Render_OpenGL, "Error compiling fragment shader:\n%s",
+                             &fragment_shader_error[0]);
             }
         }
     }
@@ -128,20 +128,20 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
         if (result == GL_TRUE) {
             LOG_DEBUG(Render_OpenGL, "%s", &program_error[0]);
         } else {
-            LOG_ERROR(Render_OpenGL, "Error linking shader:\n%s", &program_error[0]);
+            LOG_CRITICAL(Render_OpenGL, "Error linking shader:\n%s", &program_error[0]);
         }
     }
 
     // If the program linking failed at least one of the shaders was probably bad
     if (result == GL_FALSE) {
         if (vertex_shader) {
-            LOG_ERROR(Render_OpenGL, "Vertex shader:\n%s", vertex_shader);
+            LOG_CRITICAL(Render_OpenGL, "Vertex shader:\n%s", vertex_shader);
         }
         if (geometry_shader) {
-            LOG_ERROR(Render_OpenGL, "Geometry shader:\n%s", geometry_shader);
+            LOG_CRITICAL(Render_OpenGL, "Geometry shader:\n%s", geometry_shader);
         }
         if (fragment_shader) {
-            LOG_ERROR(Render_OpenGL, "Fragment shader:\n%s", fragment_shader);
+            LOG_CRITICAL(Render_OpenGL, "Fragment shader:\n%s", fragment_shader);
         }
     }
     ASSERT_MSG(result == GL_TRUE, "Shader not linked");
diff --git a/src/video_core/renderer_opengl/gl_state.h b/src/video_core/renderer_opengl/gl_state.h
index 940575dfa..c1f4efc8c 100644
--- a/src/video_core/renderer_opengl/gl_state.h
+++ b/src/video_core/renderer_opengl/gl_state.h
@@ -85,7 +85,7 @@ public:
     struct {
         GLuint texture_2d; // GL_TEXTURE_BINDING_2D
         GLuint sampler;    // GL_SAMPLER_BINDING
-    } texture_units[3];
+    } texture_units[32];
 
     struct {
         GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
new file mode 100644
index 000000000..d847317ac
--- /dev/null
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -0,0 +1,50 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <glad/glad.h>
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "video_core/engines/maxwell_3d.h"
+
+namespace MaxwellToGL {
+
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+
+inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
+    switch (attrib.type) {
+    case Maxwell::VertexAttribute::Type::UnsignedNorm: {
+
+        switch (attrib.size) {
+        case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
+            return GL_UNSIGNED_BYTE;
+        }
+
+        LOG_CRITICAL(Render_OpenGL, "Unimplemented vertex size=%s", attrib.SizeString().c_str());
+        UNREACHABLE();
+        return {};
+    }
+
+    case Maxwell::VertexAttribute::Type::Float:
+        return GL_FLOAT;
+    }
+
+    LOG_CRITICAL(Render_OpenGL, "Unimplemented vertex type=%s", attrib.TypeString().c_str());
+    UNREACHABLE();
+    return {};
+}
+
+inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
+    switch (topology) {
+    case Maxwell::PrimitiveTopology::TriangleStrip:
+        return GL_TRIANGLE_STRIP;
+    }
+    LOG_CRITICAL(Render_OpenGL, "Unimplemented primitive topology=%d", topology);
+    UNREACHABLE();
+    return {};
+}
+
+} // namespace MaxwellToGL
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index 1a24855d7..78b50b227 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -100,6 +100,8 @@ RendererOpenGL::~RendererOpenGL() = default;
 
 /// Swap buffers (render frame)
 void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) {
+    Core::System::GetInstance().perf_stats.EndSystemFrame();
+
     // Maintain the rasterizer's state as a priority
     OpenGLState prev_state = OpenGLState::GetCurState();
     state.Apply();
@@ -114,20 +116,19 @@ void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&
             // performance problem.
             ConfigureFramebufferTexture(screen_info.texture, *framebuffer);
         }
+
+        // Load the framebuffer from memory, draw it to the screen, and swap buffers
         LoadFBToScreenInfo(*framebuffer, screen_info);
+        DrawScreen();
+        render_window->SwapBuffers();
     }
 
-    DrawScreens();
-
-    Core::System::GetInstance().perf_stats.EndSystemFrame();
-
-    // Swap buffers
     render_window->PollEvents();
-    render_window->SwapBuffers();
 
     Core::System::GetInstance().frame_limiter.DoFrameLimiting(CoreTiming::GetGlobalTimeUs());
     Core::System::GetInstance().perf_stats.BeginSystemFrame();
 
+    // Restore the rasterizer state
     prev_state.Apply();
     RefreshRasterizerSetting();
 }
@@ -141,11 +142,6 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
     const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel};
     const VAddr framebuffer_addr{framebuffer.address + framebuffer.offset};
 
-    // TODO(bunnei): The framebuffer region should only be invalidated if it is written to, not
-    // every frame. When we find the right place for this, the below line can be removed.
-    Memory::RasterizerFlushVirtualRegion(framebuffer_addr, size_in_bytes,
-                                         Memory::FlushMode::Invalidate);
-
     // Framebuffer orientation handling
     framebuffer_transform_flags = framebuffer.transform_flags;
 
@@ -283,7 +279,7 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
         gl_framebuffer_data.resize(texture.width * texture.height * 4);
         break;
     default:
-        UNIMPLEMENTED();
+        UNREACHABLE();
     }
 
     state.texture_units[0].texture_2d = texture.resource.handle;
@@ -297,8 +293,8 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
     state.Apply();
 }
 
-void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w,
-                                      float h) {
+void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x, float y, float w,
+                                         float h) {
     const auto& texcoords = screen_info.display_texcoords;
     auto left = texcoords.left;
     auto right = texcoords.right;
@@ -309,7 +305,7 @@ void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, fl
             right = texcoords.left;
         } else {
             // Other transformations are unsupported
-            LOG_CRITICAL(HW_GPU, "unsupported framebuffer_transform_flags=%d",
+            LOG_CRITICAL(Render_OpenGL, "Unsupported framebuffer_transform_flags=%d",
                          framebuffer_transform_flags);
             UNIMPLEMENTED();
         }
@@ -334,7 +330,7 @@ void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, fl
 /**
  * Draws the emulated screens to the emulator window.
  */
-void RendererOpenGL::DrawScreens() {
+void RendererOpenGL::DrawScreen() {
     const auto& layout = render_window->GetFramebufferLayout();
     const auto& screen = layout.screen;
 
@@ -350,8 +346,8 @@ void RendererOpenGL::DrawScreens() {
     glActiveTexture(GL_TEXTURE0);
     glUniform1i(uniform_color_texture, 0);
 
-    DrawSingleScreen(screen_info, (float)screen.left, (float)screen.top, (float)screen.GetWidth(),
-                     (float)screen.GetHeight());
+    DrawScreenTriangles(screen_info, (float)screen.left, (float)screen.top,
+                        (float)screen.GetWidth(), (float)screen.GetHeight());
 
     m_current_frame++;
 }
diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h
index 29516baf4..fffd0f9f4 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.h
+++ b/src/video_core/renderer_opengl/renderer_opengl.h
@@ -55,8 +55,8 @@ private:
     void InitOpenGLObjects();
     void ConfigureFramebufferTexture(TextureInfo& texture,
                                      const Tegra::FramebufferConfig& framebuffer);
-    void DrawScreens();
-    void DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w, float h);
+    void DrawScreen();
+    void DrawScreenTriangles(const ScreenInfo& screen_info, float x, float y, float w, float h);
     void UpdateFramerate();
 
     // Loads framebuffer from emulated memory into the display information structure
diff --git a/src/video_core/video_core.cpp b/src/video_core/video_core.cpp
index 864691baa..289140f31 100644
--- a/src/video_core/video_core.cpp
+++ b/src/video_core/video_core.cpp
@@ -26,7 +26,7 @@ bool Init(EmuWindow* emu_window) {
     if (g_renderer->Init()) {
         LOG_DEBUG(Render, "initialized OK");
     } else {
-        LOG_ERROR(Render, "initialization failed !");
+        LOG_CRITICAL(Render, "initialization failed !");
         return false;
     }
     return true;
diff --git a/src/yuzu/debugger/graphics/graphics_surface.cpp b/src/yuzu/debugger/graphics/graphics_surface.cpp
index 8e6509adc..1e4844b57 100644
--- a/src/yuzu/debugger/graphics/graphics_surface.cpp
+++ b/src/yuzu/debugger/graphics/graphics_surface.cpp
@@ -339,11 +339,10 @@ void GraphicsSurfaceWidget::OnUpdate() {
                                 static_cast<size_t>(Source::RenderTarget0)];
 
         surface_address = rt.Address();
-        surface_width = rt.horiz;
-        surface_height = rt.vert;
-        if (rt.format != 0) {
-            surface_format =
-                ConvertToTextureFormat(static_cast<Tegra::RenderTargetFormat>(rt.format));
+        surface_width = rt.width;
+        surface_height = rt.height;
+        if (rt.format != Tegra::RenderTargetFormat::NONE) {
+            surface_format = ConvertToTextureFormat(rt.format);
         }
 
         break;