optimized build script and started to split shaders. not yet working, opengl only handles one asset file and vulkan and directx still need the remaining code impl.

asset/AssetArchive.h

@@ -13,7 +13,6 @@
 #include "../utils/StringUtils.h"
 #include "../utils/EndianUtils.h"
 #include "../utils/Utils.h"
-#include "../stdlib/simd/SIMD_I32.h"
 #include "../memory/RingMemory.h"
 #include "../memory/BufferMemory.h"
 #include "../image/Image.cpp"
@@ -25,6 +24,12 @@
 #include "../ui/UITheme.h"
 #include "AssetManagementSystem.h"
 
+#if __aarch64__
+    #include "../stdlib/sve/SVE_I32.h"
+#else
+    #include "../stdlib/simd/SIMD_I32.h"
+#endif
+
 #if _WIN32
     #include <windows.h>
     #include "../platform/win32/FileUtils.cpp"
@@ -277,6 +282,7 @@ Asset* asset_archive_asset_load(const AssetArchive* archive, int32 id, AssetMana
                 theme_from_data(file.content, theme);
             } break;
             default: {
+                UNREACHABLE();
             }
         }
     }

camera/Camera.h

@@ -144,7 +144,9 @@ void camera_movement(Camera* camera, CameraMovement* movement, f32 dt, bool rela
                 case CAMERA_MOVEMENT_ZOOM_OUT: {
                         camera->zoom -= velocity;
                     } break;
-                default: {}
+                default: {
+                    UNREACHABLE();
+                }
             }
         }
     } else {
@@ -215,7 +217,9 @@ void camera_movement(Camera* camera, CameraMovement* movement, f32 dt, bool rela
                 case CAMERA_MOVEMENT_ZOOM_OUT: {
                         camera->zoom -= velocity;
                     } break;
-                default: {}
+                default: {
+                    UNREACHABLE();
+                }
             }
         }
     }

font/Font.h

@@ -5,7 +5,12 @@
 #include "../memory/BufferMemory.h"
 #include "../utils/EndianUtils.h"
 #include "../utils/Utils.h"
-#include "../stdlib/simd/SIMD_I32.h"
+
+#if __aarch64__
+    #include "../stdlib/sve/SVE_I32.h"
+#else
+    #include "../stdlib/simd/SIMD_I32.h"
+#endif
 
 #if _WIN32
     #include "../platform/win32/FileUtils.cpp"

gpuapi/RenderUtils.h

@@ -20,8 +20,8 @@
 #include "../ui/UIElement.h"
 #include "../ui/UIAlignment.h"
 
-// @performance Create improved vertice generation for components (input + button, chat, ...) where we don't use as many
-//      degenerate triangled
+// @performance Create improved vertex generation for components (input + button, chat, ...) where we don't use as many
+//      degenerate triangle
 
 // @todo in many places we use ->value_int. We should load it as a value_float and also define it as float in the theme.
 // This way we wouldn't have to convert the value

gpuapi/direct3d/GpuApiContainer.h

@@ -0,0 +1,33 @@
+/**
+ * Jingga
+ *
+ * @copyright Jingga
+ * @license   OMS License 2.0
+ * @version   1.0.0
+ * @link      https://jingga.app
+ */
+#ifndef TOS_GPUAPI_DIRECTX_GPU_API_CONTAINER
+#define TOS_GPUAPI_DIRECTX_GPU_API_CONTAINER
+
+#include <windows.h>
+#include <d3d12.h>
+#include <dxgi1_6.h>
+#include <d3dcompiler.h>
+#include <wrl.h>
+
+struct GpuApiContainer {
+    Microsoft::WRL::ComPtr<ID3D12Device> device;
+    Microsoft::WRL::ComPtr<IDXGISwapChain3> swapChain;
+
+    Microsoft::WRL::ComPtr<ID3D12CommandQueue> commandQueue;
+    Microsoft::WRL::ComPtr<ID3D12DescriptorHeap> rtvHeap;
+    Microsoft::WRL::ComPtr<ID3D12Resource> renderTargets[2];
+    Microsoft::WRL::ComPtr<ID3D12CommandAllocator> commandAllocator;
+    Microsoft::WRL::ComPtr<ID3D12GraphicsCommandList> commandList;
+    Microsoft::WRL::ComPtr<ID3D12PipelineState> pipelineState;
+    Microsoft::WRL::ComPtr<ID3D12RootSignature> rootSignature;
+    Microsoft::WRL::ComPtr<ID3D12Fence> fence;
+    UINT64 fenceValue = 0;
+};
+
+#endif

gpuapi/opengl/GpuApiContainer.h

@@ -0,0 +1,15 @@
+/**
+ * Jingga
+ *
+ * @copyright Jingga
+ * @license   OMS License 2.0
+ * @version   1.0.0
+ * @link      https://jingga.app
+ */
+#ifndef TOS_GPUAPI_OPENGL_GPU_API_CONTAINER
+#define TOS_GPUAPI_OPENGL_GPU_API_CONTAINER
+
+struct GpuApiContainer {
+};
+
+#endif

gpuapi/vulkan/GpuApiContainer.h

@@ -0,0 +1,28 @@
+/**
+ * Jingga
+ *
+ * @copyright Jingga
+ * @license   OMS License 2.0
+ * @version   1.0.0
+ * @link      https://jingga.app
+ */
+#ifndef TOS_GPUAPI_VULKAN_GPU_API_CONTAINER
+#define TOS_GPUAPI_VULKAN_GPU_API_CONTAINER
+
+#include <vulkan/vulkan.h>
+
+struct GpuApiContainer {
+    VkInstance instance;
+    VkSurfaceKHR surface;
+    VkDevice device;
+    VkSwapchainKHR swapchain;
+    VkPipelineLayout pipelineLayout;
+    VkFramebuffer framebuffer;
+    VkQueue graphicsQueue;
+    VkRenderPass renderPass;
+    VkPipeline pipeline;
+    VkCommandPool commandPool;
+    VkCommandBuffer commandBuffer;
+};
+
+#endif

image/Image.cpp

@@ -87,7 +87,7 @@ int32 image_pixel_size_from_type(byte type)
             return 4;
         } break;
         default: {
-            return 0;
+            UNREACHABLE();
         }
     }
 }

log/Debug.cpp

@@ -411,7 +411,9 @@ void log(const char* format, LogDataType data_type, void* data, bool should_log,
         case LOG_DATA_FLOAT64: {
             sprintf(temp, format, *((f64 *) data));
         } break;
-        default: {}
+        default: {
+            UNREACHABLE();
+        }
     }
 
     if (save || debug_container->log_memory.size - debug_container->log_memory.pos < MAX_LOG_LENGTH) {

math/matrix/VectorFloat32.h

@@ -10,7 +10,12 @@
 #define TOS_MATH_MATRIX_VECTOR_FLOAT32_H
 
 #include "../../utils/MathUtils.h"
-#include "../../stdlib/simd/SIMD_F32.h"
+
+#if __aarch64__
+    #include "../../../GameEngine/stdlib/sve/SVE_F32.h"
+#else
+    #include "../../../GameEngine/stdlib/simd/SIMD_F32.h"
+#endif
 
 struct v3_f32_4 {
     union {

math/matrix/VectorFloat64.h

@@ -10,6 +10,11 @@
 #define TOS_MATH_MATRIX_VECTOR_FLOAT64_H
 
 #include "../../utils/MathUtils.h"
-#include "../../stdlib/simd/SIMD_F64.h"
+
+#if __aarch64__
+    #include "../../../GameEngine/stdlib/sve/SVE_F64.h"
+#else
+    #include "../../../GameEngine/stdlib/simd/SIMD_F64.h"
+#endif
 
 #endif

math/matrix/VectorInt32.h

@@ -13,7 +13,12 @@
 #include <xmmintrin.h>
 
 #include "../../utils/MathUtils.h"
-#include "../../stdlib/simd/SIMD_I32.h"
+
+#if __aarch64__
+    #include "../../../GameEngine/stdlib/sve/SVE_I32.h"
+#else
+    #include "../../../GameEngine/stdlib/simd/SIMD_I32.h"
+#endif
 
 struct v3_int32_4 {
     union {

math/matrix/VectorInt64.h

@@ -13,7 +13,12 @@
 #include <xmmintrin.h>
 
 #include "../../utils/MathUtils.h"
-#include "../../stdlib/simd/SIMD_I64.h"
+
+#if __aarch64__
+    #include "../../../GameEngine/stdlib/sve/SVE_I64.h"
+#else
+    #include "../../../GameEngine/stdlib/simd/SIMD_I64.h"
+#endif
 
 struct v3_int64_2 {
     union {

memory/Queue.h

@@ -14,21 +14,26 @@
 
 typedef RingMemory Queue;
 
+// @question Consider to add the element size into the Queue struct -> we don't need to pass it after initialization as parameter
+
 inline
 void queue_alloc(Queue* queue, uint64 element_count, uint32 element_size, int32 alignment = 64)
 {
+    // @bug The alignment needs to be included in EVERY element
     ring_alloc(queue, element_count * element_size, alignment);
 }
 
 inline
 void queue_init(Queue* queue, BufferMemory* buf, uint64 element_count, uint32 element_size, int32 alignment = 64)
 {
+    // @bug The alignment needs to be included in EVERY element
     ring_init(queue, buf, element_count * element_size, alignment);
 }
 
 inline
 void queue_init(Queue* queue, byte* buf, uint64 element_count, uint32 element_size, int32 alignment = 64)
 {
+    // @bug The alignment needs to be included in EVERY element
     ring_init(queue, buf, element_count * element_size, alignment);
 }
 

memory/ThreadedQueue.h

@@ -48,9 +48,12 @@ struct ThreadedQueue {
     sem_t full;
 };
 
+// @question Consider to add the element size into the Queue struct -> we don't need to pass it after initialization as parameter
+
 inline
 void thrd_queue_alloc(ThreadedQueue* queue, uint32 element_count, uint64 element_size, int32 alignment = 64)
 {
+    // @bug The alignment needs to be included in EVERY element
     ring_alloc((RingMemory *) queue, element_count * element_size, alignment);
 
     pthread_mutex_init(&queue->mutex, NULL);
@@ -63,6 +66,7 @@ void thrd_queue_alloc(ThreadedQueue* queue, uint32 element_count, uint64 element
 inline
 void thrd_queue_init(ThreadedQueue* queue, BufferMemory* buf, uint32 element_count, uint64 element_size, int32 alignment = 64)
 {
+    // @bug The alignment needs to be included in EVERY element
     ring_init((RingMemory *) queue, buf, element_count * element_size, alignment);
 
     pthread_mutex_init(&queue->mutex, NULL);
@@ -75,6 +79,7 @@ void thrd_queue_init(ThreadedQueue* queue, BufferMemory* buf, uint32 element_cou
 inline
 void thrd_queue_init(ThreadedQueue* queue, byte* buf, uint32 element_count, uint64 element_size, int32 alignment = 64)
 {
+    // @bug The alignment needs to be included in EVERY element
     ring_init((RingMemory *) queue, buf, element_count * element_size, alignment);
 
     pthread_mutex_init(&queue->mutex, NULL);

models/settings/setting_types.h

@@ -10,11 +10,6 @@
 #define SETTING_TYPE_GPU_ULTRA 0x6
 #define SETTING_TYPE_GPU_NEXTGEN 0x7
 
-#define SETTING_TYPE_GPU_API_NONE 0x0
-#define SETTING_TYPE_GPU_API_DIRECTX11 0x1
-#define SETTING_TYPE_GPU_API_DIRECTX12 0x2
-#define SETTING_TYPE_GPU_API_OPENGL 0x3
-
 #define SETTING_TYPE_PERSPECTIVE_FIRST 0x00
 #define SETTING_TYPE_PERSPECTIVE_THIRD 0x01
 #define SETTING_TYPE_PERSPECTIVE_ISOMETRIC 0x02

object/Mesh.h

@@ -19,10 +19,15 @@
 #endif
 
 #include "../memory/RingMemory.h"
-#include "../stdlib/simd/SIMD_I32.h"
 #include "../utils/EndianUtils.h"
 #include "../utils/StringUtils.h"
 
+#if __aarch64__
+    #include "../../../GameEngine/stdlib/sve/SVE_I32.h"
+#else
+    #include "../../../GameEngine/stdlib/simd/SIMD_I32.h"
+#endif
+
 #define MESH_VERSION 1
 
 // @todo how to handle different objects and groups?
@@ -378,6 +383,9 @@ void mesh_from_file_txt(
                     }
                     text[i] = '\0';
                 } break;
+            default: {
+                UNREACHABLE();
+            }
         }
     }
 

platform/linux/SystemInfo.cpp

@@ -12,12 +12,17 @@
 #include <stdio.h>
 #include <stdint.h>
 #include "../../stdlib/Types.h"
-#include "../../stdlib/simd/SIMD_Helper.h"
-#include "../SystemInfo.h"
+#include "../../system/SystemInfo.h"
 
 #include <locale.h>
 #include <cpuid.h>
 
+#if __aarch64__
+    #include "../../stdlib/simd/SIMD_Helper.h"
+#else
+    #include "../../stdlib/sve/SVE_Helper.h"
+#endif
+
 // @todo implement for arm?
 
 uint16 system_language_code()

platform/win32/Library.h

@@ -16,7 +16,7 @@
 #include "../../stdlib/Types.h"
 #include "FileUtils.cpp"
 #include "../../utils/StringUtils.h"
-#include "../Library.h"
+#include "../../system/Library.h"
 
 // @todo Rename file to Library.cpp
 

platform/win32/SystemInfo.cpp

@@ -12,8 +12,7 @@
 #include <stdio.h>
 #include <stdint.h>
 #include "../../stdlib/Types.h"
-#include "../../stdlib/simd/SIMD_Helper.h"
-#include "../SystemInfo.h"
+#include "../../system/SystemInfo.h"
 
 #include <psapi.h>
 #include <winsock2.h>
@@ -27,6 +26,12 @@
 #include <winnls.h>
 #include <hidsdi.h>
 
+#if __aarch64__
+    #include "../../stdlib/sve/SVE_Helper.h"
+#else
+    #include "../../stdlib/simd/SIMD_Helper.h"
+#endif
+
 // @performance Do we really need all these libs, can't we simplify that?!
 #include <intrin.h>
 #pragma comment(lib, "Advapi32.lib")

platform/win32/UtilsWindows.h

@@ -85,12 +85,15 @@ void window_create(Window* __restrict window, void* proc)
 
     WNDPROC wndproc = (WNDPROC) proc;
     WNDCLASSEXA wc = {};
-    HINSTANCE hinstance = GetModuleHandle(0);
+
+    if (!window->hInstance) {
+        window->hInstance = GetModuleHandle(0);
+    }
 
     wc.cbSize = sizeof(WNDCLASSEXA);
     wc.style = CS_OWNDC;
     wc.lpfnWndProc = wndproc;
-    wc.hInstance = hinstance;
+    wc.hInstance = window->hInstance;
     wc.hCursor = LoadCursor(NULL, IDC_ARROW);
     wc.lpszClassName = (LPCSTR) window->name;
 
@@ -123,7 +126,7 @@ void window_create(Window* __restrict window, void* proc)
         window->x, window->y,
         window->width,
         window->height,
-        NULL, NULL, hinstance, window
+        NULL, NULL, window->hInstance, window
     );
 
     ASSERT_SIMPLE(window->hwnd);

platform/win32/Window.h

@@ -11,24 +11,9 @@
 
 #include <windows.h>
 #include "../../stdlib/Types.h"
+#include "../../system/Window.h"
 
-struct WindowState {
-    uint16 width;
-    uint16 height;
-
-    uint16 x;
-    uint16 y;
-    uint64 style;
-};
-
-enum WindowStateChanges : byte {
-    WINDOW_STATE_CHANGE_NONE = 0,
-    WINDOW_STATE_CHANGE_SIZE = 1,
-    WINDOW_STATE_CHANGE_POS = 2,
-    WINDOW_STATE_CHANGE_FOCUS = 4,
-    WINDOW_STATE_CHANGE_FULLSCREEN = 8,
-    WINDOW_STATE_CHANGE_ALL = 16,
-};
+typedef HINSTANCE WindowInstance;
 
 struct Window {
     uint16 width;
@@ -48,6 +33,7 @@ struct Window {
     HWND hwnd;
     HDC hdc;
     HGLRC openGLRC;
+    HINSTANCE hInstance;
 
     char name[32];
     WindowState state_old;

platform/win32/input/RawInput.h

@@ -197,9 +197,7 @@ int rawinput_init_controllers(HWND hwnd, Input* __restrict states, RingMemory* r
                         }
                     }
                 } break;
-            default: {
-
-            }
+            default: {}
         }
     }
 

stdlib/HashMap.h

@@ -72,6 +72,10 @@ struct HashMap {
     ChunkMemory buf;
 };
 
+// @performance Implement more like gperf, our implementation is slow. However, keep it around since it is very general purpose
+// Alternatively, also create a version that creates perfect hashes (input requires a hash function and a seed for that hash function)
+// Both would be saved in the hash impl.
+
 // WARNING: element_size = element size + remaining HashEntry data size
 void hashmap_create(HashMap* hm, int32 count, int32 element_size, RingMemory* ring)
 {

stdlib/IntrinsicsArm.h

@@ -67,20 +67,4 @@ inline float oms_ceil(float a) {
     return svget1_f32(result);
 }
 
-inline void atomic_increment(int32_t* a, int32_t b) {
-    __atomic_add_fetch(a, b, __ATOMIC_SEQ_CST);
-}
-
-inline void atomic_increment(int64_t* a, int64_t b) {
-    __atomic_add_fetch(a, b, __ATOMIC_SEQ_CST);
-}
-
-inline void atomic_decrement(int32_t* a, int32_t b) {
-    __atomic_sub_fetch(a, b, __ATOMIC_SEQ_CST);
-}
-
-inline void atomic_decrement(int64_t* a, int64_t b) {
-    __atomic_sub_fetch(a, b, __ATOMIC_SEQ_CST);
-}
-
 #endif

stdlib/Types.h

@@ -17,9 +17,13 @@
     #define PACKED_STRUCT  __pragma(pack(push, 1))
     #define UNPACKED_STRUCT __pragma(pack(pop))
     typedef SSIZE_T ssize_t;
+
+    #define UNREACHABLE() __assume(0)
 #else
     #define PACKED_STRUCT  __attribute__((__packed__))
     #define UNPACKED_STRUCT ((void) 0)
+
+    #define UNREACHABLE() __builtin_unreachable()
 #endif
 
 #define ARRAY_COUNT(a) (sizeof(a) / sizeof((a)[0]))

system/Library.h

@@ -6,8 +6,8 @@
  * @version   1.0.0
  * @link      https://jingga.app
  */
-#ifndef TOS_PLATFORM_LIBRARY_H
-#define TOS_PLATFORM_LIBRARY_H
+#ifndef TOS_SYSTEM_LIBRARY_H
+#define TOS_SYSTEM_LIBRARY_H
 
 #include "../stdlib/Types.h"
 

system/SystemInfo.h

@@ -6,8 +6,8 @@
  * @version   1.0.0
  * @link      https://jingga.app
  */
-#ifndef TOS_PLATFORM_SYSTEM_INFO_H
-#define TOS_PLATFORM_SYSTEM_INFO_H
+#ifndef TOS_SYSTEM_INFO_H
+#define TOS_SYSTEM_INFO_H
 
 #include "../stdlib/Types.h"
 

system/Window.h

@@ -0,0 +1,32 @@
+/**
+ * Jingga
+ *
+ * @copyright Jingga
+ * @license   OMS License 2.0
+ * @version   1.0.0
+ * @link      https://jingga.app
+ */
+#ifndef TOS_SYSTEM_WINDOW_H
+#define TOS_SYSTEM_WINDOW_H
+
+#include "../stdlib/Types.h"
+
+struct WindowState {
+    uint16 width;
+    uint16 height;
+
+    uint16 x;
+    uint16 y;
+    uint64 style;
+};
+
+enum WindowStateChanges : byte {
+    WINDOW_STATE_CHANGE_NONE = 0,
+    WINDOW_STATE_CHANGE_SIZE = 1,
+    WINDOW_STATE_CHANGE_POS = 2,
+    WINDOW_STATE_CHANGE_FOCUS = 4,
+    WINDOW_STATE_CHANGE_FULLSCREEN = 8,
+    WINDOW_STATE_CHANGE_ALL = 16,
+};
+
+#endif

utils/StringUtils.h

@@ -282,6 +282,42 @@ str_concat(
     return len_total + len;
 }
 
+inline void
+str_add(char* base, const char* src)
+{
+    while (*base) {
+        ++base;
+    }
+
+    strcpy(base, src);
+}
+
+inline void
+str_add(char* base, const char* src, size_t src_length)
+{
+    while (*base) {
+        ++base;
+    }
+
+    memcpy(base, src, src_length);
+    base[src_length] = '\0';
+}
+
+inline int64
+str_add(char* base, size_t base_length, const char* src, size_t src_length)
+{
+    memcpy(&base[base_length], src, src_length);
+    base[base_length + src_length] = '\0';
+
+    return base_length + src_length;
+}
+
+inline void
+str_add(char* base, size_t base_length, const char* src)
+{
+    strcpy(&base[base_length], src);
+}
+
 inline int64
 str_concat(
     const char* src1, size_t src1_length,