started work on async code

build.sh

@@ -76,9 +76,10 @@ mkdir -p build
 
 cd build
 
+mkdir -p ../src/file_data
+
 if [ -v vulkan ]; then
 	mkdir -p ./shaders/glsl
-	mkdir -p ../src/file_data
 	
 	for file in ../src/shaders/glsl/*.glsl; do
 		base_name=$(basename -- "$file" .glsl)
@@ -107,6 +108,16 @@ if [ -v vulkan ]; then
 
 fi
 
+rm -f ../src/file_data/images.c
+touch ../src/file_data/images.c
+
+for file in ../assets/*.png; do
+	base_name=$(basename -- "$file" .png)
+
+	xxd -n "image_${base_name}" -i $file >> ../src/file_data/images.c
+done
+
+
 $cpp_compiler $vma_compile_flags $vma_source_files $vma_out $vma_obj
 
 ar rcs libvma.a vma.o

src/entry_linux.c

@@ -8,6 +8,45 @@
 #include "renderer.c"
 #include "game.c"
 
+const char *strs[10] = {
+	"String 1",
+	"String 2",
+	"String 3",
+	"String 4",
+	"String 5",
+	"String 6",
+	"String 7",
+	"String 8",
+	"String 9",
+	"String 10",
+};
+u32 volatile str_index = 0;
+
+#include <unistd.h>
+
+void *ThreadFunc(void *i)
+{
+	for (;;)
+	{
+		u32 val = __atomic_fetch_add(&str_index, 1, __ATOMIC_RELEASE);
+		if (val < 10)
+		{
+			Printfln("Thread %d: %s", *(u32 *)i, strs[val]);
+			sleep(1);
+		}
+		else 
+			break;
+
+	}
+
+	pthread_exit(NULL);
+}
+
+void RunThreadFunc(pthread_t *th, u32 *u, void *func)
+{
+	pthread_create(th, NULL, func, u);
+}
+
 int main(int argc, char **argv) 
 {
 	u8 *mem = (u8 *)MemAllocZeroed(MB(64));

src/entry_linux.h

@@ -2,6 +2,8 @@
 
 #pragma once
 
+#define _GNU_SOURCE
+
 #define STB_SPRINTF_IMPLEMENTATION
 
 #define WINDOW_NAME "Video Game"

src/platform_linux.c

@@ -558,3 +558,10 @@ b32 _ShouldQuit()
 {
 	return false;
 }
+
+u32 AvailableCPUCount()
+{
+	cpu_set_t cpu_set;
+	sched_getaffinity(0, sizeof(cpu_set), &cpu_set);
+	return CPU_COUNT(&cpu_set);
+}

src/platform_linux.h

@@ -1,5 +1,6 @@
 #pragma once
 
+#include <pthread.h>
 #include <limits.h>
 #include <stdint.h>
 #include <stdbool.h>
@@ -17,6 +18,9 @@
 #include <stdlib.h>
 #include <dlfcn.h>
 #include <nmmintrin.h>
+#include <immintrin.h>
+
+#include <sched.h>
 
 // syscall defines
 #define SYS_ERR -1
@@ -89,6 +93,7 @@ void RepaintWindow();
 
 // General Utils
 b32 CheckSyscallErr(void *ptr);
+u32 AvailableCPUCount();
 
 // Write Utils
 i32 Write(int fd, void const *str, isize count);

src/renderer.h

@@ -34,6 +34,8 @@ typedef enum VertexAttrType_e
 
 // Declarations
 
+typedef u16 DescHandle;
+
 // @requirement RenderBuffer type;
 // @requirement u32 size;
 typedef struct RenderBuffer_t RenderBuffer;
@@ -51,27 +53,29 @@ typedef struct RenderBuffers_t
 
 // Back End API
 
-// Initialization
+// ::Initialization::Header::
 b32 InitRenderer(Arena *arena);
 void DestroyRenderer();
 
-// Buffers
+// ::Buffers::Header::
 static b32 CreateBuffer(RenderBuffer *buffer);
 static void FreeBuffers(RenderBuffer *buffers, u32 buffer_count);
-static b32 UploadToBuffer(RenderBuffer *buffer, rawptr ptr);
+static b32 UploadToBuffer(RenderBuffer *buffer, rawptr ptr, u8 thr_ix);
 static b32 CreateAndUploadToBuffer(RenderBuffer *buffer, rawptr ptr);
 static void BindVertexBuffer(RenderBuffer *buffer);
 static void BindIndexBuffer(RenderBuffer *buffer);
 
-// Uniforms/PushConstants
+// ::Uniforms::Header:: ::PushConstants::Header::
 static void GetViewportSize(Vec2 *size);
 static void SetGlobalUniform(ShaderGlobals *globals);
 static void SetPushConstants(PushConst *pc);
+static DescHandle UploadImageUniform();
 
-// Config
+// ::Config::Header::
 static void SetRenderResolution(u32 x, u32 y);
+static void SetRendererAvailableThreads(u32 n);
 
-// Rendering
+// ::Rendering::Header::
 static b32 BeginFrame();
 static b32 FinishFrame();
 static void DrawIndexed(u32 index_count, u32 instance_count);

src/renderer_vulkan.c

@@ -702,19 +702,19 @@ static b32 VLayersSupported()
 
 static b32 CreateFrameStructures()
 {
-	b32 success = true;
+	b32 success           = true;
 	FrameStructures *data = &renderer.vk.frame;
-	u32 img_count = renderer.vk.sc.img_count;
+	u32 img_count         = renderer.vk.sc.img_count;
 
 	pool_create_info.queueFamilyIndex = renderer.vk.queues.graphics;
 
-	renderer.vk.frame.pools = ArenaAlloc(renderer.perm_arena, sizeof(VkCommandPool) * img_count);
+	renderer.vk.frame.pools                 = ArenaAlloc(renderer.perm_arena, sizeof(VkCommandPool) * img_count);
-	renderer.vk.frame.buffers = ArenaAlloc(renderer.perm_arena, sizeof(VkCommandBuffer) * img_count);
+	renderer.vk.frame.buffers               = ArenaAlloc(renderer.perm_arena, sizeof(VkCommandBuffer) * img_count);
-	renderer.vk.frame.swapchain_sems = ArenaAlloc(renderer.perm_arena, sizeof(VkSemaphore) * img_count);
+	renderer.vk.frame.swapchain_sems        = ArenaAlloc(renderer.perm_arena, sizeof(VkSemaphore) * img_count);
-	renderer.vk.frame.render_sems = ArenaAlloc(renderer.perm_arena, sizeof(VkSemaphore) * img_count);
+	renderer.vk.frame.render_sems           = ArenaAlloc(renderer.perm_arena, sizeof(VkSemaphore) * img_count);
-	renderer.vk.frame.render_fences = ArenaAlloc(renderer.perm_arena, sizeof(VkFence) * img_count);
+	renderer.vk.frame.render_fences         = ArenaAlloc(renderer.perm_arena, sizeof(VkFence) * img_count);
 	renderer.vk.frame.buffer_destroy_queues = ArenaAlloc(renderer.perm_arena, sizeof(RenderBuffer *) * FRAME_OVERLAP);
-	renderer.vk.frame.buffer_counts = ArenaAlloc(renderer.perm_arena, sizeof(u32) * FRAME_OVERLAP);
+	renderer.vk.frame.buffer_counts         = ArenaAlloc(renderer.perm_arena, sizeof(u32) * FRAME_OVERLAP);
 
 	for (u32 i = 0; i < FRAME_OVERLAP; i++)
 	{
@@ -751,25 +751,40 @@ static b32 CreateFrameStructures()
 
 static b32 CreateImmediateStructures()
 {
-	b32 success = true;
+	b32 success                       = true;
 	VkResult result;
-	VkDevice device = renderer.vk.device;
+	VkDevice device                   = renderer.vk.device;
-	ImmediateStructures *imm = &renderer.vk.imm;
+	ImmediateStructures *imm          = &renderer.vk.imm;
+	u8 thread_count                   = 1;
 	pool_create_info.queueFamilyIndex = renderer.vk.queues.transfer;
 	
-	result = vkCreateCommandPool(device, &pool_create_info, NULL, &imm->pool);
+	if (renderer.vk_conf.avail_threads >= 10)
-	if (result != VK_SUCCESS)
+		thread_count = 3;
-		success = false;
+	else if (renderer.vk_conf.avail_threads >= 8)
+		thread_count = 2;
 
-	cmd_buf_info.commandPool = imm->pool;
+	imm->pools          = ArenaAlloc(renderer.perm_arena, sizeof(VkCommandPool) * thread_count);
+	imm->cmds           = ArenaAlloc(renderer.perm_arena, sizeof(VkCommandBuffer) * thread_count);
+	imm->fences         = ArenaAlloc(renderer.perm_arena, sizeof(VkFence) * thread_count);
+	imm->queued_buffers = ArenaAlloc(renderer.perm_arena, sizeof(RenderBuffer) * BUFFER_QUEUE_LEN);
+	imm->data           = ArenaAlloc(renderer.perm_arena, sizeof(void *) * BUFFER_QUEUE_LEN);
 	
-	result = vkAllocateCommandBuffers(device, &cmd_buf_info, &imm->cmd);
+	for (u32 i = 0; i < thread_count && success; i++)
-	if (result != VK_SUCCESS)
+	{
-		success = false;
+		result = vkCreateCommandPool(device, &pool_create_info, NULL, &imm->pools[i]);
+		if (result != VK_SUCCESS)
+			success = false;
 
-	result = vkCreateFence(device, &fence_create_info, NULL, &imm->fence);
+		cmd_buf_info.commandPool = imm->pools[i];
-	if (result != VK_SUCCESS)
+
-		success = false;
+		result = vkAllocateCommandBuffers(device, &cmd_buf_info, &imm->cmds[i]);
+		if (result != VK_SUCCESS)
+			success = false;
+
+		result = vkCreateFence(device, &fence_create_info, NULL, &imm->fences[i]);
+		if (result != VK_SUCCESS)
+			success = false;
+	}
 
 	return success;
 }
@@ -997,6 +1012,7 @@ static b32 CreateDescriptors()
 	{
 		bindings[i].free = ArenaAlloc(renderer.perm_arena, sizeof(u16) * DESC_MAX_BINDINGS);
 		bindings[i].used = ArenaAlloc(renderer.perm_arena, sizeof(u16) * DESC_MAX_BINDINGS);
+		bindings[i].handle_indices = ArenaAlloc(renderer.perm_arena, sizeof(DescHandle) * DESC_MAX_BINDINGS);
 
 		u16 free_count = 0;
 		for (i32 j = DESC_MAX_BINDINGS-1; j >= 0; j--)
@@ -1115,6 +1131,54 @@ static b32 CreateShaderModule(u8 *bytes, u32 len, VkShaderModule *module)
 	return success;
 }
 
+#ifdef __linux__
+
+void *VkLoaderStart(void *i)
+{
+	u32 index = *(u32 *)i;
+	pthread_t self = pthread_self();
+
+	for (;;)
+	{
+		u32 job_count = __atomic_load_n(&renderer.vk_conf.job_count, __ATOMIC_RELEASE);
+		if (job_count < 0)
+		{
+			pthread_exit(NULL);
+		}
+
+		else if (job_count == 0)
+		{
+			__atomic_add_fetch(&renderer.vk_conf.sleeping_count, __ATOMIC_RELEASE);
+			pthread_suspend(self);
+		}
+
+		else if (__atomic_compare_exchange_n(&renderer.vk_conf.job_count, &job_count, job_count-1, false, __ATOMIC_RELEASE, __ATOMIC_RELAXED))
+		{
+			job_count -= 1;
+
+		}
+			
+	}
+}
+
+static b32 StartVkLoaderThreads()
+{
+	u32 count = renderer.vk_conf.avail_threads;
+	pthread_t *threads = ArenaAlloc(renderer.perm_arena, sizeof(pthread_t) * count);
+
+	for (u32 i = 0; i < count; i++)
+	{
+		vk_thread_indices[i] = i;
+		pthread_create(&threads[i], NULL, VkLoaderStart, &vk_thread_indices[i]);
+	}
+}
+
+#elif _WIN32
+
+#error not yet implemented
+
+#endif
+
 static void DestroySwapchain()
 {
 	for (u32 i = 0; i < renderer.vk.sc.img_count; i++) 

src/renderer_vulkan.h

@@ -16,7 +16,7 @@
 
 #include "file_data/spv.c"
 
-// Macros
+// ::Vulkan::Macros::Header::
 
 #define VK_DECLARE(fn) static PFN_##fn fn = NULL
 #define STYPE(type) VK_STRUCTURE_TYPE_ ## type
@@ -36,19 +36,14 @@
 	if (!name) return false;																					\
 } while (0)
 
-#define FRAME_OVERLAP 2
+// ::Vulkan::GlobalFunctions::Header::
-#define DESC_MAX_BINDINGS 256
 
-// Macros END
-
-// Vulkan Functions
-
-// Global
 VK_DECLARE(vkGetInstanceProcAddr);
 VK_DECLARE(vkCreateInstance);
 VK_DECLARE(vkEnumerateInstanceLayerProperties);
 
-// Instance
+// ::Vulkan::InstanceFunctions::Header::
+
 VK_DECLARE(vkEnumeratePhysicalDevices);
 VK_DECLARE(vkCreateDevice);
 VK_DECLARE(vkGetPhysicalDeviceQueueFamilyProperties);
@@ -61,6 +56,8 @@ VK_DECLARE(vkGetPhysicalDeviceSurfaceFormatsKHR);
 VK_DECLARE(vkGetPhysicalDeviceSurfaceCapabilitiesKHR);
 VK_DECLARE(vkGetPhysicalDeviceImageFormatProperties);
 
+// ::Vulkan::PlatformFunctions::Header::
+
 #ifdef __linux__
 VK_DECLARE(vkCreateXcbSurfaceKHR);
 #elif _WIN32
@@ -77,7 +74,8 @@ VK_DECLARE(vkCreateDebugUtilsMessengerEXT);
 VK_DECLARE(vkDestroyDebugUtilsMessengerEXT);
 #endif
 
-// Device
+// ::Vulkan::DeviceFunctions::Header::
+
 VK_DECLARE(vkGetDeviceProcAddr);
 VK_DECLARE(vkCreateSwapchainKHR);
 VK_DECLARE(vkCreateImage);
@@ -139,14 +137,16 @@ VK_DECLARE(vkCmdDraw);
 VK_DECLARE(vkDeviceWaitIdle);
 VK_DECLARE(vkCmdClearColorImage);
 
-// Vulkan Functions END
-
 #include "vma/vk_mem_alloc.h"
 
-// Defines
+// ::Vulkan::Defines::Header::
+
+#define FRAME_OVERLAP 2
+#define DESC_MAX_BINDINGS 256
+#define BUFFER_QUEUE_LEN 32
 #define HOST_VISIBLE_BUFFERS (RENDER_BUFFER_TYPE_UNIFORM | RENDER_BUFFER_TYPE_STAGING)
 
-// Types
+// ::Vulkan::Types::Header::
 
 typedef enum DescType_e
 {
@@ -165,18 +165,18 @@ typedef struct ShaderGlobals_t
 
 typedef struct RenderBuffer_t
 {
-	RenderBufferType type;
+	RenderBufferType  type;
-	VkBuffer buffer;
+	VkBuffer          buffer;
-	VmaAllocation alloc;
+	VmaAllocation     alloc;
 	VmaAllocationInfo info;
-	u32 size;
+	u32               size;
-	i32 mem_index; // TODO(MA): use this
+	i32               mem_index; // TODO(MA): use this
 } RenderBuffer;
 
 typedef struct
 {
 	RenderBuffer index_buf, vertex_buf;
-	u32 index_count;
+	u32          index_count;
 } MeshBuffer;
 
 typedef struct
@@ -186,21 +186,22 @@ typedef struct
 
 typedef struct 
 {
-	u16 *free;
+	u16        *free;
-	u16 free_count;
+	u16        free_count;
-	u16 *used;
+	u16        *used;
-	u16 used_count;
+	u16        used_count;
+	DescHandle *handle_indices;
 } DescBindings;
 
 typedef struct 
 {
-	VkPipelineLayout pipeline_layout;
+	VkPipelineLayout      pipeline_layout;
-	VkDescriptorPool pool;
+	VkDescriptorPool      pool;
 	VkDescriptorSetLayout layouts[DESC_TYPE_MAX];
-	VkDescriptorSet sets[DESC_TYPE_MAX];
+	VkDescriptorSet       sets[DESC_TYPE_MAX];
-	DescBindings *bindings;
+	DescBindings          *bindings;
-	u16 bindings_count;
+	u16                   bindings_count;
-	VkPipeline pipelines[PIPELINE_MAX];
+	VkPipeline            pipelines[PIPELINE_MAX];
 } PipelineStructures;
 
 typedef struct PushConst_t
@@ -210,70 +211,75 @@ typedef struct PushConst_t
 
 typedef struct 
 {
-	VkCommandPool *pools;
+	VkCommandPool   *pools;
 	VkCommandBuffer *buffers;
-	VkSemaphore *swapchain_sems;
+	VkSemaphore     *swapchain_sems;
-	VkSemaphore *render_sems;
+	VkSemaphore     *render_sems;
-	VkFence *render_fences;
+	VkFence         *render_fences;
-	RenderBuffer **buffer_destroy_queues;
+	RenderBuffer    **buffer_destroy_queues;
-	u32 *buffer_counts;
+	u32             *buffer_counts;
 } FrameStructures;
 
 typedef struct 
 {
-	VkCommandPool pool;
+	VkCommandPool   *pools;
-	VkCommandBuffer cmd;
+	VkCommandBuffer *cmds;
-	VkFence fence;
+	VkFence         *fences;
+	RenderBuffer 		**queued_buffers;
+	void 						*data;
+	i32 volatile 		job_count;
+	i32 volatile 		completed_count;
+	i32 volatile 		sleeping_count;
 } ImmediateStructures;
 
 typedef struct { 
-	i32 graphics, transfer; 
+	i32     graphics, transfer; 
 	VkQueue graphics_queue, transfer_queue;
 } DeviceQueues;
 
 typedef struct {
-	VkImage img;
+	VkImage       img;
-	VkImageView view;
+	VkImageView   view;
 	VmaAllocation alloc;
-	VkFormat fmt;
+	VkFormat      fmt;
 	VkImageLayout curr_layout;
 } Image;
 
 typedef struct {
-	VkFormat format;
+	VkFormat         format;
-	VkColorSpaceKHR color_space;
+	VkColorSpaceKHR  color_space;
 	VkPresentModeKHR present_mode;
-	VkExtent3D extent;
+	VkExtent3D       extent;
-	VkImage *imgs;
+	VkImage          *imgs;
-	VkImageView *views;
+	VkImageView      *views;
-	u32 img_count;
+	u32              img_count;
-	Image draw_img;
+	Image            draw_img;
-	Image depth_img;
+	Image            depth_img;
 } SwapchainStructures;
 
 typedef struct {
-	u32 img_ix;
+	u32            img_ix;
-	u64 frame_cnt;
+	u64            frame_cnt;
-	u64 prev_frame;
+	u64            prev_frame;
-	b8 begin_rendering;
+	b8             begin_rendering;
 	PipelineHandle last_pipeline;
-	RenderBuffer *prev_buffers;
+	RenderBuffer   *prev_buffers;
-	u32 prev_buffer_count;
+	u32            prev_buffer_count;
 } FrameState;
 
 typedef struct {
-	Library lib;
+	Library                  lib;
-	VkInstance inst;
+	VkInstance               inst;
-	VkSurfaceKHR surface;
+	VkSurfaceKHR             surface;
-	VkDevice device;
+	VkDevice                 device;
-	VkSwapchainKHR swapchain;
+	VkSwapchainKHR           swapchain;
-	VkPhysicalDevice phys_device;
+	VkPhysicalDevice         phys_device;
-	DeviceQueues queues;
+	DeviceQueues             queues;
-	VmaAllocator alloc;
+	VmaAllocator             alloc;
-	FrameStructures frame;
+	FrameStructures          frame;
-	ImmediateStructures imm;
+	ImmediateStructures      imm;
-	SwapchainStructures sc;
+	SwapchainStructures      sc;
-	PipelineStructures pipe;
+	PipelineStructures       pipe;
 #ifdef BUILD_DEBUG
 	VkDebugUtilsMessengerEXT debug;
 #endif
@@ -282,21 +288,33 @@ typedef struct {
 typedef struct {
 	u16 render_width;
 	u16 render_height;
-	b8 resized;
+	b8  resized;
 } PendingUpdates;
 
+typedef struct VulkanConfig_t
+{
+	u8        avail_threads;
+#ifdef __linux__
+	pthread_t *threads;
+#elif _WIN32
+	#error not yet implemented
+#endif
+} VulkanConfig;
+
 typedef struct Renderer_t 
 {
-	Vulkan_t vk;
+	Vulkan_t       vk;
-	FrameState frame_state;
+	VulkanConfig   vk_conf;
+	FrameState     frame_state;
 	PendingUpdates pending;
-	Arena *arena;
+	Arena          *arena;
-	Arena *perm_arena;
+	Arena          *perm_arena;
 } Renderer;
 
-// Renderer Function Declarations
+// ::Vulkan::Functions::Header::
+
+// ::Vulkan::Debug::Header::
 
-// Debug
 static b32 VLayersSupported();
 static VKAPI_ATTR VkBool32 DebugCallback(
 	VkDebugUtilsMessageSeverityFlagBitsEXT			message_severity,
@@ -306,7 +324,8 @@ static VKAPI_ATTR VkBool32 DebugCallback(
 );
 const char *VkResultStr(VkResult result);
 
-// Init
+// ::Vulkan::Init::Header::
+
 static b32 LoadVulkanLib();
 static b32 InitVkInstanceFunctions();
 static b32 InitVkGlobalFunctions();
@@ -326,8 +345,10 @@ static VkFormat GetImageFormat();
 static b32 CreateDescriptors();
 static b32 CreatePipelines();
 static b32 CreateShaderModule(u8 *bytes, u32 len, VkShaderModule *module);
+static b32 StartVkLoaderThreads();
+
+// ::Vulkan::Util::Header::
 
-// Util
 static inline VkCommandBuffer GetFrameCmdBuf();
 static inline VkFence *GetFrameRenderFence();
 static inline VkSemaphore GetFrameRenderSem();
@@ -338,36 +359,49 @@ static inline u32 *GetFrameBufferCount();
 static inline RenderBuffer *GetFrameRenderBuffers();
 static void BeginRendering();
 
-// Immediate Submit
+// ::Vulkan::Async::Header::
+#ifdef __linux__
+void *VkLoaderStart(void *thread_data);
+#elif _WIN32
+#error not yet implemented
+#endif
+
+// ::Vulkan::ImmediateSubmit::Header::
+
 static b32 BeginImmSubmit(VkDevice device, VkFence *fence, VkCommandBuffer cmd);
 static b32 FinishImmSubmit(VkDevice device, VkFence *fence, VkCommandBuffer cmd, VkQueue queue);
 
-// Buffers
+// ::Vulkan::Buffers::Header::
+
 static b32 UploadGUIBuffer(MeshBuffer *buf, GUIContext *ctx);
 
-// Destroy
+// ::Vulkan::Destroy::Header::
+
 static void DestroySwapchain();
 static void DestroyDrawImages();
 
-// Util
+// ::Vulkan::Util::Header::
+
 static void TransitionImage(VkCommandBuffer cmd, Image *img, VkImageLayout new);
 static void TransitionImageLayout(VkCommandBuffer cmd, VkImage img, VkImageLayout curr, VkImageLayout new);
 static void CopyImageToImage(VkCommandBuffer cmd, VkImage src, VkImage dst, VkExtent2D src_ext, VkExtent2D dst_ext);
 
-// Swapchain
+// ::Vulkan::Swapchain::Header::
+
 static VkExtent2D SelectSwapchainExtent(VkSurfaceCapabilitiesKHR *capabilities);
 static VkSurfaceFormatKHR SelectSwapchainFormat(VkSurfaceFormatKHR *formats);
 static void ResizeSwapchain();
 
-// Logging
+// ::Vulkan::Logging::Header::
+
 static void VkInfo(const char *str);
 static void VkWarn(const char *str);
 static void VkError(const char *str);
 
-// Renderer Functions Declarations END
-
 #include "vulkan_config.c"
 
+// ::Vulkan::Globals::Header::
+
 static Renderer renderer = {
 	.vk = {
 		.queues = {
@@ -382,6 +416,10 @@ static Renderer renderer = {
 	}
 };
 
+static u32 vk_thread_indices[10] = {};
+
+// ::Vulkan::Constants::Header::
+
 static const char *_VK_VALIDATION = "VK_LAYER_KHRONOS_validation";
 
 #if __linux__

src/renderer_vulkan_public.c

@@ -3,6 +3,8 @@
  */
 b32 InitRenderer(Arena *arena) 
 {
+	SetRendererAvailableThreads(AvailableCPUCount());
+
 	CustomizePipelines();
 
 	Assert(arena != NULL, "Vulkan memory is null");
@@ -39,6 +41,7 @@ b32 InitRenderer(Arena *arena)
 	Assert(CreateImmediateStructures(), "Unable to create immediate structures");
 	Assert(CreateDescriptors(), "Unable to initialize descriptors.");
 	Assert(CreatePipelines(), "Unable to initialize pipelines.");
+	Assert(StartVkLoaderThreads(), "Unable to initialize vulkan loader threads");
 
 	ArenaFree(renderer.arena);
 
@@ -74,9 +77,12 @@ void DestroyRenderer()
 
 	DestroySwapchain();
 
-	vkDestroyFence(device, imm.fence, NULL);
+	for (u32 i = 0; i < renderer.vk_conf.avail_threads; i++)
-	vkFreeCommandBuffers(device, imm.pool, 1, &imm.cmd);
+	{
-	vkDestroyCommandPool(device, imm.pool, NULL);
+		vkDestroyFence(device, imm.fences[i], NULL);
+		vkFreeCommandBuffers(device, imm.pools[i], 1, &imm.cmds[i]);
+		vkDestroyCommandPool(device, imm.pools[i], NULL);
+	}
 
 	for (u32 i = 0; i < FRAME_OVERLAP; i++)
 	{
@@ -196,14 +202,15 @@ static b32 CreateBuffer(RenderBuffer *buffer)
 	return success;
 }
 
-static b32 UploadToBuffer(RenderBuffer *buffer, rawptr ptr)
+static b32 UploadToBuffer(RenderBuffer *buffer, rawptr ptr, u8 thr_ix)
 {
 	Assert(buffer, "UploadToBuffer: buffer must not be null");
 	Assert(ptr, "UploadToBuffer: ptr must not be null");
 
 	b32 success = true;
 
-	ImmediateStructures *imm = &renderer.vk.imm;
+	VkCommandBuffer cmd = renderer.vk.imm.cmds[thr_ix];
+	VkFence fence = renderer.vk.imm.fences[thr_ix];
 	VkDevice device = renderer.vk.device;
 	VkQueue queue = renderer.vk.queues.transfer_queue;
 	VmaAllocator alloc = renderer.vk.alloc;
@@ -231,18 +238,18 @@ static b32 UploadToBuffer(RenderBuffer *buffer, rawptr ptr)
 		}
 
 		if (success)
-			success = BeginImmSubmit(device, &imm->fence, imm->cmd);
+			success = BeginImmSubmit(device, &fence, cmd);
 		
 		b32 imm_started = success;
 
 		if (success)
 		{
 			VkBufferCopy buffer_copy = { .size = (VkDeviceSize)buffer->size };
-			vkCmdCopyBuffer(imm->cmd, staging_buffer.buffer, buffer->buffer, 1, &buffer_copy);
+			vkCmdCopyBuffer(cmd, staging_buffer.buffer, buffer->buffer, 1, &buffer_copy);
 		}
 
 		if (imm_started)
-			FinishImmSubmit(device, &imm->fence, imm->cmd, queue);
+			FinishImmSubmit(device, &fence, cmd, queue);
 
 		if (buffer_created)
 		{
@@ -254,6 +261,17 @@ static b32 UploadToBuffer(RenderBuffer *buffer, rawptr ptr)
 	return success;
 }
 
+static void CreateAndUploadToBuffer(RenderBuffer *buffer, rawptr ptr)
+{
+	// TODO: revisit this to see if it could be done better
+
+	u32 i = __atomic_fetch_add(&renderer.vk.imm.job_count, 1, __ATOMIC_RELEASE);
+
+	renderer.vk.imm.queued_buffers[i] = buffer;
+	renderer.vk.imm.data[i] = ptr;
+}
+
+/* TODO: DELETE
 static b32 CreateAndUploadToBuffer(RenderBuffer *buffer, rawptr ptr)
 {
 	b32 success = true;
@@ -261,10 +279,11 @@ static b32 CreateAndUploadToBuffer(RenderBuffer *buffer, rawptr ptr)
 	success = CreateBuffer(buffer);
 	
 	if (success)
-		success = UploadToBuffer(buffer, ptr);
+		success = UploadToBuffer(buffer, ptr, 0); // TODO: DELETE this
 
 	return success;
 }
+*/
 
 static void FreeBuffers(RenderBuffer *buffers, u32 buffer_count)
 {
@@ -348,6 +367,11 @@ static void SetRenderResolution(u32 x, u32 y)
 	renderer.pending.resized = true;
 }
 
+static void SetRendererAvailableThreads(u32 n)
+{
+	renderer.vk_conf.avail_threads = n;
+}
+
 /*
  *	::Config::End::
  */