31_compute_shader.cpp

// Sample by Sascha Willems
// Contact: webmaster@saschawillems.de

#include <iostream>
#include <fstream>
#include <stdexcept>
#include <vector>
#include <cstring>
#include <cstdlib>
#include <memory>
#include <algorithm>
#include <limits>
#include <array>
#include <chrono>
#include <random>

#ifdef __INTELLISENSE__
#include <vulkan/vulkan_raii.hpp>
#else
import vulkan_hpp;
#endif

#include <vulkan/vk_platform.h>

#define GLFW_INCLUDE_VULKAN // REQUIRED only for GLFW CreateWindowSurface.
#include <GLFW/glfw3.h>

#define GLM_FORCE_RADIANS
#include <glm/glm.hpp>
#include <glm/gtc/matrix_transform.hpp>

constexpr uint32_t WIDTH = 800;
constexpr uint32_t HEIGHT = 600;
constexpr uint64_t FenceTimeout = 100000000;
constexpr uint32_t PARTICLE_COUNT = 8192;

constexpr int MAX_FRAMES_IN_FLIGHT = 2;

const std::vector validationLayers = {
    "VK_LAYER_KHRONOS_validation"
};

#ifdef NDEBUG
constexpr bool enableValidationLayers = false;
#else
constexpr bool enableValidationLayers = true;
#endif

struct UniformBufferObject {
    float deltaTime = 1.0f;
};

struct Particle {
    glm::vec2 position;
    glm::vec2 velocity;
    glm::vec4 color;

    static vk::VertexInputBindingDescription getBindingDescription() {
        return { 0, sizeof(Particle), vk::VertexInputRate::eVertex };
    }

    static std::array<vk::VertexInputAttributeDescription, 2> getAttributeDescriptions() {
        return {
            vk::VertexInputAttributeDescription( 0, 0, vk::Format::eR32G32Sfloat, offsetof(Particle, position) ),
            vk::VertexInputAttributeDescription( 1, 0, vk::Format::eR32G32B32A32Sfloat, offsetof(Particle, color) ),
        };
    }
};

class ComputeShaderApplication {
public:
    void run() {
        initWindow();
        initVulkan();
        mainLoop();
        cleanup();
    }

private:
    GLFWwindow* window = nullptr;

    vk::raii::Context  context;
    vk::raii::Instance instance = nullptr;
    vk::raii::DebugUtilsMessengerEXT debugMessenger = nullptr;
    vk::raii::SurfaceKHR surface = nullptr;

    vk::raii::PhysicalDevice physicalDevice = nullptr;
    vk::raii::Device device = nullptr;

    vk::raii::Queue graphicsQueue = nullptr;
    vk::raii::Queue computeQueue = nullptr;
    vk::raii::Queue presentQueue = nullptr;

    vk::raii::SwapchainKHR swapChain = nullptr;
    std::vector<vk::Image> swapChainImages;
    vk::SurfaceFormatKHR swapChainImageFormat;
    vk::Extent2D swapChainExtent;
    std::vector<vk::raii::ImageView> swapChainImageViews;

    vk::raii::PipelineLayout pipelineLayout = nullptr;
    vk::raii::Pipeline graphicsPipeline = nullptr;

    vk::raii::DescriptorSetLayout computeDescriptorSetLayout = nullptr;
    vk::raii::PipelineLayout computePipelineLayout = nullptr;
    vk::raii::Pipeline computePipeline = nullptr;


    std::vector<vk::raii::Buffer> shaderStorageBuffers;
    std::vector<vk::raii::DeviceMemory> shaderStorageBuffersMemory;

    std::vector<vk::raii::Buffer> uniformBuffers;
    std::vector<vk::raii::DeviceMemory> uniformBuffersMemory;
    std::vector<void*> uniformBuffersMapped;

    vk::raii::DescriptorPool descriptorPool = nullptr;
    std::vector<vk::raii::DescriptorSet> computeDescriptorSets;

    vk::raii::CommandPool commandPool = nullptr;
    std::vector<vk::raii::CommandBuffer> commandBuffers;
    std::vector<vk::raii::CommandBuffer> computeCommandBuffers;
    uint32_t graphicsAndComputeIndex = 0;

    vk::raii::Semaphore semaphore = nullptr;
    uint64_t timelineValue = 0;
    std::vector<vk::raii::Fence> inFlightFences;
    uint32_t currentFrame = 0;

    double lastFrameTime = 0.0;

    bool framebufferResized = false;

    double lastTime = 0.0f;

    std::vector<const char*> requiredDeviceExtension = {
        vk::KHRSwapchainExtensionName,
        vk::KHRSpirv14ExtensionName,
        vk::KHRSynchronization2ExtensionName,
        vk::KHRCreateRenderpass2ExtensionName
    };

    void initWindow() {
        glfwInit();

        glfwWindowHint(GLFW_CLIENT_API, GLFW_NO_API);

        window = glfwCreateWindow(WIDTH, HEIGHT, "Vulkan", nullptr, nullptr);
        glfwSetWindowUserPointer(window, this);
        glfwSetFramebufferSizeCallback(window, framebufferResizeCallback);

        lastTime = glfwGetTime();
    }

    static void framebufferResizeCallback(GLFWwindow* window, int width, int height) {
        auto app = static_cast<ComputeShaderApplication*>(glfwGetWindowUserPointer(window));
        app->framebufferResized = true;
    }

    void initVulkan() {
        createInstance();
        setupDebugMessenger();
        createSurface();
        pickPhysicalDevice();
        createLogicalDevice();
        createSwapChain();
        createImageViews();
        createComputeDescriptorSetLayout();
        createGraphicsPipeline();
        createComputePipeline();
        createCommandPool();
        createShaderStorageBuffers();
        createUniformBuffers();
        createDescriptorPool();
        createComputeDescriptorSets();
        createCommandBuffers();
        createComputeCommandBuffers();
        createSyncObjects();
    }

    void mainLoop() {
        while (!glfwWindowShouldClose(window)) {
            glfwPollEvents();
            drawFrame();
            // We want to animate the particle system using the last frames time to get smooth, frame-rate independent animation
            double currentTime = glfwGetTime();
            lastFrameTime = (currentTime - lastTime) * 1000.0;
            lastTime = currentTime;
        }

        device.waitIdle();
    }

    void cleanupSwapChain() {
        swapChainImageViews.clear();
    }

    void cleanup() const {
        glfwDestroyWindow(window);

        glfwTerminate();
    }

    void recreateSwapChain() {
        int width = 0, height = 0;
        glfwGetFramebufferSize(window, &width, &height);
        while (width == 0 || height == 0) {
            glfwGetFramebufferSize(window, &width, &height);
            glfwWaitEvents();
        }

        device.waitIdle();

        cleanupSwapChain();
        createSwapChain();
        createImageViews();
    }

    void createInstance() {
        constexpr vk::ApplicationInfo appInfo{ .pApplicationName   = "Hello Triangle",
                .applicationVersion = VK_MAKE_VERSION( 1, 0, 0 ),
                .pEngineName        = "No Engine",
                .engineVersion      = VK_MAKE_VERSION( 1, 0, 0 ),
                .apiVersion         = vk::ApiVersion14 };

        // Get the required layers
        std::vector<char const*> requiredLayers;
        if (enableValidationLayers) {
          requiredLayers.assign(validationLayers.begin(), validationLayers.end());
        }

        // Check if the required layers are supported by the Vulkan implementation.
        auto layerProperties = context.enumerateInstanceLayerProperties();
        for (auto const& requiredLayer : requiredLayers)
        {
            if (std::ranges::none_of(layerProperties,
                                     [requiredLayer](auto const& layerProperty)
                                     { return strcmp(layerProperty.layerName, requiredLayer) == 0; }))
            {
                throw std::runtime_error("Required layer not supported: " + std::string(requiredLayer));
            }
        }

        // Get the required extensions.
        auto requiredExtensions = getRequiredExtensions();

        // Check if the required extensions are supported by the Vulkan implementation.
        auto extensionProperties = context.enumerateInstanceExtensionProperties();
        for (auto const& requiredExtension : requiredExtensions)
        {
            if (std::ranges::none_of(extensionProperties,
                                     [requiredExtension](auto const& extensionProperty)
                                     { return strcmp(extensionProperty.extensionName, requiredExtension) == 0; }))
            {
                throw std::runtime_error("Required extension not supported: " + std::string(requiredExtension));
            }
        }

        vk::InstanceCreateInfo createInfo{
            .pApplicationInfo        = &appInfo,
            .enabledLayerCount       = static_cast<uint32_t>(requiredLayers.size()),
            .ppEnabledLayerNames     = requiredLayers.data(),
            .enabledExtensionCount   = static_cast<uint32_t>(requiredExtensions.size()),
            .ppEnabledExtensionNames = requiredExtensions.data() };
        instance = vk::raii::Instance(context, createInfo);
    }

    void setupDebugMessenger() {
        if (!enableValidationLayers) return;

        vk::DebugUtilsMessageSeverityFlagsEXT severityFlags( vk::DebugUtilsMessageSeverityFlagBitsEXT::eVerbose | vk::DebugUtilsMessageSeverityFlagBitsEXT::eWarning | vk::DebugUtilsMessageSeverityFlagBitsEXT::eError );
        vk::DebugUtilsMessageTypeFlagsEXT    messageTypeFlags( vk::DebugUtilsMessageTypeFlagBitsEXT::eGeneral | vk::DebugUtilsMessageTypeFlagBitsEXT::ePerformance | vk::DebugUtilsMessageTypeFlagBitsEXT::eValidation );
        vk::DebugUtilsMessengerCreateInfoEXT debugUtilsMessengerCreateInfoEXT{
            .messageSeverity = severityFlags,
            .messageType = messageTypeFlags,
            .pfnUserCallback = &debugCallback
        };
        debugMessenger = instance.createDebugUtilsMessengerEXT(debugUtilsMessengerCreateInfoEXT);
    }

    void createSurface() {
        VkSurfaceKHR       _surface;
        if (glfwCreateWindowSurface(*instance, window, nullptr, &_surface) != 0) {
            throw std::runtime_error("failed to create window surface!");
        }
        surface = vk::raii::SurfaceKHR(instance, _surface);
    }

    void pickPhysicalDevice() {
        std::vector<vk::raii::PhysicalDevice> devices = instance.enumeratePhysicalDevices();
        const auto                            devIter = std::ranges::find_if(
          devices,
          [&]( auto const & device )
          {
            // Check if the device supports the Vulkan 1.3 API version
            bool supportsVulkan1_3 = device.getProperties().apiVersion >= VK_API_VERSION_1_3;

            // Check if any of the queue families support graphics operations
                auto queueFamilies = device.getQueueFamilyProperties();
            bool supportsGraphics =
              std::ranges::any_of( queueFamilies, []( auto const & qfp ) { return !!( qfp.queueFlags & vk::QueueFlagBits::eGraphics ); } );

            // Check if all required device extensions are available
            auto availableDeviceExtensions = device.enumerateDeviceExtensionProperties();
            bool supportsAllRequiredExtensions =
              std::ranges::all_of( requiredDeviceExtension,
                                   [&availableDeviceExtensions]( auto const & requiredDeviceExtension )
                        {
                                     return std::ranges::any_of( availableDeviceExtensions,
                                                                 [requiredDeviceExtension]( auto const & availableDeviceExtension )
                                                                 { return strcmp( availableDeviceExtension.extensionName, requiredDeviceExtension ) == 0; } );
                        } );

            auto features = device.template getFeatures2<vk::PhysicalDeviceFeatures2,
                                                         vk::PhysicalDeviceVulkan13Features,
                                                         vk::PhysicalDeviceExtendedDynamicStateFeaturesEXT,
                                                         vk::PhysicalDeviceTimelineSemaphoreFeaturesKHR>();
            bool supportsRequiredFeatures = features.template get<vk::PhysicalDeviceFeatures2>().features.samplerAnisotropy &&
                                            features.template get<vk::PhysicalDeviceVulkan13Features>().dynamicRendering &&
                                            features.template get<vk::PhysicalDeviceExtendedDynamicStateFeaturesEXT>().extendedDynamicState &&
                                            features.template get<vk::PhysicalDeviceTimelineSemaphoreFeaturesKHR>().timelineSemaphore;

            return supportsVulkan1_3 && supportsGraphics && supportsAllRequiredExtensions && supportsRequiredFeatures;
        });
        if ( devIter != devices.end() )
        {
            physicalDevice = *devIter;
        }
        else
        {
            throw std::runtime_error("failed to find a suitable GPU!");
        }
    }

    void createLogicalDevice() {
        // find the index of the first queue family that supports graphics
        std::vector<vk::QueueFamilyProperties> queueFamilyProperties = physicalDevice.getQueueFamilyProperties();

        // get the first index into queueFamilyProperties which supports graphics and compute
        auto graphicsAndComputeQueueFamilyProperty = std::ranges::find_if(queueFamilyProperties, []( auto const & qfp )
            { return (qfp.queueFlags & vk::QueueFlagBits::eGraphics && qfp.queueFlags & vk::QueueFlagBits::eCompute); } );

        graphicsAndComputeIndex = static_cast<uint32_t>( std::distance( queueFamilyProperties.begin(), graphicsAndComputeQueueFamilyProperty ) );

        // determine a queueFamilyIndex that supports present
        // first check if the graphicsIndex is good enough
        auto presentIndex = physicalDevice.getSurfaceSupportKHR( graphicsAndComputeIndex, surface )
                                           ? graphicsAndComputeIndex
                                           : ~0;
        if ( presentIndex == queueFamilyProperties.size() )
        {
            // the graphicsIndex doesn't support present -> look for another family index that supports both
            // graphics and present
            for ( size_t i = 0; i < queueFamilyProperties.size(); i++ )
            {
                if ( ( (queueFamilyProperties[i].queueFlags & vk::QueueFlagBits::eGraphics && queueFamilyProperties[i].queueFlags & vk::QueueFlagBits::eCompute) ) &&
                     physicalDevice.getSurfaceSupportKHR( static_cast<uint32_t>( i ), surface ) )
                {
                    graphicsAndComputeIndex = static_cast<uint32_t>( i );
                    presentIndex  = graphicsAndComputeIndex;
                    break;
                }
            }
            if ( presentIndex == queueFamilyProperties.size() )
            {
                // there's nothing like a single family index that supports both graphics and present -> look for another
                // family index that supports present
                for ( size_t i = 0; i < queueFamilyProperties.size(); i++ )
                {
                    if ( physicalDevice.getSurfaceSupportKHR( static_cast<uint32_t>( i ), surface ) )
                    {
                        presentIndex = static_cast<uint32_t>( i );
                        break;
                    }
                }
            }
        }
        if ( ( graphicsAndComputeIndex == queueFamilyProperties.size() ) || ( presentIndex == queueFamilyProperties.size() ) )
        {
            throw std::runtime_error( "Could not find a queue for graphics or present -> terminating" );
        }

        // query for Vulkan 1.3 features
        vk::StructureChain<vk::PhysicalDeviceFeatures2,
                           vk::PhysicalDeviceVulkan13Features,
                           vk::PhysicalDeviceExtendedDynamicStateFeaturesEXT,
                           vk::PhysicalDeviceTimelineSemaphoreFeaturesKHR>
          featureChain = {
            {.features = {.samplerAnisotropy = true } },           // vk::PhysicalDeviceFeatures2
            {.synchronization2 = true, .dynamicRendering = true },  // vk::PhysicalDeviceVulkan13Features
            {.extendedDynamicState = true },                        // vk::PhysicalDeviceExtendedDynamicStateFeaturesEXT
            {.timelineSemaphore = true }                            // vk::PhysicalDeviceTimelineSemaphoreFeaturesKHR
        };

        // create a Device
        float                     queuePriority = 0.0f;
        vk::DeviceQueueCreateInfo deviceQueueCreateInfo{ .queueFamilyIndex = graphicsAndComputeIndex, .queueCount = 1, .pQueuePriorities = &queuePriority };
        vk::DeviceCreateInfo      deviceCreateInfo{ .pNext = &featureChain.get<vk::PhysicalDeviceFeatures2>(),
                                                    .queueCreateInfoCount = 1,
                                                    .pQueueCreateInfos = &deviceQueueCreateInfo,
                                                    .enabledExtensionCount = static_cast<uint32_t>(requiredDeviceExtension.size()),
                                                    .ppEnabledExtensionNames = requiredDeviceExtension.data() };

        device = vk::raii::Device( physicalDevice, deviceCreateInfo );
        graphicsQueue = vk::raii::Queue( device, graphicsAndComputeIndex, 0 );
        computeQueue = vk::raii::Queue( device, graphicsAndComputeIndex, 0 );
        presentQueue = vk::raii::Queue( device, presentIndex, 0 );
    }

    void createSwapChain() {
        auto surfaceCapabilities = physicalDevice.getSurfaceCapabilitiesKHR( surface );
        swapChainImageFormat = chooseSwapSurfaceFormat(physicalDevice.getSurfaceFormatsKHR( surface ));
        swapChainExtent = chooseSwapExtent(surfaceCapabilities);
        auto minImageCount = std::max( 3u, surfaceCapabilities.minImageCount );
        minImageCount = ( surfaceCapabilities.maxImageCount > 0 && minImageCount > surfaceCapabilities.maxImageCount ) ? surfaceCapabilities.maxImageCount : minImageCount;
        vk::SwapchainCreateInfoKHR swapChainCreateInfo{
            .flags = vk::SwapchainCreateFlagsKHR(),
            .surface = surface, .minImageCount = minImageCount,
            .imageFormat = swapChainImageFormat.format, .imageColorSpace = swapChainImageFormat.colorSpace,
            .imageExtent = swapChainExtent, .imageArrayLayers =1,
            .imageUsage = vk::ImageUsageFlagBits::eColorAttachment, .imageSharingMode = vk::SharingMode::eExclusive,
            .preTransform = surfaceCapabilities.currentTransform, .compositeAlpha = vk::CompositeAlphaFlagBitsKHR::eOpaque,
            .presentMode = chooseSwapPresentMode(physicalDevice.getSurfacePresentModesKHR(surface)),
            .clipped = true, .oldSwapchain = nullptr };

        swapChain = vk::raii::SwapchainKHR( device, swapChainCreateInfo );
        swapChainImages = swapChain.getImages();
    }

    void createImageViews() {
        vk::ImageViewCreateInfo imageViewCreateInfo{
            .viewType = vk::ImageViewType::e2D,
            .format = swapChainImageFormat.format,
            .components = {vk::ComponentSwizzle::eIdentity, vk::ComponentSwizzle::eIdentity, vk::ComponentSwizzle::eIdentity, vk::ComponentSwizzle::eIdentity},
            .subresourceRange = { vk::ImageAspectFlagBits::eColor, 0, 1, 0, 1 }
        };
        for ( auto image : swapChainImages )
        {
            imageViewCreateInfo.image = image;
            swapChainImageViews.emplace_back( device, imageViewCreateInfo );
        }
    }

    void createComputeDescriptorSetLayout() {
        std::array layoutBindings{
            vk::DescriptorSetLayoutBinding(0, vk::DescriptorType::eUniformBuffer, 1, vk::ShaderStageFlagBits::eCompute, nullptr),
            vk::DescriptorSetLayoutBinding(1, vk::DescriptorType::eStorageBuffer, 1, vk::ShaderStageFlagBits::eCompute, nullptr),
            vk::DescriptorSetLayoutBinding(2, vk::DescriptorType::eStorageBuffer, 1, vk::ShaderStageFlagBits::eCompute, nullptr)
        };

        vk::DescriptorSetLayoutCreateInfo layoutInfo{ .bindingCount = static_cast<uint32_t>(layoutBindings.size()), .pBindings = layoutBindings.data() };
        computeDescriptorSetLayout = vk::raii::DescriptorSetLayout( device, layoutInfo );
    }


    void createGraphicsPipeline() {
        vk::raii::ShaderModule shaderModule = createShaderModule(readFile("shaders/slang.spv"));

        vk::PipelineShaderStageCreateInfo vertShaderStageInfo{ .stage = vk::ShaderStageFlagBits::eVertex, .module = shaderModule, .pName = "vertMain" };
        vk::PipelineShaderStageCreateInfo fragShaderStageInfo{ .stage = vk::ShaderStageFlagBits::eFragment, .module = shaderModule, .pName = "fragMain" };
        vk::PipelineShaderStageCreateInfo shaderStages[] = {vertShaderStageInfo, fragShaderStageInfo};

        auto bindingDescription = Particle::getBindingDescription();
        auto attributeDescriptions = Particle::getAttributeDescriptions();

        vk::PipelineVertexInputStateCreateInfo vertexInputInfo{ .vertexBindingDescriptionCount = 1, .pVertexBindingDescriptions = &bindingDescription, .vertexAttributeDescriptionCount = static_cast<uint32_t>(attributeDescriptions.size()), .pVertexAttributeDescriptions = attributeDescriptions.data() };
        vk::PipelineInputAssemblyStateCreateInfo inputAssembly{ .topology = vk::PrimitiveTopology::ePointList, .primitiveRestartEnable = vk::False };
        vk::PipelineViewportStateCreateInfo viewportState{ .viewportCount = 1, .scissorCount = 1 };
        vk::PipelineRasterizationStateCreateInfo rasterizer{
            .depthClampEnable = vk::False,
            .rasterizerDiscardEnable = vk::False,
            .polygonMode = vk::PolygonMode::eFill,
            .cullMode = vk::CullModeFlagBits::eBack,
            .frontFace = vk::FrontFace::eCounterClockwise,
            .depthBiasEnable = vk::False,
            .lineWidth = 1.0f
        };
        vk::PipelineMultisampleStateCreateInfo multisampling{ .rasterizationSamples = vk::SampleCountFlagBits::e1, .sampleShadingEnable = vk::False };

        vk::PipelineColorBlendAttachmentState colorBlendAttachment{
            .blendEnable = vk::True,
            .srcColorBlendFactor = vk::BlendFactor::eSrcAlpha,
            .dstColorBlendFactor = vk::BlendFactor::eOneMinusSrcAlpha,
            .colorBlendOp = vk::BlendOp::eAdd,
            .srcAlphaBlendFactor = vk::BlendFactor::eOneMinusSrcAlpha,
            .dstAlphaBlendFactor = vk::BlendFactor::eZero,
            .alphaBlendOp = vk::BlendOp::eAdd,
            .colorWriteMask = vk::ColorComponentFlagBits::eR | vk::ColorComponentFlagBits::eG | vk::ColorComponentFlagBits::eB | vk::ColorComponentFlagBits::eA,
        };

        vk::PipelineColorBlendStateCreateInfo colorBlending{ .logicOpEnable = vk::False, .logicOp = vk::LogicOp::eCopy, .attachmentCount = 1, .pAttachments = &colorBlendAttachment };

        std::vector dynamicStates = {
            vk::DynamicState::eViewport,
            vk::DynamicState::eScissor
        };
        vk::PipelineDynamicStateCreateInfo dynamicState{ .dynamicStateCount = static_cast<uint32_t>(dynamicStates.size()), .pDynamicStates = dynamicStates.data() };

        vk::PipelineLayoutCreateInfo pipelineLayoutInfo;
        pipelineLayout = vk::raii::PipelineLayout( device, pipelineLayoutInfo );

        vk::PipelineRenderingCreateInfo pipelineRenderingCreateInfo{ .colorAttachmentCount = 1, .pColorAttachmentFormats = &swapChainImageFormat.format };
        vk::GraphicsPipelineCreateInfo pipelineInfo{ .pNext = &pipelineRenderingCreateInfo,
            .stageCount = 2,
            .pStages = shaderStages,
            .pVertexInputState = &vertexInputInfo,
            .pInputAssemblyState = &inputAssembly,
            .pViewportState = &viewportState,
            .pRasterizationState = &rasterizer,
            .pMultisampleState = &multisampling,
            .pColorBlendState = &colorBlending,
            .pDynamicState = &dynamicState,
            .layout = *pipelineLayout,
            .subpass = 0
        };

        graphicsPipeline = vk::raii::Pipeline(device, nullptr, pipelineInfo);
    }

    void createComputePipeline() {
        vk::raii::ShaderModule shaderModule = createShaderModule(readFile("shaders/slang.spv"));

        vk::PipelineShaderStageCreateInfo computeShaderStageInfo{ .stage = vk::ShaderStageFlagBits::eCompute, .module = shaderModule, .pName = "compMain" };
        vk::PipelineLayoutCreateInfo pipelineLayoutInfo{ .setLayoutCount = 1, .pSetLayouts = &*computeDescriptorSetLayout };
        computePipelineLayout = vk::raii::PipelineLayout( device, pipelineLayoutInfo );
        vk::ComputePipelineCreateInfo pipelineInfo{ .stage = computeShaderStageInfo, .layout = *computePipelineLayout };
        computePipeline = vk::raii::Pipeline(device, nullptr, pipelineInfo);
    }

    void createCommandPool() {
        vk::CommandPoolCreateInfo poolInfo{};
        poolInfo.flags = vk::CommandPoolCreateFlagBits::eResetCommandBuffer;
        poolInfo.queueFamilyIndex = graphicsAndComputeIndex;
        commandPool = vk::raii::CommandPool(device, poolInfo);
    }

    void createShaderStorageBuffers() {
        // Initialize particles
        std::default_random_engine rndEngine(static_cast<unsigned>(time(nullptr)));
        std::uniform_real_distribution rndDist(0.0f, 1.0f);

        // Initial particle positions on a circle
        std::vector<Particle> particles(PARTICLE_COUNT);
        for (auto& particle : particles) {
            float r = 0.25f * sqrtf(rndDist(rndEngine));
            float theta = rndDist(rndEngine) * 2.0f * 3.14159265358979323846f;
            float x = r * cosf(theta) * HEIGHT / WIDTH;
            float y = r * sinf(theta);
            particle.position = glm::vec2(x, y);
            particle.velocity = normalize(glm::vec2(x,y)) * 0.00025f;
            particle.color = glm::vec4(rndDist(rndEngine), rndDist(rndEngine), rndDist(rndEngine), 1.0f);
        }

        vk::DeviceSize bufferSize = sizeof(Particle) * PARTICLE_COUNT;

        // Create a staging buffer used to upload data to the gpu
        vk::raii::Buffer stagingBuffer({});
        vk::raii::DeviceMemory stagingBufferMemory({});
        createBuffer(bufferSize, vk::BufferUsageFlagBits::eTransferSrc, vk::MemoryPropertyFlagBits::eHostVisible | vk::MemoryPropertyFlagBits::eHostCoherent, stagingBuffer, stagingBufferMemory);

        void* dataStaging = stagingBufferMemory.mapMemory(0, bufferSize);
        memcpy(dataStaging, particles.data(), (size_t)bufferSize);
        stagingBufferMemory.unmapMemory();

        shaderStorageBuffers.clear();
        shaderStorageBuffersMemory.clear();

        // Copy initial particle data to all storage buffers
        for (size_t i = 0; i < MAX_FRAMES_IN_FLIGHT; i++) {
            vk::raii::Buffer shaderStorageBufferTemp({});
            vk::raii::DeviceMemory shaderStorageBufferTempMemory({});
            createBuffer(bufferSize, vk::BufferUsageFlagBits::eStorageBuffer | vk::BufferUsageFlagBits::eVertexBuffer | vk::BufferUsageFlagBits::eTransferDst, vk::MemoryPropertyFlagBits::eDeviceLocal, shaderStorageBufferTemp, shaderStorageBufferTempMemory);
            copyBuffer(stagingBuffer, shaderStorageBufferTemp, bufferSize);
            shaderStorageBuffers.emplace_back(std::move(shaderStorageBufferTemp));
            shaderStorageBuffersMemory.emplace_back(std::move(shaderStorageBufferTempMemory));
        }
    }

    void createUniformBuffers() {
        uniformBuffers.clear();
        uniformBuffersMemory.clear();
        uniformBuffersMapped.clear();

        for (size_t i = 0; i < MAX_FRAMES_IN_FLIGHT; i++) {
            vk::DeviceSize bufferSize = sizeof(UniformBufferObject);
            vk::raii::Buffer buffer({});
            vk::raii::DeviceMemory bufferMem({});
            createBuffer(bufferSize, vk::BufferUsageFlagBits::eUniformBuffer, vk::MemoryPropertyFlagBits::eHostVisible | vk::MemoryPropertyFlagBits::eHostCoherent, buffer, bufferMem);
            uniformBuffers.emplace_back(std::move(buffer));
            uniformBuffersMemory.emplace_back(std::move(bufferMem));
            uniformBuffersMapped.emplace_back( uniformBuffersMemory[i].mapMemory(0, bufferSize));
        }
    }

    void createDescriptorPool() {
        std::array poolSize {
            vk::DescriptorPoolSize( vk::DescriptorType::eUniformBuffer, MAX_FRAMES_IN_FLIGHT),
            vk::DescriptorPoolSize(  vk::DescriptorType::eStorageBuffer, MAX_FRAMES_IN_FLIGHT * 2)
        };
        vk::DescriptorPoolCreateInfo poolInfo{};
        poolInfo.flags = vk::DescriptorPoolCreateFlagBits::eFreeDescriptorSet;
        poolInfo.maxSets = MAX_FRAMES_IN_FLIGHT;
        poolInfo.poolSizeCount = poolSize.size();
        poolInfo.pPoolSizes = poolSize.data();
        descriptorPool = vk::raii::DescriptorPool(device, poolInfo);
    }

    void createComputeDescriptorSets() {
        std::vector<vk::DescriptorSetLayout> layouts(MAX_FRAMES_IN_FLIGHT, computeDescriptorSetLayout);
        vk::DescriptorSetAllocateInfo allocInfo{};
        allocInfo.descriptorPool = *descriptorPool;
        allocInfo.descriptorSetCount = MAX_FRAMES_IN_FLIGHT;
        allocInfo.pSetLayouts = layouts.data();
        computeDescriptorSets.clear();
        computeDescriptorSets = device.allocateDescriptorSets(allocInfo);

        for (size_t i = 0; i < MAX_FRAMES_IN_FLIGHT; i++) {
            vk::DescriptorBufferInfo bufferInfo(uniformBuffers[i], 0, sizeof(UniformBufferObject));

            vk::DescriptorBufferInfo storageBufferInfoLastFrame(shaderStorageBuffers[(i - 1) % MAX_FRAMES_IN_FLIGHT], 0, sizeof(Particle) * PARTICLE_COUNT);
            vk::DescriptorBufferInfo storageBufferInfoCurrentFrame(shaderStorageBuffers[i], 0, sizeof(Particle) * PARTICLE_COUNT);
            std::array descriptorWrites{
                vk::WriteDescriptorSet{ .dstSet = *computeDescriptorSets[i], .dstBinding = 0, .dstArrayElement = 0, .descriptorCount = 1, .descriptorType = vk::DescriptorType::eUniformBuffer, .pImageInfo = nullptr, .pBufferInfo = &bufferInfo, .pTexelBufferView = nullptr },
                vk::WriteDescriptorSet{ .dstSet = *computeDescriptorSets[i], .dstBinding = 1, .dstArrayElement = 0, .descriptorCount = 1, .descriptorType = vk::DescriptorType::eStorageBuffer, .pImageInfo = nullptr, .pBufferInfo = &storageBufferInfoLastFrame, .pTexelBufferView = nullptr },
                vk::WriteDescriptorSet{ .dstSet = *computeDescriptorSets[i], .dstBinding = 2, .dstArrayElement = 0, .descriptorCount = 1, .descriptorType = vk::DescriptorType::eStorageBuffer, .pImageInfo = nullptr, .pBufferInfo = &storageBufferInfoCurrentFrame, .pTexelBufferView = nullptr },
            };
            device.updateDescriptorSets(descriptorWrites, {});
        }
    }


    void createBuffer(vk::DeviceSize size, vk::BufferUsageFlags usage, vk::MemoryPropertyFlags properties, vk::raii::Buffer& buffer, vk::raii::DeviceMemory& bufferMemory) const {
        vk::BufferCreateInfo bufferInfo{};
        bufferInfo.size = size;
        bufferInfo.usage = usage;
        bufferInfo.sharingMode = vk::SharingMode::eExclusive;
        buffer = vk::raii::Buffer(device, bufferInfo);
        vk::MemoryRequirements memRequirements = buffer.getMemoryRequirements();
        vk::MemoryAllocateInfo allocInfo{};
        allocInfo.allocationSize = memRequirements.size;
        allocInfo.memoryTypeIndex = findMemoryType(memRequirements.memoryTypeBits, properties);
        bufferMemory = vk::raii::DeviceMemory(device, allocInfo);
        buffer.bindMemory(bufferMemory, 0);
    }

    [[nodiscard]] vk::raii::CommandBuffer beginSingleTimeCommands() const {
        vk::CommandBufferAllocateInfo allocInfo{};
        allocInfo.commandPool = *commandPool;
        allocInfo.level = vk::CommandBufferLevel::ePrimary;
        allocInfo.commandBufferCount = 1;
        vk::raii::CommandBuffer commandBuffer = std::move(vk::raii::CommandBuffers( device, allocInfo ).front());

        vk::CommandBufferBeginInfo beginInfo{ .flags = vk::CommandBufferUsageFlagBits::eOneTimeSubmit };
        commandBuffer.begin(beginInfo);

        return commandBuffer;
    }

    void endSingleTimeCommands(const vk::raii::CommandBuffer& commandBuffer) const {
        commandBuffer.end();

        vk::SubmitInfo submitInfo{};
        submitInfo.commandBufferCount = 1;
        submitInfo.pCommandBuffers = &*commandBuffer;
        graphicsQueue.submit(submitInfo, nullptr);
        graphicsQueue.waitIdle();
    }

    void copyBuffer(const vk::raii::Buffer & srcBuffer, const vk::raii::Buffer & dstBuffer, vk::DeviceSize size) const {
        vk::raii::CommandBuffer commandCopyBuffer = beginSingleTimeCommands();
        commandCopyBuffer.copyBuffer(srcBuffer, dstBuffer, vk::BufferCopy(0, 0, size));
        endSingleTimeCommands(commandCopyBuffer);
    }

    [[nodiscard]] uint32_t findMemoryType(uint32_t typeFilter, vk::MemoryPropertyFlags properties) const {
        vk::PhysicalDeviceMemoryProperties memProperties = physicalDevice.getMemoryProperties();

        for (uint32_t i = 0; i < memProperties.memoryTypeCount; i++) {
            if ((typeFilter & (1 << i)) && (memProperties.memoryTypes[i].propertyFlags & properties) == properties) {
                return i;
            }
        }

        throw std::runtime_error("failed to find suitable memory type!");
    }

    void createCommandBuffers() {
        commandBuffers.clear();
        vk::CommandBufferAllocateInfo allocInfo{};
        allocInfo.commandPool = *commandPool;
        allocInfo.level = vk::CommandBufferLevel::ePrimary;
        allocInfo.commandBufferCount = MAX_FRAMES_IN_FLIGHT;
        commandBuffers = vk::raii::CommandBuffers(device, allocInfo);
    }

    void createComputeCommandBuffers() {
        computeCommandBuffers.clear();
        vk::CommandBufferAllocateInfo allocInfo{};
        allocInfo.commandPool = *commandPool;
        allocInfo.level = vk::CommandBufferLevel::ePrimary;
        allocInfo.commandBufferCount = MAX_FRAMES_IN_FLIGHT;
        computeCommandBuffers = vk::raii::CommandBuffers(device, allocInfo);
    }

    void recordCommandBuffer( uint32_t imageIndex) {
        commandBuffers[currentFrame].reset();
        commandBuffers[currentFrame].begin( {} );
        // Before starting rendering, transition the swapchain image to COLOR_ATTACHMENT_OPTIMAL
        transition_image_layout(
            imageIndex,
            vk::ImageLayout::eUndefined,
            vk::ImageLayout::eColorAttachmentOptimal,
            {},                                                     // srcAccessMask (no need to wait for previous operations)
            vk::AccessFlagBits2::eColorAttachmentWrite,                // dstAccessMask
            vk::PipelineStageFlagBits2::eTopOfPipe,                   // srcStage
            vk::PipelineStageFlagBits2::eColorAttachmentOutput        // dstStage
        );
        vk::ClearValue clearColor = vk::ClearColorValue(0.0f, 0.0f, 0.0f, 1.0f);
        vk::RenderingAttachmentInfo attachmentInfo = {
            .imageView = swapChainImageViews[imageIndex],
            .imageLayout = vk::ImageLayout::eColorAttachmentOptimal,
            .loadOp = vk::AttachmentLoadOp::eClear,
            .storeOp = vk::AttachmentStoreOp::eStore,
            .clearValue = clearColor
        };
        vk::RenderingInfo renderingInfo = {
            .renderArea = { .offset = { 0, 0 }, .extent = swapChainExtent },
            .layerCount = 1,
            .colorAttachmentCount = 1,
            .pColorAttachments = &attachmentInfo
        };
        commandBuffers[currentFrame].beginRendering(renderingInfo);
        commandBuffers[currentFrame].bindPipeline(vk::PipelineBindPoint::eGraphics, *graphicsPipeline);
        commandBuffers[currentFrame].setViewport(0, vk::Viewport(0.0f, 0.0f, static_cast<float>(swapChainExtent.width), static_cast<float>(swapChainExtent.height), 0.0f, 1.0f));
        commandBuffers[currentFrame].setScissor( 0, vk::Rect2D( vk::Offset2D( 0, 0 ), swapChainExtent ) );
        commandBuffers[currentFrame].bindVertexBuffers(0, { shaderStorageBuffers[currentFrame] }, {0});
        commandBuffers[currentFrame].draw( PARTICLE_COUNT, 1, 0, 0 );
        commandBuffers[currentFrame].endRendering();
        // After rendering, transition the swapchain image to PRESENT_SRC
        transition_image_layout(
            imageIndex,
            vk::ImageLayout::eColorAttachmentOptimal,
            vk::ImageLayout::ePresentSrcKHR,
            vk::AccessFlagBits2::eColorAttachmentWrite,                 // srcAccessMask
            {},                                                      // dstAccessMask
            vk::PipelineStageFlagBits2::eColorAttachmentOutput,        // srcStage
            vk::PipelineStageFlagBits2::eBottomOfPipe                  // dstStage
        );
        commandBuffers[currentFrame].end();
    }

        void transition_image_layout(
                uint32_t imageIndex,
                vk::ImageLayout old_layout,
                vk::ImageLayout new_layout,
                vk::AccessFlags2 src_access_mask,
                vk::AccessFlags2 dst_access_mask,
                vk::PipelineStageFlags2 src_stage_mask,
                vk::PipelineStageFlags2 dst_stage_mask
                ) {
            vk::ImageMemoryBarrier2 barrier = {
                .srcStageMask = src_stage_mask,
                .srcAccessMask = src_access_mask,
                .dstStageMask = dst_stage_mask,
                .dstAccessMask = dst_access_mask,
                .oldLayout = old_layout,
                .newLayout = new_layout,
                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .image = swapChainImages[imageIndex],
                .subresourceRange = {
                    .aspectMask = vk::ImageAspectFlagBits::eColor,
                    .baseMipLevel = 0,
                    .levelCount = 1,
                    .baseArrayLayer = 0,
                    .layerCount = 1
                }
            };
            vk::DependencyInfo dependency_info = {
                .dependencyFlags = {},
                .imageMemoryBarrierCount = 1,
                .pImageMemoryBarriers = &barrier
            };
            commandBuffers[currentFrame].pipelineBarrier2(dependency_info);
        }

    void recordComputeCommandBuffer() {
        computeCommandBuffers[currentFrame].reset();
        computeCommandBuffers[currentFrame].begin({});
        computeCommandBuffers[currentFrame].bindPipeline(vk::PipelineBindPoint::eCompute, computePipeline);
        computeCommandBuffers[currentFrame].bindDescriptorSets(vk::PipelineBindPoint::eCompute, computePipelineLayout, 0, {computeDescriptorSets[currentFrame]}, {});
        computeCommandBuffers[currentFrame].dispatch( PARTICLE_COUNT / 256, 1, 1 );
        computeCommandBuffers[currentFrame].end();
    }

    void createSyncObjects() {
        inFlightFences.clear();

        vk::SemaphoreTypeCreateInfo semaphoreType{ .semaphoreType = vk::SemaphoreType::eTimeline, .initialValue = 0 };
        semaphore = vk::raii::Semaphore(device, {.pNext = &semaphoreType});
        timelineValue = 0;

        for (size_t i = 0; i < MAX_FRAMES_IN_FLIGHT; i++) {
            vk::FenceCreateInfo fenceInfo{};
            inFlightFences.emplace_back(device, fenceInfo);
        }
    }

    void updateUniformBuffer(uint32_t currentImage) {
        UniformBufferObject ubo{};
        ubo.deltaTime = static_cast<float>(lastFrameTime) * 2.0f;

        memcpy(uniformBuffersMapped[currentImage], &ubo, sizeof(ubo));
    }

    void drawFrame() {
        auto [result, imageIndex] = swapChain.acquireNextImage(UINT64_MAX, nullptr, *inFlightFences[currentFrame]);
        while ( vk::Result::eTimeout == device.waitForFences( *inFlightFences[currentFrame], vk::True, UINT64_MAX ) )
            ;
        device.resetFences(*inFlightFences[currentFrame]);

        // Update timeline value for this frame
        uint64_t computeWaitValue = timelineValue;
        uint64_t computeSignalValue = ++timelineValue;
        uint64_t graphicsWaitValue = computeSignalValue;
        uint64_t graphicsSignalValue = ++timelineValue;

        updateUniformBuffer(currentFrame);

        {
            recordComputeCommandBuffer();
            // Submit compute work
            vk::TimelineSemaphoreSubmitInfo computeTimelineInfo{
                .waitSemaphoreValueCount = 1,
                .pWaitSemaphoreValues = &computeWaitValue,
                .signalSemaphoreValueCount = 1,
                .pSignalSemaphoreValues = &computeSignalValue
            };

            vk::PipelineStageFlags waitStages[] = {vk::PipelineStageFlagBits::eComputeShader};

            vk::SubmitInfo computeSubmitInfo{
                .pNext = &computeTimelineInfo,
                .waitSemaphoreCount = 1,
                .pWaitSemaphores = &*semaphore,
                .pWaitDstStageMask = waitStages,
                .commandBufferCount = 1,
                .pCommandBuffers = &*computeCommandBuffers[currentFrame],
                .signalSemaphoreCount = 1,
                .pSignalSemaphores = &*semaphore
            };

            computeQueue.submit(computeSubmitInfo, nullptr);
        }
        {
            // Record graphics command buffer
            recordCommandBuffer(imageIndex);

            // Submit graphics work (waits for compute to finish)
            vk::PipelineStageFlags waitStage = vk::PipelineStageFlagBits::eVertexInput;
            vk::TimelineSemaphoreSubmitInfo graphicsTimelineInfo{
                .waitSemaphoreValueCount = 1,
                .pWaitSemaphoreValues = &graphicsWaitValue,
                .signalSemaphoreValueCount = 1,
                .pSignalSemaphoreValues = &graphicsSignalValue
            };

            vk::SubmitInfo graphicsSubmitInfo{
                .pNext = &graphicsTimelineInfo,
                .waitSemaphoreCount = 1,
                .pWaitSemaphores = &*semaphore,
                .pWaitDstStageMask = &waitStage,
                .commandBufferCount = 1,
                .pCommandBuffers = &*commandBuffers[currentFrame],
                .signalSemaphoreCount = 1,
                .pSignalSemaphores = &*semaphore
            };

            graphicsQueue.submit(graphicsSubmitInfo, nullptr);

            // Present the image (wait for graphics to finish)
            vk::SemaphoreWaitInfo waitInfo{
                .semaphoreCount = 1,
                .pSemaphores = &*semaphore,
                .pValues = &graphicsSignalValue
            };

            // Wait for graphics to complete before presenting
            while ( vk::Result::eTimeout ==device.waitSemaphores(waitInfo, UINT64_MAX) )
                ;

            vk::PresentInfoKHR presentInfo{
                .waitSemaphoreCount = 0, // No binary semaphores needed
                .pWaitSemaphores = nullptr,
                .swapchainCount = 1,
                .pSwapchains = &*swapChain,
                .pImageIndices = &imageIndex
            };

            result = presentQueue.presentKHR(presentInfo);
            if (result == vk::Result::eErrorOutOfDateKHR || result == vk::Result::eSuboptimalKHR || framebufferResized) {
                framebufferResized = false;
                recreateSwapChain();
            } else if (result != vk::Result::eSuccess) {
                throw std::runtime_error("failed to present swap chain image!");
            }
        }
        currentFrame = (currentFrame + 1) % MAX_FRAMES_IN_FLIGHT;
    }

    [[nodiscard]] vk::raii::ShaderModule createShaderModule(const std::vector<char>& code) const {
        vk::ShaderModuleCreateInfo createInfo{ .codeSize = code.size(), .pCode = reinterpret_cast<const uint32_t*>(code.data()) };
        vk::raii::ShaderModule shaderModule{ device, createInfo };

        return shaderModule;
    }

    static vk::SurfaceFormatKHR chooseSwapSurfaceFormat(const std::vector<vk::SurfaceFormatKHR>& availableFormats) {
        for (const auto& availableFormat : availableFormats) {
            if (availableFormat.format == vk::Format::eB8G8R8A8Srgb && availableFormat.colorSpace == vk::ColorSpaceKHR::eSrgbNonlinear) {
                return availableFormat;
            }
        }

        return availableFormats[0];
    }

    static vk::PresentModeKHR chooseSwapPresentMode(const std::vector<vk::PresentModeKHR>& availablePresentModes) {
        for (const auto& availablePresentMode : availablePresentModes) {
            if (availablePresentMode == vk::PresentModeKHR::eMailbox) {
                return availablePresentMode;
            }
        }
        return vk::PresentModeKHR::eFifo;
    }

    [[nodiscard]] vk::Extent2D chooseSwapExtent(const vk::SurfaceCapabilitiesKHR& capabilities) const {
        if (capabilities.currentExtent.width != std::numeric_limits<uint32_t>::max()) {
            return capabilities.currentExtent;
        }
            int width, height;
            glfwGetFramebufferSize(window, &width, &height);

            return {
                std::clamp<uint32_t>(width, capabilities.minImageExtent.width, capabilities.maxImageExtent.width),
                std::clamp<uint32_t>(height, capabilities.minImageExtent.height, capabilities.maxImageExtent.height)
            };
            }

    [[nodiscard]] std::vector<const char*> getRequiredExtensions() const {
        uint32_t glfwExtensionCount = 0;
        auto glfwExtensions = glfwGetRequiredInstanceExtensions(&glfwExtensionCount);

        std::vector extensions(glfwExtensions, glfwExtensions + glfwExtensionCount);
        if (enableValidationLayers) {
            extensions.push_back(vk::EXTDebugUtilsExtensionName );
        }

        return extensions;
    }

    static VKAPI_ATTR vk::Bool32 VKAPI_CALL debugCallback(vk::DebugUtilsMessageSeverityFlagBitsEXT severity, vk::DebugUtilsMessageTypeFlagsEXT type, const vk::DebugUtilsMessengerCallbackDataEXT* pCallbackData, void*) {
        if (severity == vk::DebugUtilsMessageSeverityFlagBitsEXT::eError || severity == vk::DebugUtilsMessageSeverityFlagBitsEXT::eWarning) {
            std::cerr << "validation layer: type " << to_string(type) << " msg: " << pCallbackData->pMessage << std::endl;
        }

        return vk::False;
    }

    static std::vector<char> readFile(const std::string& filename) {
        std::ifstream file(filename, std::ios::ate | std::ios::binary);

        if (!file.is_open()) {
            throw std::runtime_error("failed to open file!");
        }
        std::vector<char> buffer(file.tellg());
        file.seekg(0, std::ios::beg);
        file.read(buffer.data(), static_cast<std::streamsize>(buffer.size()));
        file.close();

        return buffer;
    }
};

int main() {
    try {
        ComputeShaderApplication app;
        app.run();
    } catch (const std::exception& e) {
        std::cerr << e.what() << std::endl;
        return EXIT_FAILURE;
    }

    return EXIT_SUCCESS;
}