Adds tests for the new Morton Code class

07_StagingAndMultipleQueues/app_resources/common.hlsl

@@ -1,8 +1,8 @@
 #include "nbl/builtin/hlsl/cpp_compat.hlsl"
 
-NBL_CONSTEXPR uint32_t WorkgroupSizeX = 16;
-NBL_CONSTEXPR uint32_t WorkgroupSizeY = 16;
-NBL_CONSTEXPR uint32_t WorkgroupSize = WorkgroupSizeX*WorkgroupSizeY;
+NBL_CONSTEXPR_INLINE_NSPC_SCOPE_VAR uint32_t WorkgroupSizeX = 16;
+NBL_CONSTEXPR_INLINE_NSPC_SCOPE_VAR uint32_t WorkgroupSizeY = 16;
+NBL_CONSTEXPR_INLINE_NSPC_SCOPE_VAR uint32_t WorkgroupSize = WorkgroupSizeX*WorkgroupSizeY;
 
 static const uint32_t FRAMES_IN_FLIGHT = 3u;
 

14_Mortons/CMakeLists.txt

@@ -0,0 +1,24 @@
+include(common RESULT_VARIABLE RES)
+if(NOT RES)
+	message(FATAL_ERROR "common.cmake not found. Should be in {repo_root}/cmake directory")
+endif()
+
+nbl_create_executable_project("" "" "" "" "${NBL_EXECUTABLE_PROJECT_CREATION_PCH_TARGET}")
+
+if(NBL_EMBED_BUILTIN_RESOURCES)
+	set(_BR_TARGET_ ${EXECUTABLE_NAME}_builtinResourceData)
+	set(RESOURCE_DIR "app_resources")
+
+	get_filename_component(_SEARCH_DIRECTORIES_ "${CMAKE_CURRENT_SOURCE_DIR}" ABSOLUTE)
+	get_filename_component(_OUTPUT_DIRECTORY_SOURCE_ "${CMAKE_CURRENT_BINARY_DIR}/src" ABSOLUTE)
+	get_filename_component(_OUTPUT_DIRECTORY_HEADER_ "${CMAKE_CURRENT_BINARY_DIR}/include" ABSOLUTE)
+
+    file(GLOB_RECURSE BUILTIN_RESOURCE_FILES RELATIVE "${CMAKE_CURRENT_SOURCE_DIR}/${RESOURCE_DIR}" CONFIGURE_DEPENDS "${CMAKE_CURRENT_SOURCE_DIR}/${RESOURCE_DIR}/*")
+    foreach(RES_FILE ${BUILTIN_RESOURCE_FILES})
+      LIST_BUILTIN_RESOURCE(RESOURCES_TO_EMBED "${RES_FILE}")
+    endforeach()
+
+	ADD_CUSTOM_BUILTIN_RESOURCES(${_BR_TARGET_} RESOURCES_TO_EMBED "${_SEARCH_DIRECTORIES_}" "${RESOURCE_DIR}" "nbl::this_example::builtin" "${_OUTPUT_DIRECTORY_HEADER_}" "${_OUTPUT_DIRECTORY_SOURCE_}")
+
+	LINK_BUILTIN_RESOURCES_TO_TARGET(${EXECUTABLE_NAME} ${_BR_TARGET_})
+endif()

14_Mortons/ITester.h

@@ -0,0 +1,279 @@
+#ifndef _NBL_EXAMPLES_TESTS_22_CPP_COMPAT_I_TESTER_INCLUDED_
+#define _NBL_EXAMPLES_TESTS_22_CPP_COMPAT_I_TESTER_INCLUDED_
+
+#include <nabla.h>
+#include "app_resources/common.hlsl"
+#include "nbl/application_templates/MonoDeviceApplication.hpp"
+
+using namespace nbl;
+
+class ITester 
+{
+public:
+    virtual ~ITester()
+    {
+        m_outputBufferAllocation.memory->unmap();
+    };
+
+    struct PipelineSetupData
+    {
+        std::string testShaderPath;
+        core::smart_refctd_ptr<video::ILogicalDevice> device;
+        core::smart_refctd_ptr<video::CVulkanConnection> api;
+        core::smart_refctd_ptr<asset::IAssetManager> assetMgr;
+        core::smart_refctd_ptr<system::ILogger> logger;
+        video::IPhysicalDevice* physicalDevice;
+        uint32_t computeFamilyIndex;
+    };
+
+    template<typename InputStruct, typename OutputStruct>
+    void setupPipeline(const PipelineSetupData& pipleineSetupData)
+    {
+        // setting up pipeline in the constructor
+        m_device = core::smart_refctd_ptr(pipleineSetupData.device);
+        m_physicalDevice = pipleineSetupData.physicalDevice;
+        m_api = core::smart_refctd_ptr(pipleineSetupData.api);
+        m_assetMgr = core::smart_refctd_ptr(pipleineSetupData.assetMgr);
+        m_logger = core::smart_refctd_ptr(pipleineSetupData.logger);
+        m_queueFamily = pipleineSetupData.computeFamilyIndex;
+        m_semaphoreCounter = 0;
+        m_semaphore = m_device->createSemaphore(0);
+        m_cmdpool = m_device->createCommandPool(m_queueFamily, video::IGPUCommandPool::CREATE_FLAGS::RESET_COMMAND_BUFFER_BIT);
+        if (!m_cmdpool->createCommandBuffers(video::IGPUCommandPool::BUFFER_LEVEL::PRIMARY, 1u, &m_cmdbuf))
+            logFail("Failed to create Command Buffers!\n");
+
+        // Load shaders, set up pipeline
+        core::smart_refctd_ptr<asset::IShader> shader;
+        {
+            asset::IAssetLoader::SAssetLoadParams lp = {};
+            lp.logger = m_logger.get();
+            lp.workingDirectory = ""; // virtual root
+            auto assetBundle = m_assetMgr->getAsset(pipleineSetupData.testShaderPath, lp);
+            const auto assets = assetBundle.getContents();
+            if (assets.empty())
+                return logFail("Could not load shader!");
+
+            // It would be super weird if loading a shader from a file produced more than 1 asset
+            assert(assets.size() == 1);
+            core::smart_refctd_ptr<asset::IShader> source = asset::IAsset::castDown<asset::IShader>(assets[0]);
+
+            shader = m_device->compileShader({source.get()});
+        }
+
+        if (!shader)
+            logFail("Failed to create a GPU Shader, seems the Driver doesn't like the SPIR-V we're feeding it!\n");
+
+        video::IGPUDescriptorSetLayout::SBinding bindings[2] = {
+            {
+                .binding = 0,
+                .type = asset::IDescriptor::E_TYPE::ET_STORAGE_BUFFER,
+                .createFlags = video::IGPUDescriptorSetLayout::SBinding::E_CREATE_FLAGS::ECF_NONE,
+                .stageFlags = ShaderStage::ESS_COMPUTE,
+                .count = 1
+            },
+            {
+                .binding = 1,
+                .type = asset::IDescriptor::E_TYPE::ET_STORAGE_BUFFER,
+                .createFlags = video::IGPUDescriptorSetLayout::SBinding::E_CREATE_FLAGS::ECF_NONE,
+                .stageFlags = ShaderStage::ESS_COMPUTE,
+                .count = 1
+            }
+        };
+
+        core::smart_refctd_ptr<video::IGPUDescriptorSetLayout> dsLayout = m_device->createDescriptorSetLayout(bindings);
+        if (!dsLayout)
+            logFail("Failed to create a Descriptor Layout!\n");
+
+        m_pplnLayout = m_device->createPipelineLayout({}, core::smart_refctd_ptr(dsLayout));
+        if (!m_pplnLayout)
+            logFail("Failed to create a Pipeline Layout!\n");
+
+        {
+            video::IGPUComputePipeline::SCreationParams params = {};
+            params.layout = m_pplnLayout.get();
+            params.shader.entryPoint = "main";
+            params.shader.shader = shader.get();
+            if (!m_device->createComputePipelines(nullptr, { &params,1 }, &m_pipeline))
+                logFail("Failed to create pipelines (compile & link shaders)!\n");
+        }
+
+        // Allocate memory of the input buffer
+        {
+            constexpr size_t BufferSize = sizeof(InputStruct);
+
+            video::IGPUBuffer::SCreationParams params = {};
+            params.size = BufferSize;
+            params.usage = video::IGPUBuffer::EUF_STORAGE_BUFFER_BIT;
+            core::smart_refctd_ptr<video::IGPUBuffer> inputBuff = m_device->createBuffer(std::move(params));
+            if (!inputBuff)
+                logFail("Failed to create a GPU Buffer of size %d!\n", params.size);
+
+            inputBuff->setObjectDebugName("emulated_float64_t output buffer");
+
+            video::IDeviceMemoryBacked::SDeviceMemoryRequirements reqs = inputBuff->getMemoryReqs();
+            reqs.memoryTypeBits &= m_physicalDevice->getHostVisibleMemoryTypeBits();
+
+            m_inputBufferAllocation = m_device->allocate(reqs, inputBuff.get(), video::IDeviceMemoryAllocation::EMAF_NONE);
+            if (!m_inputBufferAllocation.isValid())
+                logFail("Failed to allocate Device Memory compatible with our GPU Buffer!\n");
+
+            assert(inputBuff->getBoundMemory().memory == m_inputBufferAllocation.memory.get());
+            core::smart_refctd_ptr<video::IDescriptorPool> pool = m_device->createDescriptorPoolForDSLayouts(video::IDescriptorPool::ECF_NONE, { &dsLayout.get(),1 });
+
+            m_ds = pool->createDescriptorSet(core::smart_refctd_ptr(dsLayout));
+            {
+                video::IGPUDescriptorSet::SDescriptorInfo info[1];
+                info[0].desc = core::smart_refctd_ptr(inputBuff);
+                info[0].info.buffer = { .offset = 0,.size = BufferSize };
+                video::IGPUDescriptorSet::SWriteDescriptorSet writes[1] = {
+                    {.dstSet = m_ds.get(),.binding = 0,.arrayElement = 0,.count = 1,.info = info}
+                };
+                m_device->updateDescriptorSets(writes, {});
+            }
+        }
+
+        // Allocate memory of the output buffer
+        {
+            constexpr size_t BufferSize = sizeof(OutputStruct);
+
+            video::IGPUBuffer::SCreationParams params = {};
+            params.size = BufferSize;
+            params.usage = video::IGPUBuffer::EUF_STORAGE_BUFFER_BIT;
+            core::smart_refctd_ptr<video::IGPUBuffer> outputBuff = m_device->createBuffer(std::move(params));
+            if (!outputBuff)
+                logFail("Failed to create a GPU Buffer of size %d!\n", params.size);
+
+            outputBuff->setObjectDebugName("emulated_float64_t output buffer");
+
+            video::IDeviceMemoryBacked::SDeviceMemoryRequirements reqs = outputBuff->getMemoryReqs();
+            reqs.memoryTypeBits &= m_physicalDevice->getHostVisibleMemoryTypeBits();
+
+            m_outputBufferAllocation = m_device->allocate(reqs, outputBuff.get(), video::IDeviceMemoryAllocation::EMAF_NONE);
+            if (!m_outputBufferAllocation.isValid())
+                logFail("Failed to allocate Device Memory compatible with our GPU Buffer!\n");
+
+            assert(outputBuff->getBoundMemory().memory == m_outputBufferAllocation.memory.get());
+            core::smart_refctd_ptr<video::IDescriptorPool> pool = m_device->createDescriptorPoolForDSLayouts(video::IDescriptorPool::ECF_NONE, { &dsLayout.get(),1 });
+
+            {
+                video::IGPUDescriptorSet::SDescriptorInfo info[1];
+                info[0].desc = core::smart_refctd_ptr(outputBuff);
+                info[0].info.buffer = { .offset = 0,.size = BufferSize };
+                video::IGPUDescriptorSet::SWriteDescriptorSet writes[1] = {
+                    {.dstSet = m_ds.get(),.binding = 1,.arrayElement = 0,.count = 1,.info = info}
+                };
+                m_device->updateDescriptorSets(writes, {});
+            }
+        }
+
+        if (!m_outputBufferAllocation.memory->map({ 0ull,m_outputBufferAllocation.memory->getAllocationSize() }, video::IDeviceMemoryAllocation::EMCAF_READ))
+            logFail("Failed to map the Device Memory!\n");
+
+        // if the mapping is not coherent the range needs to be invalidated to pull in new data for the CPU's caches
+        const video::ILogicalDevice::MappedMemoryRange memoryRange(m_outputBufferAllocation.memory.get(), 0ull, m_outputBufferAllocation.memory->getAllocationSize());
+        if (!m_outputBufferAllocation.memory->getMemoryPropertyFlags().hasFlags(video::IDeviceMemoryAllocation::EMPF_HOST_COHERENT_BIT))
+            m_device->invalidateMappedMemoryRanges(1, &memoryRange);
+
+        assert(memoryRange.valid() && memoryRange.length >= sizeof(OutputStruct));
+
+        m_queue = m_device->getQueue(m_queueFamily, 0);
+    }
+
+    enum class TestType
+    {
+        CPU,
+        GPU
+    };
+
+    template<typename T>
+    void verifyTestValue(const std::string& memberName, const T& expectedVal, const T& testVal, const TestType testType)
+    {
+        if (expectedVal == testVal)
+            return;
+
+        std::stringstream ss;
+        switch (testType)
+        {
+        case TestType::CPU:
+            ss << "CPU TEST ERROR:\n";
+            break;
+        case TestType::GPU:
+            ss << "GPU TEST ERROR:\n";
+        }
+
+        ss << "nbl::hlsl::" << memberName << " produced incorrect output!" << '\n';
+
+        m_logger->log(ss.str().c_str(), system::ILogger::ELL_ERROR);
+    }
+
+protected:
+    uint32_t m_queueFamily;
+    core::smart_refctd_ptr<video::ILogicalDevice> m_device;
+    core::smart_refctd_ptr<video::CVulkanConnection> m_api;
+    video::IPhysicalDevice* m_physicalDevice;
+    core::smart_refctd_ptr<asset::IAssetManager> m_assetMgr;
+    core::smart_refctd_ptr<system::ILogger> m_logger;
+    video::IDeviceMemoryAllocator::SAllocation m_inputBufferAllocation = {};
+    video::IDeviceMemoryAllocator::SAllocation m_outputBufferAllocation = {};
+    core::smart_refctd_ptr<video::IGPUCommandBuffer> m_cmdbuf = nullptr;
+    core::smart_refctd_ptr<video::IGPUCommandPool> m_cmdpool = nullptr;
+    core::smart_refctd_ptr<video::IGPUDescriptorSet> m_ds = nullptr;
+    core::smart_refctd_ptr<video::IGPUPipelineLayout> m_pplnLayout = nullptr;
+    core::smart_refctd_ptr<video::IGPUComputePipeline> m_pipeline;
+    core::smart_refctd_ptr<video::ISemaphore> m_semaphore;
+    video::IQueue* m_queue;
+    uint64_t m_semaphoreCounter;
+
+    template<typename InputStruct, typename OutputStruct>
+    OutputStruct dispatch(const InputStruct& input)
+    {
+        // Update input buffer
+        if (!m_inputBufferAllocation.memory->map({ 0ull,m_inputBufferAllocation.memory->getAllocationSize() }, video::IDeviceMemoryAllocation::EMCAF_READ))
+            logFail("Failed to map the Device Memory!\n");
+
+        const video::ILogicalDevice::MappedMemoryRange memoryRange(m_inputBufferAllocation.memory.get(), 0ull, m_inputBufferAllocation.memory->getAllocationSize());
+        if (!m_inputBufferAllocation.memory->getMemoryPropertyFlags().hasFlags(video::IDeviceMemoryAllocation::EMPF_HOST_COHERENT_BIT))
+            m_device->invalidateMappedMemoryRanges(1, &memoryRange);
+
+        std::memcpy(static_cast<InputStruct*>(m_inputBufferAllocation.memory->getMappedPointer()), &input, sizeof(InputStruct));
+
+        m_inputBufferAllocation.memory->unmap();
+
+        // record command buffer
+        m_cmdbuf->reset(video::IGPUCommandBuffer::RESET_FLAGS::NONE);
+        m_cmdbuf->begin(video::IGPUCommandBuffer::USAGE::NONE);
+        m_cmdbuf->beginDebugMarker("test", core::vector4df_SIMD(0, 1, 0, 1));
+        m_cmdbuf->bindComputePipeline(m_pipeline.get());
+        m_cmdbuf->bindDescriptorSets(nbl::asset::EPBP_COMPUTE, m_pplnLayout.get(), 0, 1, &m_ds.get());
+        m_cmdbuf->dispatch(1, 1, 1);
+        m_cmdbuf->endDebugMarker();
+        m_cmdbuf->end();
+
+        video::IQueue::SSubmitInfo submitInfos[1] = {};
+        const video::IQueue::SSubmitInfo::SCommandBufferInfo cmdbufs[] = { {.cmdbuf = m_cmdbuf.get()} };
+        submitInfos[0].commandBuffers = cmdbufs;
+        const video::IQueue::SSubmitInfo::SSemaphoreInfo signals[] = { {.semaphore = m_semaphore.get(), .value = ++m_semaphoreCounter, .stageMask = asset::PIPELINE_STAGE_FLAGS::COMPUTE_SHADER_BIT} };
+        submitInfos[0].signalSemaphores = signals;
+
+        m_api->startCapture();
+        m_queue->submit(submitInfos);
+        m_api->endCapture();
+
+        m_device->waitIdle();
+        OutputStruct output;
+        std::memcpy(&output, static_cast<OutputStruct*>(m_outputBufferAllocation.memory->getMappedPointer()), sizeof(OutputStruct));
+        m_device->waitIdle();
+
+        return output;
+    }
+
+private:
+    template<typename... Args>
+    inline void logFail(const char* msg, Args&&... args)
+    {
+        m_logger->log(msg, system::ILogger::ELL_ERROR, std::forward<Args>(args)...);
+        exit(-1);
+    }
+};
+
+#endif