This repository contains many examples showing different aspects of Vulkan, debugging tips, and using other Nvidia tools. See the samples section below.

Each sample has its own documentation that describes the functionality and where to get more information.

Dependencies

• nvpro_core : Vulkan helper classes and other small utilities.

Cloning repositories

git clone --recursive --shallow-submodules https://github.com/nvpro-samples/nvpro_core.git
git clone https://github.com/nvpro-samples/vk_mini_samples.git

Generate solution

cd vk_mini_samples
mkdir build
cd build
cmake ..

The Aftermath sample requires to download the SDK separately: Nsight Aftermath SDK

By default, the samples are created using GLSL. However, many of them also have equivalent shaders written in HLSL and SLANG. To switch between them, select either USE_HLSL or USE_SLANG. Afterward, regenerate CMake and the solution will update with compatible projects and their respective shaders.

If you are new to this repository, the first samples to read to better understand the framework are solid color and rectangle.

Name	Description	Image	GLSL	HLSL	Slang
solid_color	Set a user custom color to a pixel wide texture and display it.		[x]	[x]	[x]
rectangle	Render a 2D rectangle to GBuffer.		[x]	[x]	[x]
aftermath	Integrate the Nsight Aftermath SDK to an existing application		[x]	[x]	[ ]
image_ktx	Display KTX image and apply tonemap post processing		[x]	[x]	[x]
image_viewer	Load an image, allow to zoom and pan under mouse cursor		[x]	[x]	[x]
mm_displacement	Micro-mesh displacement		[x]	[x]	[x]
mm_opacity	Micromap opacity		[x]	[x]	[x]
msaa	Hardware Multi-Sampling Anti-Aliasing		[x]	[x]	[x]
shader_printf	Add printf to shader and display in a log window		[x]	[x]	[x]
ray_trace	Simple ray tracer using metalic-roughness shading, reflection and shadows and simple sky shader.		[x]	[x]	[x]
shading execution reorder	Known also as SER, this shows how to reorder execution rays to gain a better usage of the GPU.		[x]	[x]	[x]
simple_polygons	Rasterizing multiple polygonal objects.		[x]	[x]	[x]
offscreen	Render without window context and save image to disk.		[x]	[x]	[x]
tiny_shader_toy	Compile shader on the fly, diplay compilation errors, multiple pipeline stages.		[x]	[ ]	[ ]
barycentric_wireframe	Draw wifreframe in a a single pass using gl_BaryCoordNV		[x]	[x]	[x]
texture 3d	Create a 3D texture and do ray marching.		[x]	[x]	[x]
position fetch	Using VK_KHR_ray_tracing_position_fetch.		[x]	[ ]	[x]
ray_query	Doing inline raytracing in a compute shader		[x]	[x]	[x]
ray_query_position_fetch	Using VK_KHR_ray_tracing_position_fetch in ray quary		[x]	[ ]	[x]
shader_object	Using shader object and dynamic pipline		[x]	[x]	[x]
compute_only	Simple compute and display example		[x]	[x]	[x]
gpu_monitor	Showing GPU usage		[x]	[x]	[x]
realtime_analysis	Displaying in realtime GPU informations		[x]	[ ]	[x]
motion_blur	Tracing object having motion, using the NVIDIA raytrace extension		[x]	[ ]	[x]

The examples utilize many helpers from the nvpro_core repository. Each sample's core uses the Application class to create a window, initialize the UI, and create a swapchain with the ImGui framework. The Application class is a modified version of the Dear ImGui Vulkan example.

Samples are attached to the Application class as Engines. While the application is running, the sample will be called to render its UI or to perform rendering operations in the current frame.

The init() function will create the Vulkan context using nvvk::Context, create the GLFW window and create the swapchains by calling ImGui_ImplVulkanH_CreateOrResizeWindow.

The run() function is an infinite loop until the close event is triggered. Within the loop, each engine will be called (in order) with:

• onResize : Called when the viewport size is changing
• onUIMenu : To add functions to the menubar
• onUIRender : Called for anything related to UI
• onRender : For anything to render within a frame, with the command buffer of the frame.

At the end of each loop the frame is rendered with frameRender() then the frame is presented with framePresent().

Instead of directly consuming shaders in a human-readable text format, Vulkan uses SPIR-V as an intermediate representation. This allows the use of shader languages other than GLSL as long as they can target the Vulkan SPIR-V environment.

The samples can use any of the shader languages, and the CMake option is : USE_SLANG our USE_HLSL. Only one of the two can be true and if both are false, it default to using GLSL.

Slang is a high-level based shader language, similar to C++. It is one of the primary language used by researchers at NVIDIA, as it can generate not only Spir-V, but also DX12, CUDA, CPP.

Note: To change the version of the Slang compiler, select the path to SLANG_COMPILER in CMake.

Microsoft's High Level Shading Language (HLSL), used by DirectX, can generate Spir-V code. The DXC compiler comes by default with recent Vulkan SDK.

Note: To use a different dxc binary, open Vulkan and change the path to Vulkan_dxc_EXECUTABLE.

HLSL

• HLSL to SPIR-V: Feature Mapping Manual
• Ray Tracing: HLSL
• Porting to HLSL
  • GLSL variables
  • GLSL types
  • Global Variables
  • Mapping between HLSL and GLSL

SLANG

• Github: https://github.com/shader-slang/slang
• Releases: https://github.com/shader-slang/slang/releases
• Getting Started: https://shader-slang.com/getting-started.html
• User Guide: http://shader-slang.com/slang/user-guide/index.html
• Various Documentations: https://github.com/shader-slang/slang/tree/master/docs
• Interop GLSL, SPIR-V: https://shader-slang.com/slang/user-guide/a1-04-interop.html

• GL_EXT_spirv_intrinsics
• KHR Extensions
• JSON

Copyright 2024 NVIDIA CORPORATION. Released under Apache License, Version 2.0. See "LICENSE" file for details.