kitware / vtk-m

Mirror of https://gitlab.kitware.com/vtk/vtk-m

VTK-m

VTK-m is a toolkit of scientific visualization algorithms for emerging processor architectures. VTK-m supports the fine-grained concurrency for data analysis and visualization algorithms required to drive extreme scale computing by providing abstract models for data and execution that can be applied to a variety of algorithms across many different processor architectures.

You can find out more about the design of VTK-m on the VTK-m Wiki.

Learning Resources

• A high-level overview is given in the IEEE Vis talk "VTK-m: Accelerating the Visualization Toolkit for Massively Threaded Architectures."

• The VTK-m Users Guide provides extensive documentation. It is broken into multiple parts for learning and references at multiple different levels.

  • "Part 1: Getting Started" provides the introductory instruction for building VTK-m and using its high-level features.
  • "Part 2: Using VTK-m" covers the core fundamental components of VTK-m including data model, worklets, and filters.
  • "Part 3: Developing with VTK-m" covers how to develop new worklets and filters.
  • "Part 4: Advanced Development" covers topics such as new worklet types and custom device adapters.

• A practical VTK-m Tutorial based in what users want to accomplish with VTK-m:

  • Building VTK-m and using existing VTK-m data structures and filters.
  • Algorithm development with VTK-m.
  • Writing new VTK-m filters.

• Community discussion takes place on the VTK-m users email list.

• Doxygen-generated reference documentation is available for both:

  • Last Nightly build VTK-m Doxygen nightly
  • Last release VTK-m Doxygen latest

Contributing

There are many ways to contribute to VTK-m, with varying levels of effort.

• Ask a question on the VTK-m users email list.

• Submit new or add to discussions of a feature requests or bugs on the VTK-m Issue Tracker.

• Submit a Pull Request to improve VTK-m

  • See CONTRIBUTING.md for detailed instructions on how to create a Pull Request.
  • See the VTK-m Coding Conventions that must be followed for contributed code.

• Submit an Issue or Pull Request for the VTK-m Users Guide

Dependencies

VTK-m Requires:

• C++14 Compiler. VTK-m has been confirmed to work with the following
  • GCC 5.4+
  • Clang 5.0+
  • XCode 5.0+
  • MSVC 2015+
  • Intel 17.0.4+
• CMake
  • CMake 3.12+
  • CMake 3.13+ (for CUDA support)

Optional dependencies are:

• Kokkos Device Adapter
  • Kokkos 3.7+
• CUDA Device Adapter
  • Cuda Toolkit 9.2, >= 10.2
  • Note CUDA >= 10.2 is required on Windows
• TBB Device Adapter
  • TBB
• OpenMP Device Adapter
  • Requires a compiler that supports OpenMP >= 4.0.
• OpenGL Rendering
  • The rendering module contains multiple rendering implementations including standalone rendering code. The rendering module also includes (optionally built) OpenGL rendering classes.
  • The OpenGL rendering classes require that you have a extension binding library and one rendering library. A windowing library is not needed except for some optional tests.
• Extension Binding
  • GLEW
• On Screen Rendering
  • OpenGL Driver
  • Mesa Driver
• On Screen Rendering Tests
  • GLFW
  • GLUT
• Headless Rendering
  • OS Mesa
  • EGL Driver

VTK-m has been tested on the following configurations:c

• On Linux
  • GCC 5.4.0, 5.4, 6.5, 7.4, 8.2, 9.2; Clang 5, 8; Intel 17.0.4; 19.0.0
  • CMake 3.12, 3.13, 3.16, 3.17
  • CUDA 9.2, 10.2, 11.0, 11.1
  • TBB 4.4 U2, 2017 U7
• On Windows
  • Visual Studio 2015, 2017
  • CMake 3.12, 3.17
  • CUDA 10.2
  • TBB 2017 U3, 2018 U2
• On MacOS
  • AppleClang 9.1
  • CMake 3.12
  • TBB 2018

Building

VTK-m supports all majors platforms (Windows, Linux, OSX), and uses CMake to generate all the build rules for the project. The VTK-m source code is available from the VTK-m download page or by directly cloning the VTK-m git repository.

The basic procedure for building VTK-m is to unpack the source, create a build directory, run CMake in that build directory (pointing to the source) and then build. Here are some example *nix commands for the process (individual commands may vary).

$ tar xvzf ~/Downloads/vtk-m-v2.0.0.tar.gz
$ mkdir vtkm-build
$ cd vtkm-build
$ cmake-gui ../vtk-m-v2.0.0
$ cmake --build -j .              # Runs make (or other build program)

A more detailed description of building VTK-m is available in the VTK-m Users Guide.

Example

The VTK-m source distribution includes a number of examples. The goal of the VTK-m examples is to illustrate specific VTK-m concepts in a consistent and simple format. However, these examples only cover a small portion of the capabilities of VTK-m.

Below is a simple example of using VTK-m to create a simple data set and use VTK-m's rendering engine to render an image and write that image to a file. It then computes an isosurface on the input data set and renders this output data set in a separate image file:

#include <vtkm/cont/Initialize.h>
#include <vtkm/source/Tangle.h>

#include <vtkm/rendering/Actor.h>
#include <vtkm/rendering/CanvasRayTracer.h>
#include <vtkm/rendering/MapperRayTracer.h>
#include <vtkm/rendering/MapperVolume.h>
#include <vtkm/rendering/MapperWireframer.h>
#include <vtkm/rendering/Scene.h>
#include <vtkm/rendering/View3D.h>

#include <vtkm/filter/contour/Contour.h>

using vtkm::rendering::CanvasRayTracer;
using vtkm::rendering::MapperRayTracer;
using vtkm::rendering::MapperVolume;
using vtkm::rendering::MapperWireframer;

int main(int argc, char* argv[])
{
  vtkm::cont::Initialize(argc, argv, vtkm::cont::InitializeOptions::Strict);

  auto tangle = vtkm::source::Tangle(vtkm::Id3{ 50, 50, 50 });
  vtkm::cont::DataSet tangleData = tangle.Execute();
  std::string fieldName = "tangle";

  // Set up a camera for rendering the input data
  vtkm::rendering::Camera camera;
  camera.SetLookAt(vtkm::Vec3f_32(0.5, 0.5, 0.5));
  camera.SetViewUp(vtkm::make_Vec(0.f, 1.f, 0.f));
  camera.SetClippingRange(1.f, 10.f);
  camera.SetFieldOfView(60.f);
  camera.SetPosition(vtkm::Vec3f_32(1.5, 1.5, 1.5));
  vtkm::cont::ColorTable colorTable("inferno");

  // Background color:
  vtkm::rendering::Color bg(0.2f, 0.2f, 0.2f, 1.0f);
  vtkm::rendering::Actor actor(tangleData.GetCellSet(),
                               tangleData.GetCoordinateSystem(),
                               tangleData.GetField(fieldName),
                               colorTable);
  vtkm::rendering::Scene scene;
  scene.AddActor(actor);
  // 2048x2048 pixels in the canvas:
  CanvasRayTracer canvas(2048, 2048);
  // Create a view and use it to render the input data using OS Mesa

  vtkm::rendering::View3D view(scene, MapperVolume(), canvas, camera, bg);
  view.Paint();
  view.SaveAs("volume.png");

  // Compute an isosurface:
  vtkm::filter::contour::Contour filter;
  // [min, max] of the tangle field is [-0.887, 24.46]:
  filter.SetIsoValue(3.0);
  filter.SetActiveField(fieldName);
  vtkm::cont::DataSet isoData = filter.Execute(tangleData);
  // Render a separate image with the output isosurface
  vtkm::rendering::Actor isoActor(
    isoData.GetCellSet(), isoData.GetCoordinateSystem(), isoData.GetField(fieldName), colorTable);
  // By default, the actor will automatically scale the scalar range of the color table to match
  // that of the data. However, we are coloring by the scalar that we just extracted a contour
  // from, so we want the scalar range to match that of the previous image.
  isoActor.SetScalarRange(actor.GetScalarRange());
  vtkm::rendering::Scene isoScene;
  isoScene.AddActor(isoActor);

  // Wireframe surface:
  vtkm::rendering::View3D isoView(isoScene, MapperWireframer(), canvas, camera, bg);
  isoView.Paint();
  isoView.SaveAs("isosurface_wireframer.png");

  // Smooth surface:
  vtkm::rendering::View3D solidView(isoScene, MapperRayTracer(), canvas, camera, bg);
  solidView.Paint();
  solidView.SaveAs("isosurface_raytracer.png");

  return 0;
}

A minimal CMakeLists.txt such as the following one can be used to build this example.

cmake_minimum_required(VERSION 3.12...3.15 FATAL_ERROR)
project(VTKmDemo CXX)

#Find the VTK-m package
find_package(VTKm REQUIRED QUIET)

if(TARGET vtkm::rendering)
  add_executable(Demo Demo.cxx)
  target_link_libraries(Demo PRIVATE vtkm::filter vtkm::rendering vtkm::source)
endif()

License

VTK-m is distributed under the OSI-approved BSD 3-clause License. See LICENSE.txt for details.

vtk-m's Issues

vtkm does not compile for CUDA architectures < 60 (Solution included)

Problem

I found a compile error in:

• vtkm 2.0.0
• on Windows MSVC,
• Using CUDA 11.8, and
• CMAKE_CUDA_ARCHITECTURES "all-major" (or probably any configuration with CUDA versions below 60)

Looking at the source code, the problem is also in master, but I haven't tested it on master, because compilation takes forever. The problem should be reproducible when compiling with CUDA architecture 'sm_35' though.

The compile fails in Atomic.h, because atomicAdd() is not natively supported for double in cuda architectures < 60.

Solution

The fallback implementation is actually in the source file Atomic.h, but it is named wrongly.

Since there are no usages of vtkmAtomicAdd() in the whole code base, I assume it should be named AtomicAddImpl() and was just a mistake during refactoring. The same applies to the implementation for vtkm::Float32 directly above, which should be renamed from vtkmAtomicAddImpl() to AtomicAddImpl().

Renaming both of the implementations to AtomicAddImpl() lets me compile on above config using "all-major", which for me goes down to 'sm_35'.