Skip to content

Latest commit

 

History

History
133 lines (84 loc) · 4.48 KB

README.md

File metadata and controls

133 lines (84 loc) · 4.48 KB

Submission 07

Updates:

  1. Fixed bugs. Now the simulation won't explode.

Submission 06

IMPORTANT:

  1. After generating the project using cmake and running it once to generate executable, please copy tbb_debug.dll and Half.dll from nuparu/lib/win into where the executable resides.

  2. Libraries used were compiled with VS 14 x64 compiler. They may not work with your compiler if you are using a different one.

Min OpenGL requirement: 4.3

Visual Studio requirement: 14 (2015) x64

Updates:

  1. Implemented FLIP using GPU kernels;

Performance Analysis:

  • 20 x 20 x 20 grid:

    Particle Count FPS
    19,200 12
    153,600 8
    518,400 4
    1,228,800 2
    2,400,000 1.2
    (Grid Size, Particle Count) FPS
    (10x10x10, 2,400) 45
    (20x20x20, 19,200) 12
    (40x40x40, 166,400) 4.8
    (50x50x50, 320,200) 2.7
    (80x80x80, 1,331,200) 0.7

Existing Problems:

  1. The simulation is unstable. Some local area will explode during the simulation.

Submission 05

IMPORTANT:

  1. After generating the project using cmake and running it once to generate executable, please copy tbb_debug.dll and Half.dll from nuparu/lib/win into where the executable resides.

  2. Libraries used were compiled with VS 14 x64 compiler. They may not work with your compiler if you are using a different one.

  3. Uncomment buildSDF() in FS_Grid::render() to enable writing VDB files. You may want to change the ouput directory specified in FS_Grid::buildSDF().

Min OpenGL requirement: 4.3

Visual Studio requirement: 14 (2015) x64

Updates:

  1. Implemented Conjugate Gradient method with Diagonal Preconditioning using GPU kernels. Current implementation is not optimal due to heavy CPU intervention but it is still about 10 times faster than the CPU implementation.

Submission 04

IMPORTANT:

  1. After generating the project using cmake and running it once to generate executable, please copy tbb_debug.dll and Half.dll from nuparu/lib/win into where the executable resides.

  2. Libraries used were compiled with VS 14 x64 compiler. They may not work with your compiler if you are using a different one.

  3. Uncomment buildSDF() in FS_Grid::render() to enable writing VDB files. You may want to change the ouput directory specified in FS_Grid::buildSDF().

Min OpenGL requirement: 4.3 Visual Studio requirement: 14 (2015) x64

Parallelization Performance Analysis (10 x 10 grid with 8 particles per cell):

Topic Serial(FPS) Parallel(FPS) Comparison Reason(if negative)
Interpolation (grid to particles) ~= 5.0 ~= 7.0 +40%
Interpolation (particles to grid) ~= 7.0 ~= 6.0 -14.2% Mutex overhead
Extrapolation ~= 6.5 ~= 7.0 +7.7%

Updates:

  1. Pressure solve
  2. Parallelization
  3. Meshing using OpenVDB

Submission 03

IMPORTANT: after generating the project using cmake and running it once to generate executable, please copy tbb_debug.dll into where the executable resides.

  1. Implemented all required features;
  2. Used parallel_for in trilinear interpolation.

Submission 02

Min OpenGL requirement: 4.3 Visual Studio requirement: 14 (2015) x64

Updates:

  1. Grid and MACGrid classes;
  2. Construction and initialization of MAC grid;
  3. Transfer particle velocity to grid;
  4. A trilinear interpolation for PIC (not used yet. Only gravity affects particles right now);
  5. Debug rendering of the grid, particles, and velocity components.

Submission 01

Basic viewer funtionality implemented.

Tested on VS 2015, Windows 10.

Inorder to compile the code, you need to change two #define's in main.hpp:

  1. SCENE_FILE_NAME: the path + name of the scene json file (e.g. "C:/path_to_scene_file/scene.json")
  2. SHADERS_DIR: the directory where the shaders locate without trailing slash (e.g. "C:/path_to_shaders")