GDC 2005

1
Physical Simulation on GPUs
Jim Van VerthOpenGL Software EngineerNVIDIAjv
anverth_at_nvidia.comwww.nvidia.comwww.essentialmat
h.com
2
Physics on GPU

• Topics of discussion
• Ways of parallelizing physics
• Examples of GPU physics
• CUDA and you

3
Parallelizing Physics

• What weve talked about so far

Display
CPU
GPU
Game Logic, AI, Physics
Graphics
Awfully busy improve performance?
4
Parallelizing Physics

• Solution 1 Multicore CPU

Display
CPU
GPU
Game Logic, AI
Physics
Graphics
5
Parallelizing Physics

• Solution 2a Cell processor

Game Logic, AI
Display
CPU
GPU
Graphics
SPU
SPU
SPU
SPU
SPU
SPU
Physics
6
Parallelizing Physics

• Solution 2b AGEIA processor

Game Logic, AI
Display
CPU
GPU
Graphics

PPU
Physics
7
Parallelizing Physics

• Solution 3 Programmable GPU

Display
CPU
GPU
Game Logic, AI
Graphics, Physics
8
Parallelizing Physics

• Solution 3b SLI

GPU
Display
Graphics
CPU
Game Logic, AI
GPU
Physics
9
GPU Computing

• Modern GPU has many independent processors
• GeForce 8800 GTX 128 SPs
• GeForce 8800 GT 112 SPs
• Mostly processing power, not cache
• GeForce 8800 GTX 300-400 Gflops
• GeForce 8800 GT 500 Gflops
• A lot of parallel power for physics!

10
GPU Physics Example

• From GPU Gems 3
• Takahiro Harada, Real-time Rigid Body Simulation
  on GPUs
• Simple physics engine, all running on GPU

11
GPU Physics Example

• Idea GPU is good at
• Many similar computations
• Simple data
• So
• Particles for collision representation
• Grid for collision detection
• Simple collision response

12
Object Representation

• Global object data in texture pairs
• Alternate frame to frame

Linear Momentum
Angular Momentum
Position
Orientation
13
Object Representation

• Collision rep Solid (or shell) of particles
• Store as
• Fixed radius
• Displacement from center of mass

14
Object Representation

• Smaller particles better fit
• But more processing

15
Object Representation

• Particle data stored in texture and three
  rendertargets
• Update position, velocity each frame from global
  object data
• Update force from collisions

Displacement
Velocity
Force
Position
16
Pipeline
Update Particles
Calculate Grid
Compute Collisions
Integrate
17
Update Particles
Update Particles

• For each object do
• Iterate through all particles
• Update particle position, velocity

Calculate Grid
Compute Collisions
Integrate
18
Grid Representation
Update Particles

• Stored as slabs within 2D rendertarget
• Voxel stored as texel
• Four particle indices per texel

Calculate Grid
Compute Collisions
Integrate
19
Grid Creation
Update Particles

• For each particle do
• Compute grid index
• Write particle index to appropriate component at
  that location

Calculate Grid
Compute Collisions
Integrate
20
Collision Resolution
Update Particles

• For each voxel do
• For each particle in voxel do
• Compute force based on particles in this and 27
  neighboring voxels
• Regardless of collision!
• Spring force
• Damping from relative vel.
• Tangential force

Calculate Grid
Compute Collisions
Integrate
21
Integration
Update Particles

• Compute new linear and angular momenta based on
  collision (and other) forces
• Force/torque on rigid body is weighted sum of
  forces from each particle
• Compute new position and orientation from momenta

Calculate Grid
Compute Collisions
Integrate
22
Demo
23
Other approaches

• Simon Greens particles
• N-body
• Parallelize one piece
• Ex. Broad Phase (GPU Gems 3)
• Do smaller problem
• Ex. Fluid dynamics (Hellfire London)

24
GPU Computing

• How to program?
• Pre-G80, had to use Cg, GLSL, HLSL
• Problems
• Requires specialized shader knowledge
• Data is often texture or rendertarget
• Cant scatter data easily

25
CUDA

• Solution is CUDA
• Stands for Compute Unified Device Architecture
• Extensions on C/C
• Interoperable with D3D and OpenGL
• www.nvidia.com/cuda
• Use it!

26
CUDA

• Updating our example
• Instead of Cg, use standard C w/CUDA extensions
• Instead of textures or rendertargets, just use
  CUDA arrays
• Instead of vertex shader, use scatter operation

27
NVIDIA Presentations _at_ GDC 2008

• NVIDIA SessionsRoom 3003 West Hall
• Advanced Skin Rendering in NVIDIA's Human Head
  Demo400-500 pm, Wednesday, Feb. 20
• Particle-based Fluid Simulation For
  Games900-930 am, Thursday, Feb. 21
• 3D Stereoscopic Game Development - How to Make
  Your Game Look Like Beowulf 3D930-1000 am,
  Thursday, Feb. 21
• GPU Optimization with the Latest NVIDIA
  Performance Tools1030-1130 am, Thursday, Feb.
  21
• NVIDIA FX Composer 2 Shader Development
  Unleashed1200-100 pm, Thursday, Feb. 21

• General GDC Sessions
• Advanced Visual Effects with Direct
  3D1000-1800, Monday, Feb 18Room 2007 - West
  Hall
• Beyond Printf Debugging Graphics Through
  Tools1200-100 pm, Thursday, Feb. 21 Room 131
  - North Hall
• Real-Time Ambient Occlusion1450-1510, Friday,
  Feb 22 Room 3004 - West Hall
• Physics for Game Programmers1000-1800,
  Tuesday, Feb 19 Room 2018 - West Hall

28
The first-ever visual computing mega-event.. The
universe of visual computing in one place at one
time

• 15,000 Visitors
• 10,000 amateur and pro gamers, demoscencers, Game
  modders, machinima creators, and 3D artists
• 3,000 technical and marketing professionals from
  multiple industries
• 2000 visitors
• 300 press analysts

San Jose downtown Aug 25-27 2008
http//www.nvision2008.com
29
Questions?