Introduction to Geometry Shaders

1
Introduction to Geometry Shaders

• Patrick Cozzi
• Analytical Graphics, Inc.

2
Overview

• Geometry Shaders in the Pipeline
• Primitive Types
• Applications
• Performance

3
Birds Eye View

• Create or destroy primitives on the GPU
• Requires DirectX 10 or GL_ARB_geometry_shader4

Geometry Shader
4
Geometry Shaders in the Pipeline
5
Geometry Shaders in the Pipeline
6
Primitive Types
Geometry Shader
Output primitives can be disconnected
7
Primitive Types

• Input Primitives
• GL_POINTS
• GL_LINES
• GL_TRIANGLES
• Adjacency

• Output Primitives
• GL_POINTS
• GL_LINE_STRIP
• GL_TRIANGLE_STRIP

8
Primitive Types

• Input primitive type doesnt have to equal output
  primitive type

9
Applications

• Implement glPolygonMode
• Triangles Points or Line Strips
• Emulate GL_ARB_point_sprite
• Points Triangle Strips

10
Applications

• Displacement Mapping
• Single pass generate a cube map
• Extrusions
• Shadow volumes
• Fins along silhouettes for fur rendering

11
Applications Fur in Lost Planet

• Render surface, write buffers for
• Fur Color
• Angle
• Length
• GS turns each pixel into a translucent polyline
• Automatic LOD

12
Applications Fur in Lost Planet
color
angle
length
Images from meshula.net/wordpress/?p124
13
Performance

• Duplicates per-vertex operations for vertices
  shared by primitives

5 vertices processed
9 vertices processed
Geometry Shader
Vertex Shader
14
Performance

• Must guarantee order in order out

15
Performance

• Order guarantee affects parallelism

16
Performance

• Buffer size needs to support a number of threads
  running in parallel

17
Performance

• Maximum number of vertices a GS will output
• GEOMETRY_VERTICES_OUT_ARB
• Determines the speed of GS execution
• Make this and vertex sizes as small as possible

18
Performance

• GeForce 8, 9, and GTX2xx
• Output size vertex size GEOMETRY_VERTICES_OUT_
  ARB
• Maximum output size 1,024 scalars
• Performance is inversely proportional to output
  size
• Not a continuous function
• 1-20 scalars Peak Performance
• 27-40 scalars 50 Performance
• On GeForce 8800 GTX

19
Performance

• Benefits
• Reduces vertex buffer memory usage
• Compute in GS, e.g. normals
• Create more geometry
• No need to duplicate (e.g. compared to equivalent
  VS implementation)
• Less memory less bus traffic
• Reduces vertex attribute setup cost

20
Summary

• Modify incoming primitive or make a limited
  number of copies
• Not for
• Large scale amplification
• Instancing

21
Resources
Introduction to Direct3D 10 SIGGRAPH 2007 Course
Notes

• www.microsoft.com/downloads/details.aspx?FamilyId
  96CD28D5-4C15-475E-A2DC-1D37F67FA6CDdisplaylange
  n

22
Resources
GL_ARB_geometry_shader4

• www.opengl.org/registry/specs/ARB/geometry_shader4
  .txt

23
Resources
Section 3.5

• www.realtimerendering.com

24
Resources
Section 4.6

• developer.nvidia.com/object/gpu_programming_guide.
  html