HP j6700 OpenGL Implementation Guide for HP-UX 11.x - Page 60

optimization of lighting, occlusion culling, high frame rate applications, GL_LINE, GL_SHININESS

Page 60 highlights

programming hints OpenGL performance hints • Grouping your state changes together (that is, several modal state changes at one time), and then rendering primitives, will provide better performance than doing the modal state changes one by one and intermixing them with primitives. • Grouping primitives that require the same modal state together to minimize modal state changes. For example, if only part of a scene's primitives are lighted, draw all the lighted primitives, then turn off lighting and draw all the unlighted primitives, rather than enabling/disabling lighting many times. Some states negatively impact performance, such as two-sided lighting, polygon mode GL_LINE, and wide lines. optimization of lighting HP's implementation of OpenGL optimizes the lighting case such that the performance degradation from one light to two or more lights is linear. Lighting performance does not degrade noticeably when you enable a second light. In addition, the GL_SHININESS material parameter is not particularly expensive to change. occlusion culling The proper use of HP's occlusion culling extension can dramatically improve rendering performance. This extension defines a mechanism for determining the non-visibility of complex geometry based on the non-visibility of a bounding geometry. This feature can greatly reduce the amount of geometry processing and rendering required by an application, thereby, increasing the applications performance. For more information on occlusion culling, see the section "Occlusion Extension" found in Chapter 1. high frame rate applications To achieve maximum performance for buffer swap and clear operations across the entire family of Visualize and OEM graphics devices, avoid the following: • Scissor rectangle smaller than the window size • Non-trivial plane masks, i.e., any value other than all zeros or all ones 58 Chapter 5

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31
  • 32
  • 33
  • 34
  • 35
  • 36
  • 37
  • 38
  • 39
  • 40
  • 41
  • 42
  • 43
  • 44
  • 45
  • 46
  • 47
  • 48
  • 49
  • 50
  • 51
  • 52
  • 53
  • 54
  • 55
  • 56
  • 57
  • 58
  • 59
  • 60
  • 61
  • 62

programming hints
OpenGL performance hints
Chapter 5
58
Grouping your state changes together (that is, several modal state
changes at one time), and then rendering primitives, will provide
better performance than doing the modal state changes one by one
and intermixing them with primitives.
Grouping primitives that require the same modal state together to
minimize modal state changes. For example, if only part of a scene’s
primitives are lighted, draw all the lighted primitives, then turn off
lighting and draw all the unlighted primitives, rather than
enabling/disabling lighting many times.
Some states negatively impact performance, such as two-sided lighting,
polygon mode
GL_LINE
, and wide lines.
optimization of lighting
HP’s implementation of OpenGL optimizes the lighting case such that
the performance degradation from one light to two or more lights is
linear. Lighting performance does not degrade noticeably when you
enable a second light. In addition, the
GL_SHININESS
material parameter
is not particularly expensive to change.
occlusion culling
The proper use of HP’s occlusion culling extension can dramatically
improve rendering performance. This extension defines a mechanism for
determining the non-visibility of complex geometry based on the
non-visibility of a bounding geometry. This feature can greatly reduce the
amount of geometry processing and rendering required by an
application, thereby, increasing the applications performance. For more
information on occlusion culling, see the section “Occlusion Extension”
found in Chapter 1.
high frame rate applications
To achieve maximum performance for buffer swap and clear operations
across the entire family of Visualize and OEM graphics devices, avoid the
following:
Scissor rectangle smaller than the window size
Non-trivial plane masks, i.e., any value other than all zeros or all
ones