| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
and no more switches inside loops. To do: give the 1D/3D/cube routines the
same treatment.
|
|
Use fast approximation to log().
Check for dq==0 to avoid a per-pixel divide.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Big clean-up of line drawing code.
Removed many obsolete span processing functions.
|
|
ditch the pb (pixel buffer) code.
Converted point drawing, bitmaps and aa lines to use new span functions.
|
|
Major clean-up of pixel zoom code.
|
|
|
|
Updated comments, fixed indentation, etc.
|
|
Removed last of span.filled* flags and Klaus's macros.
More simplification of triangle functions.
|
|
_mesa_rasterize_span() is gone, replaced by new _mesa_write_textured_span().
Removed some unneeded triangle functions - more simplification possible.
|
|
struct sw_span is used throughout span/fragment processing.
This is leading to less code and more chances for optimization.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Chimera bug)
|
|
Moved vertex program hash table into shared context state.
Implemented reference counting for vertex programs.
Replaced tnl "ProjectedClip" with "Ndc" (normalized device coordinates).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
GL_ARB_shadow_ambient
|
|
|
