Age | Commit message (Collapse) | Author |
|
Also, quite a bit of re-org of the tile caches and surface mapping/unmapping.
Leave surfaces mapped between primitives now.
|
|
|
|
|
|
Instead of using read/write_quad() functions, do framebuffer accesses via
get/put_tile(). A cache of tiles is used to avoid frequent get/put() calls.
Only implemented for color buffers right now.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This controls whether texcoords are interpreted as-is or scaled up from [0,1].
Fixes glDrawPixels/glBitmap problems on i915 when image is non power-of-two.
Also, cleans up the CSO sampler state for i915 a bit.
|
|
|
|
|
|
|
|
ignore stencil
|
|
|
|
ignored for now
|
|
texture and level!=0)
|
|
|
|
|
|
Previously, output[0] was always Z and output[1] was color. Now output[0]
will be color if Z is not written.
In shade_quad() use the semantic info to determine which quantity is in
which output slot.
|
|
fragment shader writes Z.
|
|
|
|
|
|
|
|
Unfortunately, the generated fragment shader code is effectively unusable until
it handles quad->mask.
|
|
structure so allocate a private copy of it.
|
|
in the state functions.
|
|
|
|
This, plus expanding all instructions ahead of time, seems to have improved
the performance of program execution by 8x or so.
|
|
|
|
Fixes crash when GL_BASE_LEVEL!=0.
Also, remove old assertion.
|
|
|
|
This reverts commit 6dcfddb8e2ec2bfb6187b912807fa65f28da2c5e.
|
|
|
|
|
|
|