| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
tested with reflect on 32-bit.. not sure why depthHasSurface isn't needed
|
|
|
|
|
|
|
|
it, and guessing that the two remaining entries in the 3-bit values were the
new funcs. Tested with modified stencilwrap test. Also, remove the commented
fallback stuff -- more modification to stencilwrap suggests that those issues
were just from span readback, not stencil implementation.
|
|
stencilwrap reported many issues with various modes. Some of these were
complicated by the fact that spans are broken (Bug #1615), but some appear to be
real bugs. However, while spans remain broken, I found that visual results were
better by avoiding fallbacks rather than avoiding just a broken stencil
implementation. Note that this required changing the depth spans at 24+8bpp
into read-modify-write cycles. It would be nicer as a single write with
a mask, but the kernel span blits turn off masking.
Reviewed by: ajax
|
|
flipping the signs on two factors. While this contradicts the DDK, it matches
what r200 does, and it fixes the test.
|
|
Rage 128 stencil support.
|
|
|
|
|
|
|
|
|
|
Changes to current operation:
-Elts are no longer converted to 16-bit format
-Cube maps
|
|
|
|
back on. Tested using seccolor modified to use the blue channel instead of
green, since green stays in the same place across RGB/BGR mistakes. Also hook
in UpdateSpecular on COLOR_EXT change, which might have resulted in missing
statechanges before.
|
|
output from indirect rendering).
|
|
|
|
wrote for sis, which has also exposed a few bugs elsewhere.
|
|
not respecting the coords (or perhaps interpreting them differently?) in my
testing. However, in the process it led to a fix of a secondary color handling
issue where it would be taken from the wrong offset, I believe, based off of
reading the r200 driver. Also add a minor tweak to save time in the
fog-but-not-specular case.
|
|
width and height stayed the same. Rely on front.offset pointing at the origin
in the window (rather than the origin of the front buffer), and fix the coords
in swapbuffers, which were wrong. This is different from other drivers because
we've got a back buffer with a 0,0 origin at the window's origin, which makes it
more convenient to have the front buffer match it.
|
|
last drop I saw, which was in turn based on a code drop of mine. Texturing,
culling, and several extensions are unimplemented, and some features could
probably be improved. It's untested, but there don't appear to be regressions on
the 300-series code, so I'd like to get these bits in now so that it can be
worked on as not a huge diff.
|
|
|
|
just do it per blit. No noticable performance difference (or behavior
difference, in my tests), and a significant difference in sanity.
|
|
wouldn't get confused when adding support for new formats. However, it resulted
in dying in texenv instead of drawing something undefined for GL_DECAL. Do what
GLcore apparently does and just pass the fragment through.
|
|
concerned that TEXEL_YUV422 looks like TEXEL_YVU422 and TEXEL_VUY422 looks like
TEXEL_VYU422 in yuvrect, but I'm happy enough for now.
|
|
|
|
tuxracer and the upcoming support for secondary color).
|
|
|
|
|
|
we start copying. I haven't seen it go wrong before, but I'm pretty sure this
fix is correct.
|
|
ones require custom texstore, so they aren't done yet.
- Add YCBCR support commented out, since the yuvsquare test didn't work with
just the bits I've added.
- Add the no-compression GL_ARB_texture_compression support.
- Add the driconf texture depth option and try to respect it.
|
|
the float values to bytes before packing them. Tested with texenv.
|
|
render state, not texture state). While here, move a SIS_FIREVERTICES to a more
appropriate location, though it was harmless.
|
|
tested using texwrap, fixing the sourcing of border constant values (use the
bytes, not the floats!). Remove the fallback on GL_CLAMP_TO_EDGE, which acts
properly, and add a note to GL_CLAMP about nonconformance (whcih is what the
fallback for CLAMP_TO_EDGE was probably targeting).
|
|
|
|
the register header was meaningful. It turns out those were the proper values
for the plain INCR/DECR ops, while what we were using as INCR/DECR were the
_WRAP versions. Tested with stencil_wrap (didn't expose normal/_WRAP swapping)
and stencilwrap (exposed it nicely) tests. Props to idr for poking me about
adding this.
|
|
Change the initialization to fail if an unknown pci id is detected, instead of accepting all values which are not r100-based radeons.
|
|
mklib changes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
