4de293bb9a
nv50: loops part 2
...
At least the mesa demo glsl/mandelbrot should work now.
2010-08-15 21:40:00 +02:00
34e0db4c50
nv50: more constant folding
2010-08-15 21:39:57 +02:00
3a68fcfb6b
nv50: begin implementing loops
2010-08-10 17:36:25 +02:00
fc1d72d15d
nv50: fix reg count
2010-08-10 17:35:26 +02:00
aaa8802a22
nv50: build proper phi functions in the first place
2010-08-05 00:50:00 +02:00
720e0c430d
nv50: fix constbuf validation
...
We only uploaded up to the highest offset a program would use,
and if the constant buffer isn't changed when a new program is
used, the new program is missing the rest of them.
Might want to introduce a "fill state" for user mem constbufs.
2010-08-05 00:50:00 +02:00
2c695d38e6
nv50: don't eliminate loads to dedicated values
2010-08-05 00:50:00 +02:00
fa67cabe7a
nv50: fixes for nested IFs
2010-07-31 18:32:35 +02:00
5705b45b6a
nv50: explicitly set src type for SET ops
...
Need to do this more nicely for all ops.
2010-07-31 18:32:35 +02:00
5de5e4fd5c
nv50: insert MOVs also for PHI sources from dominating block
...
Otherwise we get live range conflicts for operands that are written
only in e.g. an ELSE block but not the IF block.
2010-07-31 18:32:35 +02:00
582311ca97
nv50: fix for empty BBs
2010-07-31 18:32:34 +02:00
28ded2585c
nv50: add signed RGTC1 to format table, allow 2_10_10_10 for vbufs
2010-07-31 18:32:34 +02:00
7d34e79e44
nv50: add missing 2nd source for POW multiplication
2010-07-26 11:21:05 +02:00
e1ad3bd2f2
nv50: permit usage of undefined TGSI TEMPs
2010-07-26 11:20:52 +02:00
a3ba99b303
nv50: fix constant_operand opt mul by 2 case
2010-07-26 11:20:39 +02:00
5811c69264
nv50: simple reload elimination and local CSE
2010-07-26 11:20:28 +02:00
bb9d634730
nv50: add/fix some license headers
2010-07-24 22:16:40 +02:00
4baaf1d4c3
nv50: change back accidentally swapped UNORM,SNORM vertex type
2010-07-24 21:20:46 +02:00
1d1bb20612
nv50: don't produce MOV immediate to output reg in store opt
2010-07-24 21:20:40 +02:00
d7aac107e6
nv50: introduce the big formats table
2010-07-24 14:48:19 +02:00
f3af1201c5
nouveau: update nouveau_class.h
...
Adds nvc0, new vertex formats, and dual source blending values.
2010-07-24 14:48:15 +02:00
633f5ac612
nv50: import new compiler
2010-07-23 21:35:00 +02:00
c65f4fd5ae
i965: Cleanly fail programs with unsupported array access.
...
This should be more useful for developers and for bug triaging than
just generating wrong code.
2010-07-23 10:54:08 -07:00
35bbbf4742
i965: Add support for VS relative addressing of temporary arrays.
...
Fixes glsl-vs-arrays. Bug #27388 .
2010-07-23 10:46:42 -07:00
41e7347a17
draw: add small ybias factor for drawing wide points
...
Fixes minor rasterization error detected by some tests.
2010-07-23 10:32:58 -06:00
8122baf8ba
softpipe: Check for NULL pointer in sp_destroy_tile_cache().
2010-07-23 16:20:37 +02:00
a0fc83b277
softpipe: Check for NULL pointer in sp_destroy_tex_tile_cache().
2010-07-23 16:20:36 +02:00
0811b09669
nv50: implement depth clamp
2010-07-23 15:53:15 +02:00
a96feddddb
xorg/vmwgfx: Implement early mode pruning based on max fb size.
...
Also move some initialization from screen init to pre-init, now
that it is possible.
Also import a new vmwgfx drm (1.3) header.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com >
2010-07-23 04:48:33 +02:00
cec7d6a4de
st/xorg: Init the Gallium3D / libkms resources in pre-init.
...
This makes it possible to prune modes already in pre-init.
We also keep these resources alive across server generations, and
they are implicitly closed on server exit.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com >
2010-07-23 04:48:32 +02:00
f3ddffc392
st/xorg: Kill a couple of compilation warnings
...
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com >
2010-07-23 04:48:32 +02:00
6ffa23b8fe
st/xorg vmwgfx/xorg: Add a pre-init customizer callback
...
Add a customizer callback just before initial config setting, so that the
customizer code can initialize the mode validator using the drm
file-descriptor.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com >
2010-07-23 04:48:32 +02:00
a7a126bdfa
st/xorg: Add a possibility to prune modes and limit fb allocation size based on max fb size.
...
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com >
2010-07-23 04:48:32 +02:00
2299ff4c6b
llvmpipe: Partially fix resource texture from_handle
2010-07-22 20:26:35 -07:00
dc544d87a2
llvmpipe: Don't align values already aligned
2010-07-22 20:26:35 -07:00
095e99ddf6
i915g: Rename winsys debug options
2010-07-22 20:26:35 -07:00
5ccab575fd
i915g: Allow wrapping with software pipes
2010-07-22 20:26:35 -07:00
37dabfeef7
i915g: Set total_nblocksy in from_handle
2010-07-22 20:26:35 -07:00
5de2678b4e
i915g: Add some debug prints in texture code
2010-07-22 20:26:35 -07:00
4eaa671052
i915g: Ifdef out debug code on non-debug builds
2010-07-22 20:26:35 -07:00
e36a082214
tgsi: Fix error message on invalid swizzle parse
2010-07-22 20:26:35 -07:00
eeaab2047c
glx: Drop screen argument to GetGLXDRIDrawable
...
We'll just get it from the returned drawable when we need it.
2010-07-22 22:36:37 -04:00
7b7845a076
glx: Move WaitGL, WaitX, UseXFont to context vtable functions
2010-07-22 22:24:00 -04:00
31819830b6
glx: Allocate the __GLXcontext in the DRI drivers
2010-07-22 22:16:32 -04:00
ab434f6b76
glx: Use _Xglobal_lock for protecting extension display list
...
Avoids double locking glXLock in the X wire to event handlers.
2010-07-22 21:11:20 -04:00
c20a3628c7
glsl: remove invalid _mesa_problem() call
...
Fixes fd.o bug 29206.
2010-07-22 16:35:46 -06:00
815e79e72c
draw: re-order optimization passes depending on LLVM version, 32/64-bit
...
This is a work-around for an apparent bug in LLVM seen with piglit's
glsl-vs-sqrt-zero test.
2010-07-22 16:09:03 -06:00
ca3238f3fc
draw: added new assertions to clipping code
2010-07-22 11:18:54 -06:00
63ba1ec3c1
i965: Respect VS/VP point size result when enabled.
...
Fixes glsl-vs-point-size.
2010-07-22 10:06:55 -07:00
4ea71cbd0e
i965: Fix the disasm output for da16 src widths.
...
This has confused me twice now. It's a fixed width of 4 (usually a
region description of <4,4,1>), not 1. If it was 1, we'd have been
skipping all over register space.
2010-07-22 10:06:55 -07:00
Christoph Bumiller