Vinson Lee
fddd5834ff
llvmpipe: Remove unnecessary header.
2010-06-23 15:17:00 -07:00
Corbin Simpson
e6ee4e1bdf
glhd: Simple rasterizer checks.
...
From the documentation.
2010-06-23 11:34:51 -07:00
Corbin Simpson
666fdc01c4
id, glhd: Unbreak texturing.
...
Argfl.
2010-06-23 11:25:52 -07:00
Corbin Simpson
9526305834
glhd: Use an environment variable (GALAHAD) to enable. Off by default.
2010-06-23 11:06:42 -07:00
Corbin Simpson
ee2c6d748d
id, glhd: Fix malloc/calloc of struct.
...
( >&)
2010-06-22 23:40:11 -07:00
Corbin Simpson
a6cc914874
glhd: Add query protection.
...
Not quite copied from r300g. This is slightly more API-compliant.
2010-06-22 23:11:01 -07:00
Corbin Simpson
aa451d509d
glhd: Grab framebuffer state checks from r300g.
2010-06-22 23:00:44 -07:00
Corbin Simpson
3f758d4ed4
glhd: Grab resource_copy_region from r300g.
2010-06-22 22:49:13 -07:00
Corbin Simpson
de7b181e99
glhd: Grab is_format_supported warning from r300g.
2010-06-22 22:49:13 -07:00
Corbin Simpson
7d551eb03d
glhd: Add glhd_warn for unconditional stderr output.
...
Design decision. Bad API errors should never be silenced.
Of course, you can always not use galahad.
2010-06-22 22:49:13 -07:00
Corbin Simpson
75612aa199
id, glhd: Fix segfault with misreferenced pipe member.
...
And remove the offending member to keep that from happening again.
2010-06-22 22:49:13 -07:00
Corbin Simpson
40faedfe7b
targets, radeong: Add Galahad.
...
Currently unconditional and causes segfaults.
2010-06-22 22:49:13 -07:00
Corbin Simpson
d3ad6fa579
gallium/drivers: Create Galahad from identity.
...
Galahad is a sanity-checking layer meant to replace the crufty and
scattered sanity checks inside drivers with a robust, non-silenceable,
useful set of warnings and errors that can be used to keep misbehaving
state trackers from going unnoticed.
2010-06-22 22:49:13 -07:00
Jakob Bornecrantz
41e0f6bc2f
i965g: Move bootstrap code to targets
2010-06-23 03:43:48 +02:00
Jakob Bornecrantz
2b15e37348
i965g: Moved pci_id to winsys struct
2010-06-23 03:20:45 +02:00
Marek Olšák
9ed7325848
r300g: attempt to fix texture corruption on RV505
2010-06-23 01:35:11 +02:00
Marek Olšák
d84bf6d44d
r300g: index buffer range checking
2010-06-23 00:20:11 +02:00
Jakob Bornecrantz
e130f524a9
i915g: Revert debug hunks from last commit
2010-06-22 22:55:13 +02:00
Jakob Bornecrantz
ca43b6ec9d
i915g: Reduce state emission by using a index bias
2010-06-22 20:04:56 +02:00
Jakob Bornecrantz
255d4f24e0
i915g: Drop fifo code in vbuf
2010-06-22 20:04:56 +02:00
Jakob Bornecrantz
4dd742cec3
i915g: Change state code in vbuf code
2010-06-22 20:04:56 +02:00
Jakob Bornecrantz
abbb1bde06
i915g: Rework debug print code
2010-06-22 20:04:55 +02:00
Jakob Bornecrantz
e694f3fd48
i915g: Switch to state atoms
2010-06-22 20:04:55 +02:00
Jakob Bornecrantz
ed675bb460
i915g: i915_state_sampler.c code style
2010-06-22 20:04:55 +02:00
Jakob Bornecrantz
2e7a90546d
i915g: i915_state_immediate.c code style
2010-06-22 20:04:55 +02:00
Jakob Bornecrantz
f93e378bfb
i915g: i915_state_dynamic.c code style
2010-06-22 20:04:55 +02:00
Jakob Bornecrantz
3a3e80ff96
llvmpipe: Ignores
2010-06-22 20:04:32 +02:00
Zack Rusin
b5e381d978
llvmpipe: make geometry shaders and stream output work
2010-06-22 12:58:04 -04:00
Zack Rusin
e433b73dd2
gallium: add a timestamp disjoint query
...
allows application to not only request the frequency of the TIME_ELAPSED
clock but also to detect if that frequency was consistent throughout the
entire bracketed range of graphics commands.
2010-06-22 12:58:04 -04:00
Zack Rusin
b6c360b46c
gallium: add a new query
...
to figure out if gpu is finished with all of the previously issues commands
2010-06-22 12:58:04 -04:00
Marek Olšák
eb7ef433bb
r300g: optimize the immediate mode emission path a bit
2010-06-22 06:50:23 +02:00
Marek Olšák
bd52246248
r300g: do not emit ZTOP every draw call
2010-06-22 06:50:22 +02:00
Marek Olšák
59df6725af
r300g: fix crash when destroying the context on r3xx-r4xx
...
FDO bug #28652 ,
2010-06-21 23:46:38 +02:00
José Fonseca
197636814e
llvmpipe: s/lpfs/shader/
2010-06-21 14:11:15 +01:00
Marek Olšák
d82f625333
r300g: fix the KIL opcode for r3xx-r4xx (v4)
2010-06-21 04:07:11 +02:00
Marek Olšák
646e9c2fd7
r300g: fix assigning the texture cache region for 1 texture
2010-06-20 14:06:14 +02:00
Vinson Lee
d05e848035
i965g: Fix setup of immediate types for gen4 disasm.
...
Caught by clang.
Follow-up to commit 070bbd4fcd
2010-06-20 00:58:52 -07:00
Marek Olšák
e8f2adf8e3
r300g: manually assign texture cache regions
...
This should fix corrupted texturing on r3xx-r4xx.
2010-06-20 05:30:04 +02:00
Marek Olšák
e4fe18a620
r300g: cleanup buffer_{from, get}_handle
2010-06-20 03:59:32 +02:00
Marek Olšák
6e96cfa39a
r300g: cleanup includes
2010-06-20 03:59:31 +02:00
Marek Olšák
a916669468
r300g: texture alignment code cleanup + use special alignment for AA textures
2010-06-20 03:59:28 +02:00
Marek Olšák
9b2ebcaf4b
r300g: fix random pixels appearing / incomplete rendering
...
This should fix the FDO bug #28612 .
Also, these piglit tests have been fixed:
- fbo-copypix
- scissor-copypixels
- copytexsubimage
- texredefine
Finally, 2 flushes in the transfer path are no longer needed.
2010-06-19 20:26:16 +02:00
Marek Olšák
4c76184305
r300g: output floats instead of halfs when passed through translate
2010-06-19 17:20:40 +02:00
Marek Olšák
ce4c9a7f25
r300g: fix clamp and mirror_clamp wrap modes when used with nearest filtering
2010-06-19 06:18:17 +02:00
Marek Olšák
6acd2117c1
r300g: more informative warning in END_CS
2010-06-19 05:15:07 +02:00
Marek Olšák
371ca689ec
r300g: improve the ZTOP condition
...
The real difference here is that ZTOP is now enabled if all stencil ops
are set to KEEP. This improves performance.
2010-06-19 05:14:31 +02:00
Marek Olšák
8ea45a2c0e
r300g: fix up the CS size of the rasterizer state
2010-06-19 05:14:18 +02:00
José Fonseca
d7f4250333
llvmpipe: Fix pointer dereference after free.
2010-06-18 17:37:37 +01:00
Roland Scheidegger
17c9d7eea7
llvmpipe: limit the number of fragment shader variants kept around
...
llvmpipe can create a large number of shader variants for a single shader
(which are quite big), and they were only ever deleted if the shader itself
was deleted. This is especially apparent in things like glean
blendFunc where a new variant is created for every different subtest, chewing
up all memory.
This change limits the numbers of fragment shader variants (for all shaders)
which are kept around to a fixed number. If that would be exceeded a fixed
portion of the cached variants is deleted (since without tracking the used
variants this involves flushing we don't want to delete only one).
Always the least recently used variants (from all shaders together) are
deleted.
For now this is all per-context.
Both the number of how many variants are cached (1024) as well as how many
will be deleted at once (1/4 of the cache size) are just rough guesses and
subject to further optimization.
2010-06-18 13:52:17 +01:00
Marek Olšák
ff8e1452df
r300g: fix hardlocks in occlusion queries
...
Early Z test (=ZTOP) must be disabled before a query is started,
otherwise the GPU is dead. The order of emitted registers matters more
than you might think.
This fixes hardlocks in sauerbraten.
2010-06-18 02:17:29 +02:00
