summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* r600 : add read port allocation for uniform;Richard Li2009-11-294-62/+101
| | | | | mapping ps input based on vs output; fix bugs including constants updating for vs.
* nv50: update linkage on rasterizer changeChristoph Bumiller2009-11-291-2/+4
| | | | | We need to update VP_RESULT_MAP and/or COORD_REPLACE_MAP when light_twoside and/or point_sprite are changed.
* nv50: match VP outputs to FP inputs ourselvesChristoph Bumiller2009-11-292-33/+40
| | | | | | | | | | | For each FP input, don't assume that the VP output will be at the same position, but scan the semantics instead, then put the correct output reg indices into VP_RESULT_MAP. Position is still assumed to be the first output/input. See 07fafc7c9346aa260829603bf3188596481e9e62, which renders previous assumptions incorrect.
* nv50: don't permanently negate src in emit_ddyChristoph Bumiller2009-11-291-4/+10
|
* util: Fix bad code.Corbin Simpson2009-11-281-1/+1
| | | | Uf. How embarrassing.
* util: Improve bitcount.Corbin Simpson2009-11-281-4/+8
| | | | Sorry for not pushing this before, it got lost in stashes.
* nv50: do conversion of last insn to 64 bit format firstChristoph Bumiller2009-11-271-16/+11
| | | | | | Simplifies things since the second to last one will then be converted in the subsequent pass that ensures alignment automatically.
* nv50: bswap32 the polygon stipple patternChristoph Bumiller2009-11-271-1/+1
| | | | | The hardware wants the pattern the same way it is passed to glPolygonStipple.
* gallium/util: added util_bswap32()Christoph Bumiller2009-11-271-0/+17
|
* svga: fix for not using texture width/height/depth arraysRoland Scheidegger2009-11-272-33/+32
|
* r300g,llvmpipe: fix some more merge problemsRoland Scheidegger2009-11-272-3/+3
|
* Merge branch 'width0'Roland Scheidegger2009-11-2772-594/+586
|\ | | | | | | | | | | | | Conflicts: src/gallium/drivers/r300/r300_texture.c src/gallium/state_trackers/xorg/xorg_exa.c src/mesa/state_tracker/st_cb_texture.c
| * gallium: fix more statetrackers/drivers for not using texture ↵Roland Scheidegger2009-11-2650-373/+360
| | | | | | | | width/height/depth arrays
| * gallium: don't use arrays for texture width,height,depthKeith Whitwell2009-11-1924-221/+226
| |
* | tgsi/exec: Fix orientation of DDY.Michal Krol2009-11-261-1/+1
| |
* | tgsi/exec: Force return from a subroutine at ENDSUB.Michal Krol2009-11-261-1/+27
| |
* | draw: Fix max_index check.Keith Whitwell2009-11-261-1/+2
| | | | | | | | | | We want to fallback to draw splitting when vertex element indices might be too high for atomic draw path (currently limited to 4095).
* | tgsi/ureg: Add forgotten goto in ureg_DECL_constant().Michal Krol2009-11-261-0/+1
| |
* | llvmpipe: Fake missing SSSE3 when simulation less capabable machines.José Fonseca2009-11-261-0/+1
| | | | | | | | SSE3 != SSSE3 and so far we only use the later.
* | llvmpipe: Update/correct CPU requirements.José Fonseca2009-11-261-3/+7
| | | | | | | | | | | | | | | | There are no hard requirements at the moment. We don't really emit any sse3 yet. Just some ssse3. Thanks to Roland for spotting these incorrections.
* | Merge commit 'origin/st-shader-varients'Roland Scheidegger2009-11-2614-651/+327
|\ \ | | | | | | | | | | | | | | | Conflicts: src/mesa/state_tracker/st_atom_shader.c src/mesa/state_tracker/st_program.c
| * | mesa/st: refactor vertex and fragment shader translationKeith Whitwell2009-11-1512-469/+276
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Translate vertex shaders independently of fragment shaders. Previously tried to make fragment shader semantic indexes always start at zero and exclude holes. This was unnecessary but meant that vertex shader translation had to be adjusted to take this into account. Now use a fixed scheme for labelling special FS input semantics (color, etc), and another fixed scheme for the generics. With this, vertex shaders can be translated independently of the bound fragment shader, assuming mesa has done its own job and ensured that the vertex shader provides at least the inputs the fragment shader is looking for. The state-tracker didn't attempt to do anything about this previously, so it shouldn't be needed now.
| * | mesa/st: emit tgsi vertex shader directly for drawpixelsKeith Whitwell2009-11-151-77/+47
| | | | | | | | | | | | | | | | | | | | | One of several cases where we build mesa shaders and then translate them to tgsi. Many of those cases it's because we're combining two mesa programs and there are helpers for that, but in this case at least can go straight to tgsi.
| * | mesa/st: don't calculate unused output_flags data eitherKeith Whitwell2009-11-153-24/+4
| | |
| * | mesa/st: don't calculate unused input_flags dataKeith Whitwell2009-11-153-16/+2
| | |
| * | mesa/st: don't calculate unused vs input semantic tagsKeith Whitwell2009-11-151-69/+2
| | |
* | | scons: Make it work with MinGW build of LLVM 2.6.José Fonseca2009-11-251-25/+27
| | | | | | | | | | | | LLVM 2.5 is no longer supported on windows.
* | | r600: add ARB_texture_non_power_of_two support.Dave Airlie2009-11-252-6/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This makes the miptree rounds up to the near POT for each level for all radeons, however since mipmaps aren't support with NPOT on previous radeons this calculation shouldn't cause any problems. If it does we can just make it r600 only. I tested a few mipmap demos on r500 and they all seem to work. Signed-off-by: Dave Airlie <[email protected]>
* | | r600 : reset stack flag with one channel only.Richard Li2009-11-241-1/+4
| | |
* | | r600 : fix stack depth setting bug.Richard Li2009-11-241-4/+4
| | |
* | | llvmpipe: Update instructions.José Fonseca2009-11-241-15/+17
| | |
* | | svga: Handle comment tokens when dumping.José Fonseca2009-11-242-1/+12
| | |
* | | svga: Use consistent names for public symbol names of shader dumping facilities.José Fonseca2009-11-245-12/+12
| | |
* | | svga: Use consistent file names for dumping facilities.José Fonseca2009-11-2410-11/+11
| | |
* | | util: Describe a few more formats.José Fonseca2009-11-241-0/+10
| | |
* | | llvmpipe: Use assert instead of abort. Only verify functions on debug builds.José Fonseca2009-11-242-2/+4
| | |
* | | llvmpipe: Fix memory leak.José Fonseca2009-11-241-0/+1
| | |
* | | llvmpipe: Use the generic conversion routine for depths.José Fonseca2009-11-241-7/+18
| | | | | | | | | | | | This allows for z32f depth format to work correctly.
* | | llvmpipe: Be more conservative with the supported formats.José Fonseca2009-11-241-7/+51
| | | | | | | | | | | | | | | We'll likely support much more formats, but doing this allows to run more testsuites without immediately hit assertion failures.
* | | wgl: Fix copy'n'paste typo in comment.José Fonseca2009-11-241-1/+1
| | |
* | | llvmpipe: Fix typo in comparison operator.José Fonseca2009-11-241-1/+1
| | |
* | | tgsi: Document Declaration Semantic token and FACE semantic name.Michal Krol2009-11-241-0/+32
| | |
* | | slang: Fix allocation size.Michal Krol2009-11-241-2/+2
| | | | | | | | | | | | We don't need 16K+ to store a single pointer.
* | | mesa: Fix pointer arithmetic.Michal Krol2009-11-241-1/+3
| | |
* | | tgsi: Implement predicated instructions in exec.Michal Krol2009-11-242-5/+48
| | |
* | | tgsi: Add ureg_DECL_loop().Michal Krol2009-11-242-0/+25
| | |
* | | tgsi: Account for gallium shader token representation changes.Michal Krol2009-11-245-132/+106
| | |
* | | gallium: Refactor the instruction predicate TGSI token.Michal Krol2009-11-241-10/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | Rename it to tgsi_instruction_predicate -- it's no longer an extended token. Its presence is indicated by a new flag in tgsi_instruction that indicates whether an instruction is predicated. Also, change predicate index representation to match the other tokens that specify register indices.
* | | r300: fix swtcl bo leak problem.Dave Airlie2009-11-241-3/+5
| | | | | | | | | | | | | | | | | | | | | We can get a lot of swtcl bo allocations - need to probably abstract this a bit further. Signed-off-by: Dave Airlie <[email protected]>
* | | radeon/r200/r300/r600: make bo mapping be explicitDave Airlie2009-11-2410-23/+60
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This moves the bo mapping outside the DMA layer and makes it explicit, this should in theory make it simpler to split the clean up the dma/cmdbuf linkage that I created before that is broken. Tested on: r600, rv380 (tcl/no-tcl), rv200 (tcl/no-tcl) Signed-off-by: Dave Airlie <[email protected]>
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-28 13:02:09 +0000