summaryrefslogtreecommitdiffstats
path: root/src/gallium
Commit message (Collapse)AuthorAgeFilesLines
* nv50: generate JOINs for outermost IF clausesChristoph Bumiller2010-08-175-11/+31
|
* nv50: fix thinko in store to output reg possible checkChristoph Bumiller2010-08-171-1/+1
|
* nv50: flatten simple IF/ELSE/ENDIF constructsChristoph Bumiller2010-08-174-24/+123
| | | | Less branching means less instructions and less thread divergence.
* nv50: loops part 2Christoph Bumiller2010-08-157-90/+189
| | | | At least the mesa demo glsl/mandelbrot should work now.
* nv50: more constant foldingChristoph Bumiller2010-08-151-27/+177
|
* nv50: begin implementing loopsChristoph Bumiller2010-08-108-127/+416
|
* nv50: fix reg countChristoph Bumiller2010-08-101-2/+2
|
* nv50: build proper phi functions in the first placeChristoph Bumiller2010-08-055-147/+166
|
* nv50: fix constbuf validationChristoph Bumiller2010-08-051-1/+6
| | | | | | | | 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.
* nv50: don't eliminate loads to dedicated valuesChristoph Bumiller2010-08-051-6/+4
|
* nv50: fixes for nested IFsChristoph Bumiller2010-07-315-69/+127
|
* nv50: explicitly set src type for SET opsChristoph Bumiller2010-07-311-0/+6
| | | | Need to do this more nicely for all ops.
* nv50: insert MOVs also for PHI sources from dominating blockChristoph Bumiller2010-07-312-33/+32
| | | | | Otherwise we get live range conflicts for operands that are written only in e.g. an ELSE block but not the IF block.
* nv50: fix for empty BBsChristoph Bumiller2010-07-314-35/+34
|
* nv50: add signed RGTC1 to format table, allow 2_10_10_10 for vbufsChristoph Bumiller2010-07-311-3/+7
|
* nv50: add missing 2nd source for POW multiplicationChristoph Bumiller2010-07-261-1/+9
|
* nv50: permit usage of undefined TGSI TEMPsChristoph Bumiller2010-07-261-1/+7
|
* nv50: fix constant_operand opt mul by 2 caseChristoph Bumiller2010-07-261-5/+1
|
* nv50: simple reload elimination and local CSEChristoph Bumiller2010-07-264-24/+178
|
* nv50: add/fix some license headersChristoph Bumiller2010-07-248-32/+169
|
* nv50: change back accidentally swapped UNORM,SNORM vertex typeChristoph Bumiller2010-07-241-6/+6
|
* nv50: don't produce MOV immediate to output reg in store optChristoph Bumiller2010-07-242-10/+14
|
* nv50: introduce the big formats tableChristoph Bumiller2010-07-2410-272/+486
|
* nouveau: update nouveau_class.hChristoph Bumiller2010-07-243-84/+1159
| | | | Adds nvc0, new vertex formats, and dual source blending values.
* nv50: import new compilerChristoph Bumiller2010-07-2315-4687/+6485
|
* draw: add small ybias factor for drawing wide pointsBrian Paul2010-07-231-0/+1
| | | | Fixes minor rasterization error detected by some tests.
* softpipe: Check for NULL pointer in sp_destroy_tile_cache().Michal Krol2010-07-231-8/+10
|
* softpipe: Check for NULL pointer in sp_destroy_tex_tile_cache().Michal Krol2010-07-231-11/+13
|
* nv50: implement depth clampChristoph Bumiller2010-07-234-10/+33
|
* xorg/vmwgfx: Implement early mode pruning based on max fb size.Thomas Hellstrom2010-07-233-4/+37
| | | | | | | | | 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 <[email protected]>
* st/xorg: Init the Gallium3D / libkms resources in pre-init.Thomas Hellstrom2010-07-232-86/+75
| | | | | | | | 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 <[email protected]>
* st/xorg: Kill a couple of compilation warningsThomas Hellstrom2010-07-232-2/+2
| | | | Signed-off-by: Thomas Hellstrom <[email protected]>
* st/xorg vmwgfx/xorg: Add a pre-init customizer callbackThomas Hellstrom2010-07-233-5/+19
| | | | | | | | 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 <[email protected]>
* st/xorg: Add a possibility to prune modes and limit fb allocation size based ↵Thomas Hellstrom2010-07-233-0/+23
| | | | | | on max fb size. Signed-off-by: Thomas Hellstrom <[email protected]>
* llvmpipe: Partially fix resource texture from_handleJakob Bornecrantz2010-07-221-1/+35
|
* llvmpipe: Don't align values already alignedJakob Bornecrantz2010-07-221-2/+2
|
* i915g: Rename winsys debug optionsJakob Bornecrantz2010-07-221-2/+2
|
* i915g: Allow wrapping with software pipesJakob Bornecrantz2010-07-222-3/+11
|
* i915g: Set total_nblocksy in from_handleJakob Bornecrantz2010-07-221-0/+1
|
* i915g: Add some debug prints in texture codeJakob Bornecrantz2010-07-221-6/+10
|
* i915g: Ifdef out debug code on non-debug buildsJakob Bornecrantz2010-07-221-0/+5
|
* tgsi: Fix error message on invalid swizzle parseJakob Bornecrantz2010-07-221-1/+1
|
* draw: re-order optimization passes depending on LLVM version, 32/64-bitBrian Paul2010-07-221-2/+15
| | | | | This is a work-around for an apparent bug in LLVM seen with piglit's glsl-vs-sqrt-zero test.
* draw: added new assertions to clipping codeBrian Paul2010-07-221-1/+10
|
* softpipe: add missing support for PIPE_FORMAT_S8_USCALED surfacesBrian Paul2010-07-211-20/+20
| | | | | | And remove checks of surface depth bits. The state tracker should not turn on depth/stencil testing if the framebuffer doesn't have depth/stencil.
* softpipe: fix sp_tile_cache_flush_clear() regressionBrian Paul2010-07-211-5/+11
|
* r600g: add support for all R6XX/R7XX asicJerome Glisse2010-07-214-87/+316
| | | | | | | This configure some of the value properly based on asic so others asic than RV710 works too. Signed-off-by: Jerome Glisse <[email protected]>
* r600g: add r600 compile mode to compiler.Dave Airlie2010-07-214-13/+106
| | | | | | some of the ALU instructions are different on r6xx vs r7xx, separate the alu translation to separate files, and use family to pick which compile stage to use.
* r600g: add family retrivalDave Airlie2010-07-213-1/+14
| | | | allow pipe driver to get the family of the gpu.
* llvmpipe: say no to depth clampMarek Olšák2010-07-211-0/+2
| | | | The other drivers just return 0 without the assert.