summaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/llvmpipe
Commit message (Collapse)AuthorAgeFilesLines
* Merge branch 'mesa_7_6_branch' into mesa_7_7_branchAndre Maasikas2009-12-081-3/+3
|\
| * llvmpipe: Initialize variables in emit_instruction.Vinson Lee2009-12-071-3/+3
| |
* | llvmpipe: Fix derived blend color state.José Fonseca2009-11-101-1/+1
| |
* | llvmpipe: Ensure stack variables in unit tests are properly aligned.José Fonseca2009-11-093-12/+15
| |
* | llvmpipe: Fix build with llvm 2.6.José Fonseca2009-11-062-5/+10
| | | | | | | | Fixes bug 24949.
* | llvmpipe: Sanitise shader semantic and interpolator handling.Michal Krol2009-11-031-24/+10
| | | | | | | | | | | | | | | | Handle the remaining semantic names and indices. Respect color interpolator when not flatshading. Based on Michal's softpipe commit eb699d64ec7057032139baccedcb0694ca41d706.
* | llvmpipe: remove old prim_setup draw stageKeith Whitwell2009-11-0310-386/+59
| | | | | | | | | | | | | | Everything now goes through the draw_vbuf handler, the same as regular drivers. Based on Keith's commit 4fe0fc3eba1f79beda890a5016359d549bab6ad4.
* | llvmpipe: Respect gl_rasterization_rules in primitive setup.José Fonseca2009-11-031-18/+30
| | | | | | | | | | Based on Michal's identical commit for softpipe (ca9c413647bf9efb5ed770e3a655bc758075aec7).
* | llvmpipe: Make sure the JIT engine and X86 target are linked on MSVC build.José Fonseca2009-10-286-4/+118
| | | | | | | | | | Basically mimic the llvm 2.6 way of linking execution engines and targets.
* | llvmpipe: Implement round() for MSVC.José Fonseca2009-10-281-0/+12
| |
* | llvmpipe: Human friendlier sampler state dump.José Fonseca2009-10-252-9/+22
| |
* | llvmpipe: Dump the sampler state of the shader key.José Fonseca2009-10-251-0/+18
| |
* | llvmpipe: Immediate multiplication.José Fonseca2009-10-254-32/+64
| |
* | llvmpipe: Fast path for sampling rgba8 textures with linear filtering.José Fonseca2009-10-251-12/+256
| | | | | | | | | | | | | | | | Implement Keith's suggestion of doing most of the sampling with 16x8 and 8x16 AoS, and only doing the conversion to floating point SoA at the very last step. Improves gloss performance by 10%.
* | llvmpipe: Make lerping work for 8.8 fixed point values.José Fonseca2009-10-251-4/+50
| |
* | llvmpipe: New module to help make assertions about formats.José Fonseca2009-10-254-0/+78
| |
* | llvmpipe: New function to unpack rgba8 formats into 4 x u8n AoS.José Fonseca2009-10-252-12/+148
| |
* | llvmpipe: Split the format swizzling step from the unpacking.José Fonseca2009-10-252-16/+34
| |
* | llvmpipe: Allow different signs when unpacking.José Fonseca2009-10-251-2/+1
| |
* | llvmpipe: Add inlines to quickly generate types matching the native SIMD ↵José Fonseca2009-10-252-12/+102
| | | | | | | | register bitwidth.
* | llvmpipe: Merge lp_build_load_rgba_soa into lp_build_sample_texel.José Fonseca2009-10-251-32/+16
| |
* | llvmpipe: Factor our pixel offset computation.José Fonseca2009-10-253-43/+83
| |
* | llvmpipe: Share testing infrastructure with lp_test_format.José Fonseca2009-10-252-23/+46
| |
* | llvmpipe: Eliminate lp_build_store_rgba_aos.José Fonseca2009-10-253-49/+16
| |
* | llvmpipe: Eliminate lp_build_load_rgba_aos.José Fonseca2009-10-253-52/+15
| |
* | llvmpipe: Move a few format/sampling functions into better space.José Fonseca2009-10-257-123/+158
| |
* | gallium: remove the swizzling parts of ExtSwizzleKeith Whitwell2009-10-231-20/+6
| | | | | | | | | | | | | | | | | | These haven't been used by the mesa state tracker since the conversion to tgsi_ureg, and it seems that none of the other state trackers are using it either. This helps simplify one of the biggest suprises when starting off with TGSI shaders.
* | gallium: remove noise opcodesKeith Whitwell2009-10-231-9/+0
| | | | | | | | | | | | | | | | | | | | | | Provide a dummy implementation in the GL state tracker (move 0.5 to the destination regs). At some point, a motivated person could add a better implementation of noise. Currently not even the nvidia binary drivers do anything more than this. In any case, the place to do this is in the GL state tracker, not the poor driver.
* | llvmpipe: Avoid yet another variable size array.José Fonseca2009-10-221-2/+4
| |
* | llvmpipe: Define rdtsc for MSVC.José Fonseca2009-10-221-3/+14
| |
* | llvmpipe: Avoid variable size arrays.José Fonseca2009-10-223-4/+7
| | | | | | | | Not really variable size, but MSVC still doesn't like them.
* | llvmpipe: Use the pack/unpack functions for 8bit unsigned norm multiplication.José Fonseca2009-10-221-79/+17
| |
* | llvmpipe: Utility function to double the bit width of a type.José Fonseca2009-10-222-5/+28
| |
* | llvmpipe: Factor vector packing/unpacking to a separate source file.José Fonseca2009-10-225-237/+519
| | | | | | | | | | These functions will be needed to implement many of the 8bit operations, and they are quite complex on its own.
* | llvmpipe: Call util_cpu_detect() from the unit tests.José Fonseca2009-10-222-0/+7
| |
* | llvmpipe: Remove extraneous name in lp_type pre-declaration.José Fonseca2009-10-215-5/+5
| |
* | llvmpipe: Allocate texture storage for whole quads.José Fonseca2009-10-181-3/+11
| |
* | llvmpipe: Use ALIGN_STACK.José Fonseca2009-10-141-0/+1
| |
* | llvmpipe: Eliminate constant mapping/unmapping.José Fonseca2009-10-093-56/+17
| |
* | llvmpipe: Ensure tile cache transfers are mapped before flushing it.José Fonseca2009-10-043-2/+9
| |
* | llvmpipe: Autogenerate lp_tile_soa.c from u_format.csv.José Fonseca2009-10-047-974/+339
| | | | | | | | | | | | | | | | | | This is just a temporary change until we code generate the tile read/write functions in runtime. The new code avoids an extra memcpy that exists in u_tile.c functions, from which lp_tile_soa.c was originally based. This achieves up to 5% improvement, particularly in frames with little geometry overlap.
* | llvmpipe: Remove loop testing from format testing.José Fonseca2009-10-041-6/+0
| | | | | | | | Loop building will be rewritten.
* | llvmpipe: Match header's protection macro with filename.José Fonseca2009-10-041-3/+3
| |
* | llvmpipe: Adjust format assertion.José Fonseca2009-10-041-3/+4
| | | | | | | | We support array layout too -- if it has a single channel.
* | llvmpipe: Fetch tile only if a color buffer is bound.José Fonseca2009-10-032-2/+11
| |
* | gallium: Preparations for adding more PIPE_TRANSFER_* usage flags.Michel Dänzer2009-10-021-11/+3
| | | | | | | | | | | | Always test for PIPE_TRANSFER_READ/WRITE using the bit-wise and operator, and add a pipe_transfer_buffer_flags() helper for getting the buffer usage flags corresponding to them.
* | gallium: remove depth.occlusion_count flagKeith Whitwell2009-10-012-4/+0
| | | | | | | | | | | | | | This was redundant as drivers can just keep track of whether they are inside a begin/end query pair. We want to add more query types later and also support nested queries, none of which map well onto a flag like this. No driver appeared to be using the flag.
* | llvmpipe: Workaround for bug in llvm 2.5.José Fonseca2009-09-291-1/+14
| | | | | | | | | | | | | | | | The combination of fptosi and sitofp (necessary for trunc/floor/ceil/round implementation) somehow becomes invalid code. Skip the instruction combining pass when SSE4.1 is not available.
* | llvmpipe: First verify LLVM IR, only then run optimizing passes.José Fonseca2009-09-291-5/+5
| |
* | llvmpipe: Runtime cpu checks for lp_build_min_simple too.José Fonseca2009-09-291-10/+9
| |
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-20 18:40:28 +0000