| Commit message (Expand) | Author | Age | Files | Lines |
* | gallivm,ac: add LP_FUNC_ATTR_CONVERGENT | Marek Olšák | 2017-03-06 | 2 | -0/+2 |
* | gallivm, ac: add writeonly and inaccessiblememonly attributes | Marek Olšák | 2017-03-03 | 2 | -0/+4 |
* | gallivm,ac: add function attributes at call sites instead of declarations | Marek Olšák | 2017-03-01 | 2 | -24/+55 |
* | gallivm,ac: remove unused FUNC_ATTR_LAST enums | Marek Olšák | 2017-03-01 | 1 | -1/+0 |
* | gallivm: add no-signed-zeros-fp-math option to lp_create_builder (v2) | Marek Olšák | 2017-02-21 | 2 | -4/+19 |
* | gallivm: Reenable PPC VSX (v3) | Ben Crocker | 2017-02-20 | 1 | -1/+13 |
* | gallivm: Override getHostCPUName() "generic" w/ "pwr8" (v4) | Ben Crocker | 2017-02-20 | 1 | -0/+13 |
* | gallivm: Improve debug output (V2) | Ben Crocker | 2017-02-20 | 2 | -1/+18 |
* | gallium: remove TGSI_OPCODE_CLAMP | Marek Olšák | 2017-02-18 | 2 | -24/+0 |
* | gallium: turn PIPE_SHADER_CAP_DOUBLES into a screen capability | Nicolai Hähnle | 2017-02-02 | 1 | -2/+0 |
* | gallivm: remove explicit __STDC_.*_MACROS defines | Emil Velikov | 2017-01-27 | 1 | -8/+0 |
* | gallivm: don't try to use fast rcp for fdiv | Roland Scheidegger | 2017-01-24 | 1 | -1/+3 |
* | gallivm: (trivial) fix ddiv cpu implementation | Roland Scheidegger | 2017-01-24 | 1 | -1/+0 |
* | gallivm: use #ifdef not #if for PIPE_ARCH_BIG_ENDIAN | Dave Airlie | 2017-01-19 | 1 | -1/+1 |
* | gallivm: (trivial) fix copy/paste bug with big endian code | Roland Scheidegger | 2017-01-18 | 1 | -2/+4 |
* | gallivm: Cleanup USE_MCJIT. | Jose Fonseca | 2017-01-18 | 1 | -10/+25 |
* | tgsi: add DDIV instruction | Nicolai Hähnle | 2017-01-16 | 1 | -0/+2 |
* | gallivm: generalize 4x4f->1x16ub special case conversion | Roland Scheidegger | 2017-01-06 | 1 | -56/+118 |
* | gallivm: (trivial) fix typo bug with small AoS format unpacking | Roland Scheidegger | 2017-01-06 | 1 | -1/+1 |
* | gallivm: implement aos unpack (to unorm8) for small unorm formats | Roland Scheidegger | 2017-01-05 | 1 | -12/+152 |
* | gallivm: optimize lp_build_unpack_arith_rgba_aos slightly | Roland Scheidegger | 2017-01-05 | 1 | -19/+97 |
* | gallivm: use 2 srcs for 32->16bit conversions in lp_bld_conv_auto | Roland Scheidegger | 2017-01-05 | 1 | -2/+19 |
* | llvmpipe: (trivial) minimally simplify mask construction | Roland Scheidegger | 2017-01-05 | 1 | -0/+2 |
* | gallium: remove TGSI_OPCODE_SUB | Marek Olšák | 2017-01-05 | 2 | -39/+5 |
* | gallium: remove TGSI_OPCODE_ABS | Marek Olšák | 2017-01-05 | 3 | -17/+3 |
* | swr: fix windows build break | George Kyriazis | 2017-01-05 | 1 | -0/+7 |
* | gallivm: generalize the compressed format soa fetch a bit | Roland Scheidegger | 2016-12-21 | 1 | -37/+49 |
* | gallivm: provide soa fetch path handling formats with more than 32bit | Roland Scheidegger | 2016-12-21 | 1 | -154/+375 |
* | gallivm: optimize gather a bit, by using supplied destination type | Roland Scheidegger | 2016-12-21 | 7 | -78/+332 |
* | gallivm: optimize SoA AoS fallback fetch path a little | Roland Scheidegger | 2016-12-21 | 1 | -22/+46 |
* | gallivm: (trivial) handle non-aligned fetch for lp_build_fetch_rgba_soa | Roland Scheidegger | 2016-12-21 | 3 | -8/+12 |
* | gallivm: optimize 16bit->32bit gather path a bit | Roland Scheidegger | 2016-12-06 | 1 | -3/+39 |
* | gallivm: handle 16bit float fetches in lp_build_fetch_rgba_soa | Roland Scheidegger | 2016-12-06 | 1 | -4/+18 |
* | gallivm: use getHostCPUFeatures on x86/llvm-4.0+. | Tim Rowley | 2016-12-05 | 1 | -0/+15 |
* | gallium: add PIPE_SHADER_CAP_LOWER_IF_THRESHOLD | Marek Olšák | 2016-11-15 | 1 | -0/+1 |
* | gallivm: limit use of setFastMathFlags to LLVM 3.8 and later | Marek Olšák | 2016-11-15 | 1 | -0/+2 |
* | gallivm: add lp_create_builder with an unsafe_fpmath option | Marek Olšák | 2016-11-15 | 2 | -0/+17 |
* | gallivm: fix [IU]MUL_HI regression harder | Nicolai Hähnle | 2016-11-10 | 1 | -8/+12 |
* | gallivm: Fix build after removal of deprecated attribute API v3 | Tom Stellard | 2016-11-09 | 3 | -4/+87 |
* | gallivm: fix [IU]MUL_HI regression | Nicolai Hähnle | 2016-11-08 | 3 | -28/+90 |
* | gallivm: introduce 32x32->64bit lp_build_mul_32_lohi function | Roland Scheidegger | 2016-11-08 | 3 | -38/+172 |
* | gallivm: try to fix build with LLVM <= 3.4 due to missing CallSite.h | Marek Olšák | 2016-10-20 | 1 | -1/+5 |
* | gallivm: add wrappers for missing functions in LLVM <= 3.8 | Marek Olšák | 2016-10-20 | 2 | -0/+27 |
* | draw: improve vertex fetch (v2) | Roland Scheidegger | 2016-10-19 | 2 | -0/+30 |
* | gallivm: print out time for jitting functions with GALLIVM_DEBUG=perf | Roland Scheidegger | 2016-10-19 | 1 | -0/+11 |
* | gallivm: Use native packs and unpacks for the lerps | Roland Scheidegger | 2016-10-19 | 3 | -13/+156 |
* | gallivm: Use AVX2 gather instrinsics. | Jose Fonseca | 2016-10-04 | 1 | -0/+95 |
* | gallivm: Use 8 wide AoS sampling on AVX2. | Roland Scheidegger | 2016-10-04 | 1 | -5/+6 |
* | gallivm: Basic AVX2 support. | José Fonseca | 2016-10-04 | 4 | -28/+98 |
* | gallivm: support negation on 64-bit integers | Nicolai Hähnle | 2016-09-21 | 1 | -0/+4 |