summaryrefslogtreecommitdiff
path: root/src/gallium/auxiliary/gallivm/lp_bld_nir.c
Commit message (Expand)AuthorAgeFilesLines
* treewide: Stop lowering legacy atomicsAlyssa Rosenzweig2023-05-161-1/+0
* gallivm: Use unified atomicsAlyssa Rosenzweig2023-05-121-103/+42
* gallivm: Skip loads/stores that are definitely outside of compact vars.Emma Anholt2023-03-291-1/+28
* gallivm: Drop non-scoped barrier handlingAlyssa Rosenzweig2023-03-071-10/+0
* gallivm: Fix handling of nir_intrinsic_scoped_barrierCaio Oliveira2023-02-271-1/+4
* nir: add assertions that loops don't have a Continue ConstructDaniel Schürmann2023-02-211-0/+1
* nir_lower_fp16_casts: Allow opting out of lowering certain rounding modesJesse Natalie2023-02-111-1/+1
* nir: Eliminate nir_op_f2bIan Romanick2023-02-031-20/+0
* gallivm: Fix the type of array nir_registers.Emma Anholt2023-02-031-2/+2
* nir: Eliminate nir_op_i2bIan Romanick2022-12-141-24/+0
* gallivm: fix incorrect type for undefined texcoordsBrian Paul2022-12-091-1/+1
* gallivm: misc clean-ups in lp_bld_nir.cBrian Paul2022-12-091-26/+59
* gallivm: use valid LLVM_VERSION_MAJOR instead of LLVM_VERSIONYonggang Luo2022-11-241-1/+1
* gallivm/nir: fix fmin/fmax translationDave Airlie2022-09-271-2/+2
* gallivm: fix nir AOS swizzling issuesBrian Paul2022-09-211-21/+50
* llvmpipe: always pass non-zero writemask to assign_reg()Brian Paul2022-09-211-2/+4
* gallivm: add atomic 32-bit float supportDave Airlie2022-09-201-0/+23
* gallivm/nir/st: lower image derefs in advance.Dave Airlie2022-08-051-85/+59
* llvmpipe: add shader clock supportDave Airlie2022-08-041-0/+3
* gallivm: fix casting around read invocation.Dave Airlie2022-07-271-3/+5
* gallivm: use progress from subgroup lowering.Dave Airlie2022-07-271-2/+1
* gallivm: Optimize single-invocation SSBO stores.Emma Anholt2022-07-261-2/+4
* gallivm: asst. clean-ups in lp_bld_nir.cBrian Paul2022-06-151-232/+461
* gallivm: add subgroup shuffle supportAutumn on Tape2022-05-031-0/+20
* nir/lower_tex: Make the adding a 0 LOD to nir_op_tex in the VS optional.Emma Anholt2022-04-281-1/+2
* gallivm/nir: Add a short circuit uniform-offset mode for load_ssbo/load_shared.Emma Anholt2022-04-061-2/+4
* gallivm/nir: Add a short circuit uniform-offset mode for load_global.Emma Anholt2022-04-061-1/+2
* nir: rename nir_src_is_dynamically_uniform to nir_src_is_always_uniformDaniel Schürmann2022-03-231-4/+4
* llvmpipe: set nir_shader_compiler_options::use_scoped_barrierMike Blumenkrantz2022-03-151-0/+1
* gallivm: implement nir_op_pack_32_4x8_splitMike Blumenkrantz2022-03-151-0/+9
* gallivm/llvmpipe: add support for NIR to the linear/aos paths.Dave Airlie2022-03-031-6/+40
* gallivm/nir: split load_const out into backend helper.Dave Airlie2022-03-031-4/+1
* gallivm/nir: Call nir_lower_bool_to_int32 after nir_opt_algebraic_lateIan Romanick2022-02-101-3/+6
* llvmpipe: add missing NIR alu-op handlingPierre-Eric Pelloux-Prayer2021-11-041-0/+21
* gallivm: use llvm intrinsics for 16-bit round/trunc/roundevenDave Airlie2021-09-161-1/+8
* gallivm/nir: pass the correct float builder to ddx/yDave Airlie2021-09-161-2/+2
* gallivm/nir: call pow with correct flt builderDave Airlie2021-09-161-1/+1
* gallivm/nir: handle 16-bit exp/lod using intrinsics.Dave Airlie2021-09-161-2/+2
* gallivm: add 16-bit sin/cos via llvm intrinsicDave Airlie2021-09-161-2/+2
* gallivm/nir: fix f2b32Dave Airlie2021-09-161-0/+2
* gallivm/nir: handle conversion to 16-bit texel fetchDave Airlie2021-09-161-3/+13
* gallivm: add initial support for 16-bit float builder.Dave Airlie2021-09-161-3/+19
* gallivm/ssbo: cast ssbo index to int type.Dave Airlie2021-09-161-4/+4
* gallivm: fix texture-mapping with 16-bit resultErik Faye-Lund2021-08-181-0/+20
* gallivm: fix non-32 bit popcounts.Dave Airlie2021-08-161-0/+4
* gallivm: fix idiv/irem for 8/16/64-bit and 32-bit INT_MIN/-1Dave Airlie2021-08-161-7/+49
* gallivm: don't lower local invocation index in frontendDave Airlie2021-08-161-0/+1
* gallivm: handle fisfinite/fisnormalDave Airlie2021-08-061-0/+3
* gallivm: let nir_lower_tex handle projectorsErik Faye-Lund2021-07-301-11/+2
* llvmpipe: fix nir dot products (fsum op)Roland Scheidegger2021-07-271-1/+1