Commits · f02ac0eeb2beadef8d5fe5e480dbbf64150b214a · Roger Ferrer / llvm-epi · GitLab

Jan 12, 2017
- [X86] Replace AND+IMM64 with SRL/SHL · f02ac0ee
  Nikolai Bozhenov authored Jan 12, 2017
  f02ac0ee
- [X86] Modify BypassSlowDivision tests to match their new names (NFC) · 3db8bcdb
  Nikolai Bozhenov authored Jan 12, 2017
  3db8bcdb
- [X86] Rename tests for bypassing slow division (NFC) · 6684aeb1
  Nikolai Bozhenov authored Jan 12, 2017
  6684aeb1
- [X86] Tune bypassing of slow division for Intel CPUs · 6bdf92ce
  Nikolai Bozhenov authored Jan 12, 2017
  6bdf92ce
- [X86] Update LLC tests for slow division bypassing (NFC) · 05b40959
  Nikolai Bozhenov authored Jan 12, 2017
  05b40959
- AMDGPU: Skip fneg/select combine if it can fold into other · 45337df0
  Matt Arsenault authored Jan 12, 2017
  45337df0
- AMDGPU: Fold free fneg into sin · 31c039ef
  Matt Arsenault authored Jan 12, 2017
  31c039ef
- AMDGPU: Fold fneg into fmul_legacy · a8c325e2
  Matt Arsenault authored Jan 12, 2017
  a8c325e2
- AMDGPU: Fold fneg into rcp · ff7e5aad
  Matt Arsenault authored Jan 12, 2017
  ff7e5aad
- AMDGPU: Fold fneg into fp_round · 4242d48c
  Matt Arsenault authored Jan 12, 2017
  4242d48c
- AMDGPU: Fold fneg into fp_extend · 98d2bf10
  Matt Arsenault authored Jan 12, 2017
  98d2bf10
- [AVX-512] Improve lowering of zero_extend of v4i1 to v4i32 and v2i1 to v2i64... · 24c3a239
  Craig Topper authored Jan 12, 2017
  24c3a239
- [AVX-512] Improve lowering of sign_extend of v4i1 to v4i32 and v2i1 to v2i64... · 69ab67b2
  Craig Topper authored Jan 12, 2017
  69ab67b2
- [X86][AVX512] Fix PR31515 - Do not flip vselect condition if it's not a vXi1 mask · c5ba925e
  Elad Cohen authored Jan 12, 2017
  c5ba925e
- [AVX-512] Add more varied avx512 feature command lines to the avx512-cvt.ll... · 56f9610b
  Craig Topper authored Jan 12, 2017
  56f9610b
- AMDGPU: Fold fneg into fma or fmad · 63f95379
  Matt Arsenault authored Jan 12, 2017
  63f95379
- AMDGPU: Fold fneg into fmul · 4103a81d
  Matt Arsenault authored Jan 12, 2017
  4103a81d
- AMDGPU: Fold fneg into fadd · 2529fba9
  Matt Arsenault authored Jan 12, 2017
  2529fba9
- AMDGPU: Pull fneg/fabs out of a select · 2a04ff97
  Matt Arsenault authored Jan 11, 2017
  2a04ff97
Jan 11, 2017
- AMDGPU: Fix shrinking of addc/subb. · 24a1273a
  Matt Arsenault authored Jan 11, 2017
  24a1273a
- AMDGPU: Fix sext_inreg for i1 in i16 · 682eb439
  Matt Arsenault authored Jan 11, 2017
  682eb439
- AMDGPU: Fix breaking VOP3 v_add_i32s · 28bd4cbe
  Matt Arsenault authored Jan 11, 2017
  28bd4cbe
- AMDGPU: Fix folding immediates into mac src2 · 69e3001b
  Matt Arsenault authored Jan 11, 2017
  69e3001b
- Revert "CodeGen: Allow small copyable blocks to "break" the CFG." · efe56fed
  Kyle Butt authored Jan 11, 2017
  efe56fed
- [ARM] More aggressive matching for vpadd and vpaddl. · 3a03742c
  Eli Friedman authored Jan 11, 2017
  3a03742c
- [X86][XOP] Add vpermil2ps target shuffle -> insertps combine test · fd627bf8
  Simon Pilgrim authored Jan 11, 2017
  fd627bf8
- [ARM] Fix test CodeGen/ARM/fpcmp_ueq.ll broken by rL290616 · 56f5b893
  Evgeny Astigeevich authored Jan 11, 2017
  56f5b893
- X86 CodeGen: Optimized pattern for truncate with unsigned saturation. · 9d0e7c33
  Elena Demikhovsky authored Jan 11, 2017
  9d0e7c33
- [X86][AVX512BW] Vectorize v64i8 vector shifts · 5a81fefa
  Simon Pilgrim authored Jan 11, 2017
  5a81fefa
- [X86] Fix PR30926 - Add patterns for (v)cvtsi2s{s,d} and (v)cvtsd2s{s,d} · 0c260107
  Elad Cohen authored Jan 11, 2017
  0c260107
- Revert r291645 "[DAGCombiner] Teach DAG combiner to fold (vselect (N0 xor... · 7af39837
  Craig Topper authored Jan 11, 2017
  7af39837
- [DAGCombiner] Teach DAG combiner to fold (vselect (N0 xor AllOnes), N1, N2) ->... · 577d2585
  Craig Topper authored Jan 11, 2017
  577d2585
- DAGCombiner: Add hasOneUse checks to fadd/fma combine · e482403e
  Matt Arsenault authored Jan 11, 2017
  e482403e
- Re-commit r289955: [X86] Fold (setcc (cmp (atomic_load_add x, -C) C), COND) to... · 6573976f
  Hans Wennborg authored Jan 11, 2017
  6573976f
- [X86] Dont run combineSetCCAtomicArith() when the cmp has multiple uses · 12de6937
  Hans Wennborg authored Jan 11, 2017
  12de6937
- AMDGPU/EG,CM: Add fp16 conversion instructions · 0d6cb1ca
  Jan Vesely authored Jan 11, 2017
  0d6cb1ca
- [TM] Restore default TargetOptions in TargetMachine::resetTargetOptions. · 7d81813d
  Justin Lebar authored Jan 10, 2017
  7d81813d
- [NVPTX] Add CHECK-LABEL where appropriate to fast-math.ll test. · c1e2d97a
  Justin Lebar authored Jan 10, 2017
  c1e2d97a
- AMDGPU: Constant fold when immediate is materialized · 51818c14
  Matt Arsenault authored Jan 10, 2017
  51818c14
- CodeGen: Allow small copyable blocks to "break" the CFG. · df27aa8c
  Kyle Butt authored Jan 10, 2017
  df27aa8c