Commits · f8853ae7b3a79394c4989df098f04c84bca28ad9 · Roger Ferrer / llvm-epi · GitLab

Feb 01, 2016
- [InstCombine] Don't transform (X+INT_MAX)>=(Y+INT_MAX) -> (X<=Y) · f8853ae7
  David Majnemer authored Feb 01, 2016
  f8853ae7
- [ValueTracking] Improve isKnownNonZero for PHI of non-zero constants · ca832660
  Jun Bum Lim authored Feb 01, 2016
  ca832660
- [InstCombine] simplify masked load intrinsics with all ones or zeros masks · b695c555
  Sanjay Patel authored Feb 01, 2016
  b695c555
Jan 30, 2016
- InstCombine: fabs(x) * fabs(x) -> x * x · 56c079f3
  Matt Arsenault authored Jan 30, 2016
  56c079f3
Jan 29, 2016
- [InstCombine] avoid an insertelement transformation that induces the opposite... · 66fff73c
  Sanjay Patel authored Jan 29, 2016
  66fff73c
Jan 28, 2016
- add masked intrinsic tests to show missed opportunities · 8123f919
  Sanjay Patel authored Jan 28, 2016
  8123f919
Jan 26, 2016
- [LibCallSimplifier] fold memset(malloc(x), 0, x) --> calloc(1, x) · 980b280f
  Sanjay Patel authored Jan 26, 2016
  980b280f
Jan 22, 2016
- AMDGPU: Rename intrinsics to use amdgcn prefix · bef34e21
  Matt Arsenault authored Jan 22, 2016
  bef34e21
Jan 21, 2016
- [LibCallSimplifier] don't get fooled by a fake fmin() · fcc7c1a0
  Sanjay Patel authored Jan 21, 2016
  fcc7c1a0
Jan 20, 2016
- [LibCallSimplifier] don't get fooled by a fake sqrt() · bd2dc671
  Sanjay Patel authored Jan 20, 2016
  bd2dc671
- add tests to show missing memset/malloc optimizations (PR25892) · 582857c9
  Sanjay Patel authored Jan 19, 2016
  582857c9
Jan 19, 2016
- [LibCallSimplifier] use instruction-level fast-math-flags to shrink calls · d1f4f03f
  Sanjay Patel authored Jan 19, 2016
  d1f4f03f
- [LibCallSimplifier] use instruction-level fast-math-flags to transform pow(x,... · 81a63cd1
  Sanjay Patel authored Jan 19, 2016
  81a63cd1
Jan 17, 2016
- Push isDereferenceableAndAlignedPointer down into isSafeToLoadUnconditionally · f84dc06e
  Artur Pilipenko authored Jan 17, 2016
  f84dc06e
Jan 15, 2016
- Re-commit r257064, after it was reverted in r257340. · f29dfd36
  Silviu Baranga authored Jan 15, 2016
  f29dfd36
- [InstCombine] Rewrite bswap/bitreverse handling completely. · f01488e2
  James Molloy authored Jan 15, 2016
  f01488e2
Jan 12, 2016
- [LibCallSimplifier] use instruction-level fast-math-flags to transform pow(x, 0.5) calls · 53ba88db
  Sanjay Patel authored Jan 12, 2016
  53ba88db
- [LibCallSimplifier] use instruction-level fast-math-flags to transform pow(exp(x)) calls · 6002e78a
  Sanjay Patel authored Jan 12, 2016
  6002e78a
- consolidate exp/exp2 tests · dc19b8de
  Sanjay Patel authored Jan 12, 2016
  dc19b8de
- Add/edit tests to include instruction-level FMF on calls · 71e550fe
  Sanjay Patel authored Jan 12, 2016
  71e550fe
- [LibCallSimplifier] use instruction-level fast-math-flags to transform log calls · e896ede7
  Sanjay Patel authored Jan 11, 2016
  e896ede7
Jan 11, 2016
- [LibCallSimplifier] don't allow sqrt transform unless all ops are unsafe · 6c1ddbb7
  Sanjay Patel authored Jan 11, 2016
  6c1ddbb7
- [LibCallSimplifier] use instruction-level fast-math-flags to transform sqrt calls · 683f2973
  Sanjay Patel authored Jan 11, 2016
  683f2973
- Revert r257164 - it has caused spec2k6 failures in LTO mode · 603954ef
  Silviu Baranga authored Jan 11, 2016
  603954ef
Jan 08, 2016
- Re-commit r257064, this time with a fixed assert · 9e007efa
  Silviu Baranga authored Jan 08, 2016
  9e007efa
- [InstCombine] insert a new shuffle in a safe place (PR25999) · d72a458d
  Sanjay Patel authored Jan 08, 2016
  d72a458d
Jan 07, 2016
- Add test for r256912 · 867bbc77
  David Majnemer authored Jan 07, 2016
  867bbc77
- Revert r257064. It caused failures in some sanitizer tests. · dd68d46e
  Silviu Baranga authored Jan 07, 2016
  dd68d46e
- [InstCombine] Look through PHIs, GEPs, IntToPtrs and PtrToInts to expose more... · 57b1b909
  Silviu Baranga authored Jan 07, 2016
  57b1b909
Jan 06, 2016
- [LibCallSimplifier] use instruction-level fast-math-flags for tan/atan transform · cddcd725
  Sanjay Patel authored Jan 06, 2016
  cddcd725
Jan 05, 2016
- [LibCallSimplfier] use instruction-level fast-math-flags for fmin/fmax transforms · 29095ea1
  Sanjay Patel authored Jan 05, 2016
  29095ea1
- [InstCombine] insert a new shuffle before its uses (PR26015) · a1c53479
  Sanjay Patel authored Jan 05, 2016
  a1c53479
- [InstructionCombining] prepareICWorklistFromFunction halts in infinite loop... · c6021038
  Chen Li authored Jan 04, 2016
  c6021038
Jan 01, 2016
- [LibCallSimplifier] propagate FMF when shrinking binary calls · bee05caa
  Sanjay Patel authored Dec 31, 2015
  bee05caa
Dec 31, 2015
- [LibCallSimplifier] propagate FMF when shrinking unary calls · aa23114c
  Sanjay Patel authored Dec 31, 2015
  aa23114c
- change function names to avoid accidentally matching the substring · f6f32bca
  Sanjay Patel authored Dec 31, 2015
  f6f32bca
- add 'fast' attribute to calls to show that the flag isn't being propagated · 4e8b3004
  Sanjay Patel authored Dec 31, 2015
  4e8b3004
Dec 27, 2015
- [attrs] Extract the pure inference of function attributes into · 3a040e6d
  Chandler Carruth authored Dec 27, 2015
  3a040e6d
Dec 26, 2015
- [gc.statepoint] Change gc.statepoint intrinsic's return type to token type instead of i32 type · d71999ef
  Chen Li authored Dec 26, 2015
  d71999ef
Dec 24, 2015
- [InstCombine] transform more extract/insert pairs into shuffles (PR2109) · ae945e79
  Sanjay Patel authored Dec 24, 2015
  ae945e79