Commits · e3cef93727147603667ec33feba64f6de7955818 · Roger Ferrer / llvm-epi · GitLab

Sep 21, 2016
- [LV] When reporting about a specific instruction without debug location use loop's · e3cef937
  Adam Nemet authored Sep 21, 2016
  e3cef937
Sep 18, 2016
- [Loop Vectorizer] Consecutive memory access - fixed and simplified · 5f8cc0c3
  Elena Demikhovsky authored Sep 18, 2016
  5f8cc0c3
Sep 14, 2016
- [LV] Process pointer IVs with PHINodes in collectLoopUniforms · b25e87fc
  Matthew Simpson authored Sep 14, 2016
  b25e87fc
Sep 13, 2016
- DebugInfo: New metadata representation for global variables. · d4135bbc
  Peter Collingbourne authored Sep 13, 2016
  d4135bbc
Sep 08, 2016
- [LV] Ensure proper handling of multi-use case when collecting uniforms · bfe5e181
  Matthew Simpson authored Sep 08, 2016
  bfe5e181
- [LV] Don't mark pointers used by scalarized memory accesses uniform · 408a3abc
  Matthew Simpson authored Sep 08, 2016
  408a3abc
Sep 02, 2016
- [LV] Ensure reverse interleaved group GEPs remain uniform · b65c230e
  Matthew Simpson authored Sep 02, 2016
  b65c230e
Aug 30, 2016
- [LoopVectorizer] Predicate instructions in blocks with several incoming edges · 2954d1db
  Michael Kuperstein authored Aug 30, 2016
  2954d1db
Aug 29, 2016
- [LV] Move insertelement sequence after scalar definitions · df19502b
  Matthew Simpson authored Aug 29, 2016
  df19502b
Aug 28, 2016
- [Loop Vectorizer] Fixed memory confilict checks. · 3622fbfc
  Elena Demikhovsky authored Aug 28, 2016
  3622fbfc
Aug 24, 2016
- [LV] Unify vector and scalar maps · abd2be1e
  Matthew Simpson authored Aug 24, 2016
  abd2be1e
- [Loop Vectorizer] Support predication of div/rem · 550148b2
  Gil Rapaport authored Aug 24, 2016
  550148b2
Aug 13, 2016
- [LoopVectorize] Detect loops in the innermost loop before creating InnerLoopVectorizer · c9c0d2dc
  Tim Shen authored Aug 12, 2016
  c9c0d2dc
Aug 06, 2016
- [ValueTracking] Teach computeKnownBits about [su]min/max · a19d0f2f
  David Majnemer authored Aug 06, 2016
  a19d0f2f
Aug 05, 2016
- [LV, X86] Be more optimistic about vectorizing shifts. · 3ceac2bb
  Michael Kuperstein authored Aug 04, 2016
  3ceac2bb
Aug 02, 2016
- [LoopVectorize] Change comment for isOutOfScope in collectLoopUniforms, NFC · dc7001af
  Wei Mi authored Aug 02, 2016
  dc7001af
- [LV] Generate both scalar and vector integer induction variables · 18d88983
  Matthew Simpson authored Aug 02, 2016
  18d88983
- [LV] Untangle the concepts of uniform and scalar · 58f56288
  Matthew Simpson authored Aug 02, 2016
  58f56288
- [AVX512] Don't use i128 masked gather/scatter/load/store. Do more accurately dataWidth check. · f44b79d0
  Igor Breger authored Aug 02, 2016
  f44b79d0
Aug 01, 2016
- [AVX-512] Fix a test missed in r277327. · d2b2d745
  Craig Topper authored Aug 01, 2016
  d2b2d745
Jul 29, 2016
- Initial support for vectorization using svml (short vector math library). · a6669a1e
  Matt Masten authored Jul 29, 2016
  a6669a1e
Jul 28, 2016
- Fix the assertion error in collectLoopUniforms caused by empty Worklist before expanding. · 315bb33f
  Wei Mi authored Jul 27, 2016
  315bb33f
Jul 24, 2016
- [Loop Vectorizer] Handling loops FP induction variables. · 376a18bd
  Elena Demikhovsky authored Jul 24, 2016
  376a18bd
Jul 21, 2016
- [LV] Move vector int induction update to end of latch · 102729cf
  Matthew Simpson authored Jul 21, 2016
  102729cf
- [OptDiag,LV] Add hotness attribute to applied-optimization remarks · 7cfd5971
  Adam Nemet authored Jul 21, 2016
  7cfd5971
- [OptDiag,LV] Add hotness attribute to the derived analysis remarks · 0e0e2d5d
  Adam Nemet authored Jul 20, 2016
  0e0e2d5d
Jul 20, 2016
- [OptDiag,LV] Add hotness attribute to analysis remarks · 5b3a5cf6
  Adam Nemet authored Jul 20, 2016
  5b3a5cf6
- [LV] Add hotness attribute to missed-optimization remarks · 67c8929a
  Adam Nemet authored Jul 20, 2016
  67c8929a
Jul 19, 2016
- Recommit the patch "Use uniforms set to populate VecValuesToIgnore". · 79997a24
  Wei Mi authored Jul 19, 2016
  79997a24
Jul 18, 2016
- Revert rL275912. · f9afff71
  Wei Mi authored Jul 18, 2016
  f9afff71
- Use uniforms set to populate VecValuesToIgnore. · 1fd25726
  Wei Mi authored Jul 18, 2016
  1fd25726
Jul 14, 2016
- [LV] Allow interleaved accesses in loops with predicated blocks · 65ca32b8
  Matthew Simpson authored Jul 14, 2016
  65ca32b8
- [LV] Avoid unnecessary IV scalar-to-vector-to-scalar conversions · 3c3b4a25
  Matthew Simpson authored Jul 14, 2016
  3c3b4a25
Jul 12, 2016
- [LV] Remove wrong assumption about LCSSA · a99c46cc
  Michael Kuperstein authored Jul 12, 2016
  a99c46cc
Jul 11, 2016
- [X86] Make some cast costs more precise · f0c59330
  Michael Kuperstein authored Jul 11, 2016
  f0c59330
Jul 10, 2016
- [PM] Port LoopVectorize to the new PM. · db90d4d9
  Sean Silva authored Jul 09, 2016
  db90d4d9
Jul 07, 2016
- Fixed a bug in vectorizing GEP before gather/scatter intrinsic. · fc1e969d
  Elena Demikhovsky authored Jul 07, 2016
  fc1e969d
Jul 06, 2016
- [TTI] The cost model should not assume vector casts get completely scalarized · aa71bdd3
  Michael Kuperstein authored Jul 06, 2016
  aa71bdd3
- [LV] Don't widen trivial induction variables · 433cb1df
  Matthew Simpson authored Jul 06, 2016
  433cb1df
Jun 30, 2016
- SLPVectorizer: Move propagateMetadata to VectorUtils · 727e279a
  Matt Arsenault authored Jun 30, 2016
  727e279a