1. 28 Nov, 2022 3 commits
  2. 15 Nov, 2022 1 commit
  3. 10 Nov, 2022 1 commit
  4. 08 Nov, 2022 1 commit
    • Craig Topper's avatar
      Cherry-pick D107697 · b3d3bdaa
      Craig Topper authored and rferrer's avatar rferrer committed
      This fixes a bug in the extension of integers in which a multiplication
      with a large integer might overflow and needs to happen within i32.
      b3d3bdaa
  5. 03 Nov, 2022 2 commits
  6. 25 Oct, 2022 1 commit
    • rferrer's avatar
      Fix performance defect · c699d790
      rferrer authored
      We were enabling early clobber for a number of instructions that don't
      need it.
      c699d790
  7. 24 Oct, 2022 3 commits
  8. 03 Oct, 2022 1 commit
  9. 30 Sep, 2022 1 commit
  10. 28 Sep, 2022 5 commits
    • rferrer's avatar
      Add header · 78dd2721
      rferrer authored
      This is in preparation of building documentation while we build the
      compiler
      78dd2721
    • rferrer's avatar
      Fix typo · a46dd837
      rferrer authored
      a46dd837
    • Daniel Trujillo Viedma's avatar
      Cleanup unused/commented code · 9a6e2ed4
      Daniel Trujillo Viedma authored and rferrer's avatar rferrer committed
      9a6e2ed4
    • Daniel Trujillo Viedma's avatar
      Squashed to prevent conflicts in rebase because of existing changes in InsertVSETVLI.cpp · 85f28409
      Daniel Trujillo Viedma authored and rferrer's avatar rferrer committed
      Implement parser/encoder and printer
      
      Update the error message now printed in the invalid tests
      
      Initial implementation of bfloat
      
      Prevent use of 'this' and fix the comparison to include the alternative format bit
      
      Continue the implementation of bfloat16. Stuck at the bitcast conversion from bf16 to i16
      
      Promote bf16 to f32 when the target doesn't support it
      
      This is modeled after the half-precision fp support. Two new nodes are
      introduced for casting from and to bf16. Since casting from bf16 is a
      simple operation I opted to always directly lower it to integer
      arithmetic. The other way round is more complicated if you want to
      preserve IEEE semantics, so it's handled by a new __truncsfbf2
      compiler-rt builtin.
      
      This is of course very bare bones, but sufficient to get a semi-softened
      fadd on x86.
      
      Possible future improvements:
       - Targets with bf16 conversion instructions can now make fp_to_bf16 legal
       - The software conversion to bf16 can be replaced by a trivial
         implementation under fast math.
      
      Differential Revision: https://reviews.llvm.org/D126953
      
      Implement a missing function after cherry-picking, and cleanup lowerSTORE
      
      Fix can't legalize result type
      
      Implement tests with b16 in vsetvli and fix an error with vfnmadd
      
      Implement the alternative b16 format in EPI InstrInfo, and simplify a bit the management of widened type info
      
      Delete X86 and PowerPC tests that don't pass now
      85f28409
    • rferrer's avatar
      Allow widenings outside of OnlyLMUL1 · 1bc258a3
      rferrer authored
      Ideally we'd have a better way to deal with this but for now this should
      do.
      1bc258a3
  11. 21 Sep, 2022 1 commit
    • rferrer's avatar
      Avoid infinite loop · 7186366b
      rferrer authored
      An infinite loop was fixed in EPI, so let's backport this change as
      well.
      7186366b
  12. 20 Sep, 2022 2 commits
  13. 16 Sep, 2022 3 commits
  14. 15 Jul, 2022 6 commits
  15. 20 Jun, 2022 1 commit
  16. 16 Jun, 2022 1 commit
  17. 19 May, 2022 1 commit
    • rferrer's avatar
      Hack to enable min/max reduction detection · 7e3bf27b
      rferrer authored
      When there are no NaNs the comparisons get folded to {max,min}num but
      those were not recognised by IVDescriptor yet. Add a flag to avoid the
      fold so the form is recognised.
      7e3bf27b
  18. 21 Apr, 2022 2 commits
  19. 12 Apr, 2022 1 commit
  20. 01 Apr, 2022 1 commit
  21. 30 Mar, 2022 2 commits