[all-commits] [llvm/llvm-project] 2e612f: [MLIR][Arith] Improve accuracy of `inferDivU` (#11...

[goldsteinn via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 2e612f8d868b3fb88a44964a3d4efd61ee63e06a
      https://github.com/llvm/llvm-project/commit/2e612f8d868b3fb88a44964a3d4efd61ee63e06a
  Author: goldsteinn <35538541+goldsteinn at users.noreply.github.com>
  Date:   2024-10-29 (Tue, 29 Oct 2024)

  Changed paths:
    M mlir/lib/Interfaces/Utils/InferIntRangeCommon.cpp
    M mlir/test/Dialect/Arith/int-range-interface.mlir

  Log Message:
  -----------
  [MLIR][Arith] Improve accuracy of `inferDivU` (#113789)

1) We can always bound the maximum with the numerator.
    - https://alive2.llvm.org/ce/z/PqHvuT
2) Even if denominator min can be zero, we can still bound the minimum
   result with `lhs.umin u/ rhs.umax`.

This is similar to https://github.com/llvm/llvm-project/pull/110169



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications