[PATCH] D75509: [mlir] Removed TanHOp lowering from ConvertStandardToLLVM since there is no reasonable TanH representation in LLVM.

[Alexander Belyaev via Phabricator via llvm-commits]

pifon2a added a comment.

@llitchev The Tanh approximation is definitely needed, but the main problem of this very PR was that `StandardToLLVM` was a wrong place for this code to be in.  It might be possible to stay within Standard dialect and make it a transform, i.e.  lower `std.tanh` to std ops and only then lower to LLVM using the `StandardToLLVM` pass.
The could would live then in the same directory with https://github.com/llvm/llvm-project/blob/master/mlir/lib/Dialect/StandardOps/Transforms/ExpandAtomic.cpp.

@ftynse: Alex, do we already have any other math func approximation like for tanh submitted?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D75509/new/

https://reviews.llvm.org/D75509