projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 7bf3cb3) | patch
[mlir][arith] Convert fastmath to LLVM dialect for some arith ops.
authorSlava Zakharin <szakharin@nvidia.com>
Tue, 8 Nov 2022 03:38:36 +0000 (19:38 -0800)
committerSlava Zakharin <szakharin@nvidia.com>
Tue, 8 Nov 2022 03:39:51 +0000 (19:39 -0800)
commite72fb692104ebab682d1ea5aeec39358b11ad407
tree0a836e486fa032ea4a85e8fe51ca05c0e278e210tree | snapshot
parent7bf3cb3ee871b6707b7f5114a7fde61485df450ecommit | diff
[mlir][arith] Convert fastmath to LLVM dialect for some arith ops.

This is a follow-up on D126305 and D136225.
We can now preserve fastmath for arith::MaxFOp,MinFOp,RemFOp during
ArithToLLVM conversion.

Reviewed By: rriddle

Differential Revision: https://reviews.llvm.org/D137456
mlir/lib/Conversion/ArithToLLVM/ArithToLLVM.cpp diff | blob | history
mlir/test/Conversion/ArithToLLVM/arith-to-llvm.mlir diff | blob | history
Domain: System / Toolchain;