projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 6329c73) | patch
[mlir][arith] Support wide integer addition emulation
authorJakub Kuderski <kubak@google.com>
Fri, 9 Sep 2022 20:48:00 +0000 (16:48 -0400)
committerJakub Kuderski <kubak@google.com>
Fri, 9 Sep 2022 20:49:38 +0000 (16:49 -0400)
commitc7f64616e9e1f66059bf768324a440bfde98132c
tree2cf1acfe208429730021e43bfb5a6e751763bf03tree | snapshot
parent6329c7387f6b5cad2e2f80136932a297974728c1commit | diff
[mlir][arith] Support wide integer addition emulation

I tested this implementation for all i16 input pairs, when emulating i16
operations with i8 operations.

Reviewed By: antiagainst, Mogball

Differential Revision: https://reviews.llvm.org/D133137
mlir/lib/Dialect/Arithmetic/Transforms/EmulateWideInt.cpp diff | blob | history
mlir/test/Dialect/Arithmetic/emulate-wide-int.mlir diff | blob | history
Domain: System / Toolchain;