[PATCH] D50972: [InstCombine] Fold icmp ugt/ult (add nuw X, C2), C --> icmp ugt/ult X, (C - C2)
Nicola Zaghen via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Mon Aug 20 08:28:33 PDT 2018
Nicola added inline comments.
================
Comment at: test/Transforms/LoopVectorize/runtime-check.ll:20
; CHECK-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64, !dbg !9
; CHECK-NEXT: [[TMP2:%.*]] = add nuw nsw i64 [[TMP1]], 1, !dbg !9
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 3, !dbg !9
----------------
xbolva00 wrote:
> TMP2 is unused?
It is still used on line 36, unfortunately.
Repository:
rL LLVM
https://reviews.llvm.org/D50972
More information about the llvm-commits
mailing list