[Mlir-commits] [mlir] [mlir][SCF] Skip dynamic front-peeling for non-index loops (PR #218238)
Matthias Springer
llvmlistbot at llvm.org
Sun Aug 23 08:44:39 PDT 2026
================
@@ -236,6 +236,11 @@ LogicalResult mlir::scf::peelForLoopFirstIteration(RewriterBase &b, ForOp forOp,
if (lbInt && ubInt && stepInt && ceil(float(*ubInt - *lbInt) / *stepInt) <= 1)
return failure();
+ // The peeling bound (%lb + %step) is computed with affine.apply, which
+ // accepts only index operands.
----------------
matthias-springer wrote:
nit: Add comment why you are checking for `(!lbInt || !stepInt)`.
https://github.com/llvm/llvm-project/pull/218238
More information about the Mlir-commits
mailing list