craig.topper added inline comments.
================
Comment at: llvm/lib/Target/RISCV/RISCVISelLowering.cpp:318
+ if (Subtarget.is64Bit())
+ setOperationAction({ISD::CTLZ, ISD::CTLZ_ZERO_UNDEF}, MVT::i32, Custom);
+ }
----------------
philipp.tomsich wrote:
> craig.topper wrote:
> > without these two lines to promote i32, I suppose we would get zext i32 to
> > i64, ff1, addi? Is the sequence used for ctlzw better than that?
> The sequences will have an identical critical-path (and require the same
> number of temporaries).
> I read your comment as a recommendation to simplify the overall
> implementation (by removing the special case here and in the
> pattern-matching).
Yes that would be my recommendation if there's no reason to prefer one over the
other.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D143439/new/
https://reviews.llvm.org/D143439
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
