[PATCH] D102606: [RISCV] Fix operand order in fixed-length VM(OR|AND)NOT patterns

Fraser Cormack via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue May 18 01:53:34 PDT 2021


frasercrmck added a comment.

In D102606#2764127 <https://reviews.llvm.org/D102606#2764127>, @craig.topper wrote:

> LGTM.  Let's try to get it into the 12.0.1 release.

My bad, it seems that the 12 release doesn't have the fixed-length RVV support. Time flies.

In D102606#2764964 <https://reviews.llvm.org/D102606#2764964>, @jrtc27 wrote:

> Does the `vs2, vs1` operand order in the spec make sense? Would `vs1, vs2` not be more natural?

To me, `vs1,vs2` would certainly be more natural and it is indeed `vs1,vs2` in a select few operations. To be honest, I can't think of a way that it //has// to be the way it is. Perhaps it's more consistent and requires fewer unique operand names across the full instruction set.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D102606/new/

https://reviews.llvm.org/D102606



More information about the llvm-commits mailing list