[llvm] [RISCV] Add imply rules for zcd/zcf. (PR #91342)

Jessica Clarke via llvm-commits llvm-commits at lists.llvm.org
Fri May 10 10:44:06 PDT 2024


jrtc27 wrote:

> > I don’t think this is valid, Zc* are subsets of C but not technically compatible?
> 
> I think it is indeed correct, and follows what is described in the ISA manual: https://github.com/riscv/riscv-isa-manual/blob/7b8ddc9f58b858265525b59f45bfca961f805b6b/src/zc.adoc#L6
> 
> We haven't gone and e.g. made M imply Zmmul though.
> 
> I think in general we haven't gone back and added these implications when new subset extensions are added as we're cautious of it causing issues with no benefit. We have of course aimed to add such implications when adding new extensions that have these relationships defined and weren't previously supported in LLVM.

Interesting. There was a point when Zmmul wasn't meant to be implied by M, but things have clearly evolved since.

https://github.com/llvm/llvm-project/pull/91342


More information about the llvm-commits mailing list