https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123870
--- Comment #6 from Jeffrey A. Law <law at gcc dot gnu.org> --- Yes, the thead stuff is not in good shape; I don't think we're at the point of recommending its removal (yet) as it's largely stayed out of the way. However, I think both Robin and I are concerned about the maintainability of that code. As for these specific issues, I do recall fixing on vmvNr issue on thead-vector, it was some time ago, so definitely possible I fixed in gcc-15, but never backported to gcc-14. Though IIRC it was related to multiply-accumulates which I don't think are in play here.
