https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81069
--- Comment #5 from Tom de Vries <vries at gcc dot gnu.org> --- Created attachment 41733 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=41733&action=edit tentative patch, starts neutering alap in nvptx_single > But in this case, we're extending the sese region being neutered to include > a branch that doesn't have to be neutered. I don't think there's a purpose in > doing that. This patch postpone insertion of the diverging branch to alap, by skipping over unconditional jumps at the start of nvptx_single. This allows the test-case to pass (because we don't run into the ptx bug).
