https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87064
--- Comment #4 from Janne Blomqvist <jb at gcc dot gnu.org> --- Well, that is, ahem, interesting. So it seems that OACC does the vector_length reduction incorrectly; the correct result is obviously 10.0 and not 9.0 which then causes the failure. I'm a bit lost why that might happen, but in the time-honored fashion of shifting blame I suggest this is some latent bug in OACC rather than a bug in r263751.
