<html><body><p><tt>"Kevin P. Neal" <kpn@neutralgood.org> wrote on 02.11.2016 14:35:57:<br>> On Mon, Oct 31, 2016 at 02:33:30PM -0000, Ulrich Weigand via llvm-<br>> commits wrote:<br>> > - Add optional aliases for the -mcpu processor names according to the<br>> > level of the z/Architecture, for compatibility with other compilers<br>> > on the platform. The supported aliases are:<br>> > -mcpu=arch8 equals -mcpu=z10<br>> > -mcpu=arch9 equals -mcpu=z196<br>> > -mcpu=arch10 equals -mcpu=zEC12<br>> > -mcpu=arch11 equals -mcpu=z13<br>> <br>> This appears to match the naming used by IBM C for z/OS, where support for<br>> the z10 is option ARCH(8), z196 = ARCH(9), etc. Is this on purpose? Does<br>> z/OS count as part of the "platform"? Can we count on these lining up<br>> the same in the future such that IBM C for z/OS documentation can be used<br>> to figure out llvm options in the future?<br></tt><br><tt>Yes, this is on purpose. However, we're primarily looking for</tt><br><tt>compatibility with XLC on Linux on z, not so much with XLC on z/OS,</tt><br><tt>but for the architecture levels, both use the same scheme. We're</tt><br><tt>certainly planning to continue the archX scheme for future machines,</tt><br><tt>but apart from that there's no plan to be fully compatible with all</tt><br><tt>other *z/OS* compiler options (some don't even make sense on Linux).</tt><br><br><tt>Bye,</tt><br><tt>Ulrich</tt><br><BR>
</body></html>