... > "Phenom" doesn't exist in the /usr/share/libvirt/cpu_map.xml, therefore > libvirt? chooses probably the closest to specification, > and it is always Opteron_G3, even if cpu mode='host-passthrough'. > Is this the culprit, dunno, but both systems do have a tendency to "hose". > And this is nothing new, it is ongoing long since. >
To correct myself, there is "phenom" in the 'cpu_map.xml', although this is not the second generation. However the question is whether it is at all important to libvirt/qemu given the limited number of used properties. Also, this can be surpassed by cpu mode='host-passthrough' although even that is not fully applied, probably related to the limitation of the qemu BIOS i.e. due to the lack of susceptibility to "subtle" differences in socket - core relation. But even that did not help. Finally, one small thing helped, and this is not the daisy. 3.18.1-2.fc22.x86_64 Someone once said, the Earth is round. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
