Useful discussion of the source of the bug initiates here; https://bugzilla.redhat.com/show_bug.cgi?id=651861#c39.
Seems to be caused by the Xen Hypervisor incorrectly handling cmov instructions containing a negative offset.
The (interim) solution for this is to enable "nosegneg" flag https://bugzilla.redhat.com/show_bug.cgi?id=651861#c39 for AMI images - note that this should not be applied generally to all images produced as it would cause a noticeable performance impact.
echo "hwcap 1 nosegneg" > /etc/ld.so.conf.d/libc6-xen.conf
Amazon should eventually fix this issue directly by applying a patch to their underlying xen hypervisor.