Orit Wasserman <owasserm@xxxxxxxxxx> wrote on 21/03/2013 02:22:44 PM: > By the way do you have some performance results, how does it improve nested ? Only the old numbers we obtained "emulating" this type of feature using Nehalem processors and we --including you :)-- published in the Turtles papers: http://static.usenix.org/event/osdi10/tech/slides/ben-yehuda.pdf The results showed that up-to 40% of nested overhead was caused by L0 trapping and emulating L1 vmread/vmwrite instructions. To handle a single L2 exit, L1 performs around 10 vmread/vmwrite instructions (10 exits + 10 entries), so this feature should reduce the virtual exit/entry cost (L2->L1->L2) by at least an order of magnitude. Instead of doing a long chain of entries/exits (L2->L0->L1->L0->L1....L0->L1->L0->L2) we will have a simple and short chain (L2->L0->L1->L0->L2). -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
