On Wed, 2012-07-11 at 14:51 +0300, Avi Kivity wrote: > On 07/11/2012 02:23 PM, Jan Kiszka wrote: > >> > >> I'd appreciate a couple of examples for formality's sake. > > > > From the top of my head: NVIDIA FX3700 (granted, legacy by now), Atheros > > AR9287. For others, I need to check. > > Thanks. > > >> > >>> And then there is not easily replaceable legacy hardware like old > >>> telephony adapters or industrial I/O cards etc. that want this. > >> > >> I imagine legacy hardware will live with the speed of routing through > >> qemu, when running on modern platforms. > > > > Just because it's "legacy" doesn't mean it's "low performance" and "low > > interrupt rate". > > I meant that it was used with lower throughput hardware, so the overhead > of routing through qemu will be masked by the improved host hardware. > But most of the improvement in hardware in recent years is the increase > in core/thread count. > > > We still have classic KVM device assignment to provide fast-path INTx. > > But if we want to replace it midterm, I think it's necessary for VFIO to > > be able to provide such a path as well. > > I would like VFIO to have no regressions vs. kvm device assignment, > except perhaps in uncommon corner cases. So I agree. I ran a few TCP_RR netperf tests forcing a 1Gb tg3 nic to use INTx. Without irqchip support vfio gets a bit more than 60% of KVM device assignment. That's a little bit of an unfair comparison since it's more than just the I/O path. With the proposed interfaces here, enabling irqchip, vfio is within 10% of KVM device assignment for INTx. For MSI, I can actually make vfio come out more than 30% better than KVM device assignment if I send the eventfd from the hard irq handler. Using a threaded handler as the code currently does, vfio is still behind KVM. It's hard to beat a direct call chain. For more devices, one that seems common among the non-enterprise users are TV capture cards, like the old PVR-250/350 devices. These don't support MSI. Thanks, Alex -- 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
