On Wed, 2010-03-31 at 12:51 +0300, Michael S. Tsirkin wrote:
> On Tue, Mar 30, 2010 at 04:48:25PM -0700, Sridhar Samudrala wrote:
> > This patch increases the current hardcoded limit of NR_IOBUS_DEVS
> > from 6 to 200. We are hitting this limit when creating a guest with more
> > than 1 virtio-net device using vhost-net backend. Each virtio-net
> > device requires 2 such devices to service notifications from rx/tx queues.
> >
> > Signed-off-by: Sridhar Samudrala <s...@us.ibm.com>
> >
>
> I tried this, but observed a measurable performance
> degradation with vhost net and this patch. Did not
> investigate yet. Do you see this as well?
No. I am not seeing any degradation that is not within the range of
variation seen with kvm networking.
For ex: On a Nehalem 2 socket, 8 core system, running netperf TCP_STREAM
with 64K size packets, i am getting 12-13Gb/s guest to host and
11-12Gb/s host to guest with and without this patch.
Thanks
Sridhar
>
> > diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> > index a3fd0f9..7fb48d3 100644
> > --- a/include/linux/kvm_host.h
> > +++ b/include/linux/kvm_host.h
> > @@ -54,7 +54,7 @@ extern struct kmem_cache *kvm_vcpu_cache;
> > */
> > struct kvm_io_bus {
> > int dev_count;
> > -#define NR_IOBUS_DEVS 6
> > +#define NR_IOBUS_DEVS 200
> > struct kvm_io_device *devs[NR_IOBUS_DEVS];
> > };
> >
> >
> >
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
