On Wed, 2008-07-16 at 19:57 +0300, Avi Kivity wrote:
> Ben-Ami Yassour wrote:
> >   
> >> That CPU utilization is extremely high and somewhat illogical if native 
> >> w/vt-d has almost no CPU impact.  Have you run oprofile yet or have any 
> >> insight into where CPU is being burnt?
> >>
> >> What does kvm_stat look like?  I wonder if there are a large number of 
> >> PIO exits.  What does the interrupt count look like on native vs. KVM 
> >> with VT-d?
> >>
> >> Regards,
> >>
> >> Anthony Liguori
> >>
> >>     
> >
> > These are all good points and questions, I agree that we need to take a 
> > deeper 
> > look into the performance issues, but I think that we need to merge with 
> > the main KVM tree first.
> >   
> 
> It would be good to get the host interrupt rate, to confirm that the 
> host isn't flooded with interrupts.  A deeper analysis can wait.
> 
> 

Here is an example of a kvm_stat while running iperf:
efer_reload                   13         0
exits                   11372838     58214
fpu_reload                709524         0
halt_exits                139306         0
halt_wakeup                 3580         0
host_state_reload        5070783     29031
hypercalls                635636         1
insn_emulation           4306311     22289
insn_emulation_fail            0         0
invlpg                         0         0
io_exits                  978192       646
irq_exits                2795848     20879
irq_window                     0         0
largepages                     0         0
mmio_exits                 15437         0
mmu_cache_miss             60299         0
mmu_flooded                43213         0
mmu_pde_zapped             49610         0
mmu_pte_updated           576608         1
mmu_pte_write            1203336         1
mmu_recycled                   0         0
mmu_shadow_zapped          59348         0
nmi_window                     0         0
pf_fixed                  571970         0
pf_guest                  103889         0
remote_tlb_flush               0         0
request_irq                    0         0
signal_exits                   1         0
tlb_flush                 725366         7

The average host interrupt rate is around 22,200 per second.
Note: I would assume that this is bigger then irq_exits, because some
of the interrupts occur while we are in host mode so it does not cause
an exit.
BTW, the interrupt counter on the host and guest is very close.

Regards,
Ben



--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to