Performance of VM is better than Host

Performance of VM is better than Host

[Roja Eswaran]

[-- Attachment #1.1: Type: text/plain, Size: 515 bytes --]

Hello,

I virtualized Raspi-4 and ran *sysbench  - Memory* on bare metal and the VM.
The performance of VM is better than Host (Increase by 11%)
-> Both host and VM has the same OS
-> No Background Processes are running when sysbench is executed
-> Turned off Transparent huge pages in VM before booting
Yet getting the same performance difference.
Please let me know, What I am missing here

Thank you so much for your valuable time.





*Roja Eswaran*
*Master's in Computer Science*
*Research Assistant - OSNET*

[-- Attachment #1.2: Type: text/html, Size: 1453 bytes --]

[-- Attachment #2: Type: text/plain, Size: 151 bytes --]

_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm

Re: Performance of VM is better than Host

[Marc Zyngier]

[please do not post HTML messages, and stick to plain text]

On 2020-05-17 03:29, Roja Eswaran wrote:
> Hello,
> 
> I virtualized Raspi-4 and ran _sysbench  - Memory_ on bare metal and
> the VM.
> The performance of VM is better than Host (Increase by 11%)
> -> Both host and VM has the same OS
> -> No Background Processes are running when sysbench is executed
> -> Turned off Transparent huge pages in VM before booting
> Yet getting the same performance difference.
> Please let me know, What I am missing here

With this level of information, I can't tell you anything. Funnily 
enough
  you're not mentioning:

- what you are running (exact version and parameters of your benchmark)
- what you are running it on (host configuration, parameters of the VM
   itself and its vmm, as "virualized Raspi-4" doesn't mean much to me)
- which kernel versions
- how you are measuring this discrepancy
- the variability of the measurement

Things you could look at: using perf capture the various counters and
work out by yourself where the difference lies, look at the impact
of power management on the host, the impact of other kernel threads
running, the number of interrupts...

HTH,

         M.
-- 
Jazz is not dead. It just smells funny...
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm