cim vic jader CPU, tim pomalejsi

David Pasek david.pasek at gmail.com
Tue Apr 6 21:41:17 CEST 2021


Ahoj.

On Tue, Apr 6, 2021 at 8:45 PM Miroslav Lachman <000.fbsd at quip.cz> wrote:
>
> Po upgrade VirtualBoxu s 5.22 na 6.1 jsou vysledky vyrazne lepsi nez
> pred upgradem a zaroven o neco lepsi nez Bhyve.

Chapu spravne, ze upgrade VirtualBoxu z 5.22 na 6.1 zlepsil Tvoje
vysledky skoro 3x?

> ## A, FreeBSD 11.4 GENERIC + VBox 6.1.18
> 1 vCPU: Usr: 0.000s  Krnl: 0.005s  Totl: 0:00.00s
> 2 vCPU: Usr: 0.000s  Krnl: 0.296s  Totl: 0:00.29s
> 4 vCPU: Usr: 0.000s  Krnl: 0.947s  Totl: 0:00.94s
> 6 vCPU: Usr: 0.000s  Krnl: 1.902s  Totl: 0:01.90s
> 8 vCPU: Usr: 0.000s  Krnl: 3.073s  Totl: 0:03.07s

Ale porad tam je ten paradox, ze tento workload (loading PF pravidel)
trva ve VM-ku s 1x vCPU  0:00.00s a ve VM-ku s 8x vCPU to trva dele
(0:03.07s).
Chapu to spravne?

Umel bys mi ten test s PF nejak popsat a poslat mi ho take pouze na
muj mail jako Danovi? PF take nepouzivam a 20+ let jsem zustal u IPFW
:-)
Pustil bych to ve FreeBSD 12.1 na VM-kach (1,2,4,6,8 vCPU) na ESXi,
abychom videli, jak se to v tomto pripade chova s vSMP
naimplementovanem v ESXi hypervisoru na relativne starem hardwaru
Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz nebo Intel(R) Xeon(R) CPU
E5-2430 0 @ 2.20GHz

>
> Mirek

David


More information about the Users-l mailing list