[6.9.2] Clock settings not taken into account on VM creation

I've set up an Unraid VM according to this documentation:

https://wiki.unraid.net/Manual/VM_Guest_Support#Unraid

During this process the "Slackware" template is being used. The default settings for the <clock> element in this template is this:

  <clock offset='utc'>
    <timer name='rtc' tickpolicy='catchup'/>
    <timer name='pit' tickpolicy='delay'/>
    <timer name='hpet' present='no'/>
  </clock>

Immediately after boot of this VM I checked the available clocksources:

root@TowerVM01:~# cat /sys/devices/system/clocksource/clocksource0/available_clocksource
acpi_pm

I couldn't believe that because "acpi_pm" is not mentioned anywhere. Looking in the syslog I do see "tsc" problems. This is usual for "tsc" running in a VM, but it's not usual here because "tsc" is not mentioned in the <clock> element at all. So I did another test:

<clock offset='utc'>
    <timer name='tsc' tickpolicy='catchup' mode='auto'/>
    <timer name='rtc' tickpolicy='catchup'/>
    <timer name='pit' tickpolicy='delay'/>
    <timer name='hpet' present='no'/>
</clock>

After that "tsc" is not mentioned in the "available_clocksource" again. Seems that this line was ignored.

After playing around with the <clock> section I came to the conclusion, that the <clock> element is not taken into account at all.

Here's what the same Unraid installation looks like, if running bare bone on host:

root@Tower:~# cat /sys/devices/system/clocksource/clocksource0/available_clocksource
tsc hpet acpi_pm

The "hpet" clock is set to "present='no'" in the VM declaration so I do understand it's absence. But what about "tsc". Whatever I do, it's not mentioned - but temporarely used during boot:

root@TowerVM01:~# grep -i clocks /var/log/syslog
May 18 06:54:02 TowerVM01 kernel: clocksource: refined-jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 1910969940391419 ns
May 18 06:54:02 TowerVM01 kernel: clocksource: tsc-early: mask: 0xffffffffffffffff max_cycles: 0x285cbb610e9, max_idle_ns: 440795254255 ns
May 18 06:54:02 TowerVM01 kernel: clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 1911260446275000 ns
May 18 06:54:02 TowerVM01 kernel: clocksource: timekeeping watchdog on CPU1: Marking clocksource 'tsc-early' as unstable because the skew is too large:
May 18 06:54:02 TowerVM01 kernel: clocksource:                       'refined-jiffies' wd_now: fffb7010 wd_last: fffb6e18 mask: ffffffff
May 18 06:54:02 TowerVM01 kernel: clocksource:                       'tsc-early' cs_now: 21906dee90 cs_last: 20e092a210 mask: ffffffffffffffff
May 18 06:54:02 TowerVM01 kernel: tsc: Marking TSC unstable due to clocksource watchdog
May 18 06:54:02 TowerVM01 kernel: clocksource: Switched to clocksource refined-jiffies
May 18 06:54:02 TowerVM01 kernel: clocksource: acpi_pm: mask: 0xffffff max_cycles: 0xffffff, max_idle_ns: 2085701024 ns
May 18 06:54:02 TowerVM01 kernel: clocksource: Switched to clocksource acpi_pm

The reason why I play around with the <clock> element is, that in the same second of the VM boot, "tsc" is marked unstable and Unraid switches to "acpi_pm". This results in a 75% performance reduction. A parity check goes down from 15 hours (on tsc) to 60 hours (on acpi_pm). I do know these values because it was working 1-2 years ago.

So: What's the problem with the clocks and the <clock> section in a VM. Is this ignored?