UEFI mode doesn’t work on default settings, Installation workaround needed (Do not create sys-usb on first config, do it manually attaching one of usb-controllers in permissive mode, described in Qubes Docs). Integrated Graphics and TPM not tested (but present).
To be easier to solve mentioned sys-usb problem, here is exactly part of documentation you need
Are you still happy with your system and the compatibility with Qubes?
I’m currently trying to find a good Qubes-compatible AMD system to buy and was considering your setup.
Sorry for long absence, I was reeealy busy
Yes, I’m really happy with this configuration as it will probably have future support for new processors series. It’s easy extensible, painless (e.g. no nvidia), maybe not the best for now but in reasonable price. That’s also why I chose it.
If you are audiophile only bad thing is DAC. I could hear difference comparing to just hifiberry! so yeah… If you’re going to listen music in good quality you should buy external DAC.
Well, there are no problems without workaround as mentioned above. I couldn’t boot it in UEFI mode (so just changed to BIOS) and sys-usb needs to be created after install as not supporting PCI reset breaks installation if you do it automatically before first access to qubes.
Another good thing is speed. OS Booting up is slightly faster than e.g my TP500LAB
But the best thing is appVM boot time. It’s only 6 seconds! (Comparing to TP500LAB 10-20, I don’t remember precisely) then app in appVM starts immediately.
(I have m2 disk 32 gigs of ram and ryzen 7 3800x)
Have you ever tried Qubes before or do you know its construction (appVM etc.)? If not I’ll try to explain it more precisely.
If you have any questions, just mark me somehow, so I’ll get notification.
Epic! If you have the chance, add this result to this table where we’re collecting VM boot times:
How did you measure?
I do this:
[user@dom0] date -R && qvm-run --pass-io personal-web “date -R”
Sun, 27 Jun 2021 16:57:21 -0500
Sun, 27 Jun 2021 16:57:29 -0500
I tried with several different qubes and get reliably 8 seconds on a
ThinkPad T430 with i7-3840.
I’ll try in meantime
Firstly I just counted… But your command output tells me same:
(Tested on fedora-33 appVMs)
[<user>@dom0] date -R && qvm-run --pass-io <my-secret-qube-name> “date -R” Fri, 30 Jul 2021 19:33:33 +0200 Fir, 30 Jul 2021 19.33.39 +0200
After shutdown I retested a bit later using same command
[<user>@dom0] date -R && qvm-run --pass-io <my-secret-qube-name> “date -R” Fri, 30 Jul 2021 19:41:57 +0200 Fir, 30 Jul 2021 19.42.02 +0200
Then I picked another qube
[<user>@dom0] date -R && qvm-run --pass-io <my-second-secret-qube-name> “date -R” Fri, 30 Jul 2021 19:42:58 +0200 Fir, 30 Jul 2021 19.43.05 +0200
Then shutdown and start again:
[<user>@dom0] date -R && qvm-run --pass-io <my-second-secret-qube-name> “date -R” Fri, 30 Jul 2021 19:43:25 +0200 Fir, 30 Jul 2021 19.43.31 +0200
Average is 6 sec.
I’ll do precise tests later.
TBH I’ve bought parts thinking about qubes and future with small risk of incompatibility as vendor docs weren’t enough exhaustive.
My first HCL report probably contains other ssd than I use because I’ve changed hardware (added RAM - same model, ssd is now m2 WD black) I’ll regenerate it later with speedtest.