Return to Level1Techs.com

New PC, new VFIO Setup (EDITED)

#1

Hi there,
So… I find myself in quite a pickle. Here is the deal, I’ve been using GPU passtrough since Wendell’s first video on GTA V. Luckily, my PC supported so I went for it and got it working in a week or two without much Linux Knowledge.
At the point of writing this post my MSI 970 is either (1) Dying or (2) malfunctioning in the virtual environment due to driver/libvirt/any other software issue. While i am convinced it’s physically dying so I am looking to duplicate efforts and change my whole setup.
The thing is, apart from some graphics production I am getting into audio production. So it’s getting harder for me to meet the requirements of my current workload.

I am planing on:

  1. Using my current hardware as an HTPC running linux with integrated graphics (I’ve done some magic using Arch+KDE with KDE connect and scripts in the past, setting up a very easy to use media pc with Netflix, VLC, etc…) and adding an easy-to-start Hypervisor VM with a 1070 or vega 64 with windows to play games on my tv. Probably buying a bluetooth adapter and passing it thru to hook up my ps4 controllers.

  2. Build a brand new setup for my workstation with Ryzen 3rd gen (Currently running a windows VM with 1 4k monitor, 2 1080 vertical monitors and ArchLinux as host on a lg 29’ widescreen)

So here is my current setup:

  • I7 4790k (Yes it’s the only k processor that supports intel VTD)
  • 32 GB ram (4X8GB sticks)
  • Gigabyte gaming 3 h97 (Sth like that)
  • A few ssd, one running a SAMBA Server to share files from one machine to another.
  • GTX 970 (Dying)
  • Scarlet 2i2

I own a 1070 and a vega 64, also I’ve a spare Rx580 4GB sitting around.

I’ve mostly figured things out but the audio part is killing me. Right now I am using software so linux sees the VM as an application and plays the audio on linux, as another application. I can take that with PLUG and use it as a (very low quality) dry signal into audacity (No plug-in support, not anything, just very basic).

So… to sum up, here are some questions for the experts:

  1. How good of a quality signal can I get from using audio via software from the VM to linux?
    Right now this is how i’ve got it setted it up:
    QEMU_AUDIO_DRV=PA QEMU_PA_SAMPLES=8192 QEMU_AUDIO_TIMER_PERIOD=100 QEMU_PA_SERVER=/run/user/1000/pulse/native
    -soundhw ac97

  2. Can I just passtrough the audio interface? if so, any instructions in how to do so? I can’t just treat it as an usb device, since the driver seems to be preventing QEMU from taking control over it.

  3. Are there any hardware recommendations in terms of platform for my new workstation? (Ryzen gen 3 based, maybe any decent x470 board)

  4. Is it possible to use the Gentoo Fix for the VEGA64 under another distro? (Like arch, for example) (I don’t see why not, but maybe You can see a problem that I am not seeing)
    I can not insert links, but if you google GPU passthrough with libvirt qemu kvm fixed vega, you will find it. its on the gentoo wiki.

Sorry for the terrible english.
Thanks in advance. :grimacing:

0 Likes