Has anybody tried SR-IOV on Intel 8th gen NUC (8559U w/ Iris Plus Graphics 655)?

A

alpha754293

Member
Jan 8, 2023
108
20
23
I have an 8th gen Intel NUC (8559U w/ Iris Plus Graphics 655) and I ran through the instructions, trying to set up and deploy SR-IOV for the iGPU from here and here as well.

Neither seemed to work and I was wondering if anybody else has tried this before and were able to get it working or if those instructions really only work on 12th gen Intel processors and beyond?

Your help is greatly appreciated.

When I type in lspci | grep VGA, rather than seeing the 7 VFs, it only shows the root device, no VFs at all. I have checked that the MOK has been enrolled into secure boot successfully/properly/appropriately and it still doesn't seem to work.

When I type in dmesg | grep i915, this is what I get:

Code: 
[    0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-6.8.12-10-pve root=/dev/mapper/pve-root ro quiet intel_iommu=on iommu=pt i915.enable_guc=3 i915.max_vfs=7 initcall_blacklist=sysfb_init pcie_aspm=off
[    0.075763] Kernel command line: BOOT_IMAGE=/boot/vmlinuz-6.8.12-10-pve root=/dev/mapper/pve-root ro quiet intel_iommu=on iommu=pt i915.enable_guc=3 i915.max_vfs=7 initcall_blacklist=sysfb_init pcie_aspm=off
[    3.781975] i915: module verification failed: signature and/or required key missing - tainting kernel
[    4.307277] i915: You are using the i915-sriov-dkms module, a ported version of the i915 module with SR-IOV support.
[    4.307279] i915: Please file any bug report at https://github.com/strongtz/i915-sriov-dkms/issues/new.
[    4.307280] i915: Module Homepage: https://github.com/strongtz/i915-sriov-dkms
[    4.307654] i915 0000:00:02.0: [drm] Found COFFEELAKE/ULT (device ID 3ea5) display version 9.00 stepping N/A
[    4.307762] i915 0000:00:02.0: [drm] GT0: Incompatible option enable_guc=3 - GuC submission is N/A
[    4.308229] i915 0000:00:02.0: [drm] Found 128MB of eDRAM
[    4.308249] i915 0000:00:02.0: [drm] VT-d active for gfx access
[    4.308253] i915 0000:00:02.0: vgaarb: deactivate vga console
[    4.308291] i915 0000:00:02.0: [drm] Using Transparent Hugepages
[    4.328711] i915 0000:00:02.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[    4.332243] i915 0000:00:02.0: [drm] Finished loading DMC firmware i915/kbl_dmc_ver1_04.bin (v1.4)
[    4.342823] i915 0000:00:02.0: [drm] GT0: GuC firmware i915/kbl_guc_70.1.1.bin version 70.1.1
[    4.342829] i915 0000:00:02.0: [drm] GT0: HuC firmware i915/kbl_huc_4.0.0.bin version 4.0.0
[    4.369517] i915 0000:00:02.0: [drm] GT0: HuC: authenticated for all workloads
[    4.369521] i915 0000:00:02.0: [drm] GT0: GUC: submission disabled
[    4.369523] i915 0000:00:02.0: [drm] GT0: GUC: SLPC disabled
[    4.370408] mei_hdcp 0000:00:16.0-b638ab7e-94e2-4ea2-a552-d1c54b627f04: bound 0000:00:02.0 (ops i915_hdcp_ops [i915])
[    4.557173] [drm] Initialized i915 1.6.0 20230929 for 0000:00:02.0 on minor 0
[    4.627409] fbcon: i915drmfb (fb0) is primary device
[    4.677540] i915 [CRTC:51:pipe A] fastset requirement not met in dp_m_n (expected tu 64 data 1730150/8388608 link 144179/524288, found tu 64, data 5767168/8388608 link 480597/524288)
[    4.677546] i915 [CRTC:51:pipe A] fastset requirement not met in dpll_hw_state  
[    4.677547] i915 expected:
[    4.677548] i915 dpll_hw_state: ctrl1: 0x1, cfgcr1: 0x0, cfgcr2: 0x0
[    4.677549] i915 found:
[    4.677549] i915 dpll_hw_state: ctrl1: 0x5, cfgcr1: 0x0, cfgcr2: 0x0
[    4.677550] i915 [CRTC:51:pipe A] fastset requirement not met in port_clock (expected 540000, found 162000)
[    4.839584] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])
[    4.840665] i915 0000:00:02.0: [drm] fb0: i915drmfb frame buffer device
[    4.850504] i915 display info: display version: 9
[    4.850506] i915 display info: display stepping: **
[    4.850507] i915 display info: cursor_needs_physical: no
[    4.850508] i915 display info: has_cdclk_crawl: no
[    4.850509] i915 display info: has_cdclk_squash: no
[    4.850510] i915 display info: has_ddi: yes
[    4.850510] i915 display info: has_dp_mst: yes
[    4.850511] i915 display info: has_dsb: no
[    4.850512] i915 display info: has_fpga_dbg: yes
[    4.850512] i915 display info: has_gmch: no
[    4.850513] i915 display info: has_hotplug: yes
[    4.850513] i915 display info: has_hti: no
[    4.850514] i915 display info: has_ipc: yes
[    4.850515] i915 display info: has_overlay: no
[    4.850515] i915 display info: has_psr: yes
[    4.850516] i915 display info: has_psr_hw_tracking: yes
[    4.850516] i915 display info: overlay_needs_physical: no
[    4.850517] i915 display info: supports_tv: no
[    4.850518] i915 display info: has_hdcp: yes
[    4.850518] i915 display info: has_dmc: yes
[    4.850519] i915 display info: has_dsc: no
[    4.850520] i915 display info: rawclk rate: 24000 kHz

Your help is greatly appreciated.

Thank you.
 
GVT-g or d? g in the doc but not sure the diff completely myself - but you should be able to play a little with that but not as good as SR-IOV

it is a bummer. I have a 1157-g7 which 'would' be supported if intel finished the work on the 11th gen platform. lame
 
  • Like
Reactions: alpha754293
arghdubya said:
GVT-g or d? g in the doc but not sure the diff completely myself - but you should be able to play a little with that but not as good as SR-IOV
Click to expand...
My understanding is that GVT-d is like "normal" GPU passthrough, but without the virtual functions, which means that only one VM can take it at a time (vs. SR-IOV which creates virtual functions, which means that I can pass it through to multiple VMs simultaneously, and it will just share the physical, hardware resources).

I could be wrong, but that is my current level of understanding in regards to this (at least for intel iGPUs).
 
gotcha, but GVT-g does allow device sharing
GVT-g creates mediated devices which are assigned to VMs. Those mediated devices are effectively virtual and pass the work to the host GPU from the guest when they the guest is using them. The work is time sliced, so you get close to the full performance if only one VM is doing work on the device but if multiple VMs are used at the same time they have to share the workload so they can't have 100% performance.

this guide looks at actually using i915 gvt-g
https://cetteup.com/216/how-to-use-...rdware-accelerated-streaming-in-a-proxmox-vm/
 
Last edited:
  • Like
Reactions: alpha754293
arghdubya said:
gotcha, but GVT-g does allow device sharing
GVT-g creates mediated devices which are assigned to VMs. Those mediated devices are effectively virtual and pass the work to the host GPU from the guest when they the guest is using them. The work is time sliced, so you get close to the full performance if only one VM is doing work on the device but if multiple VMs are used at the same time they have to share the workload so they can't have 100% performance.

this guide looks at actually using i915 gvt-g
https://cetteup.com/216/how-to-use-...rdware-accelerated-streaming-in-a-proxmox-vm/
Click to expand...

I don't suppose that you would happen to have a list of how many mediated devices are available based on either processor and/or processor generation, would you?

I am asking because I am testing this on my 6500T right now and I have the mediated devices up and running, per the instructions that you shared (thank you for that), but it would appear that I can only use one mediated device at a time rather than being able to fire up two Win11 VMs using the same mediated device.

(edit: It would appear that I can use one of the "big" "vGPUs" or two "small" vGPU:

Screenshot_2025-05-19_00-46-08.png

arghdubya said:
yeah, it makes SR-IOV capability not as big a deal since that just allows better simultaneous usage by VMs.
Click to expand...
I think that one of the most significant advantages of SR-IOV for iGPUs would be the fact that I should be able to create (and run) multiple VMs at the same time, off of the same iGPU.

i.e. if that's true, my tiny humans can log into their own respective VM, and play minecraft with each other, off of the same iGPU.
 
Last edited:
You must log in or register to reply here.
Top Bottom