Hi, I am building a server with the following configuration
i7-14700KF
128GB RAM
NVIDIA RTX 6000 Ada
ASUS PRIME Z790
I want to do a gpu passthrough to vm (not vGPU), but I get error code 12 in Windows VM.
Does anyone know how to solve this?
Is gpu pass-through (NOT vGPU) not possible on enterprise GPUs?
What I have tried is as follows
i7-14700KF
128GB RAM
NVIDIA RTX 6000 Ada
ASUS PRIME Z790
I want to do a gpu passthrough to vm (not vGPU), but I get error code 12 in Windows VM.
Does anyone know how to solve this?
Is gpu pass-through (NOT vGPU) not possible on enterprise GPUs?
What I have tried is as follows
- Configure GPU passthrough as follows
https://forum.proxmox.com/threads/p...x-ve-8-installation-and-configuration.130218/ - Put the RTX 6000 Ada in compute mode with displaymodeselector
https://forum.proxmox.com/threads/pci-passthrough-not-working-with-proxmox8-and-rtx6000-ada.138848/ -
Code:
qm set 100 --args '-fw_cfg name=opt/ovmf/X-PciMmio64Mb,string=65536 -global q35-pcihost.pci-hole64-size=2048G'
https://forum.proxmox.com/threads/gpu-passthrough-code-12.110407/post-504040 - IOMMU, Resize Bar, Above 4G Decoding, SR-IOV are enabled.
Code:
root@pve:~# pveversion
pve-manager/8.2.2/9355359cd7afbae4 (running kernel: 6.8.4-2-pve)
root@pve:~# qm config 100
args: -fw_cfg name=opt/ovmf/X-PciMmio64Mb,string=65536
balloon: 0
bios: ovmf
boot: order=scsi0;ide2;net0;ide0
cores: 8
cpu: x86-64-v2-AES
efidisk0: local-lvm:vm-100-disk-0,efitype=4m,pre-enrolled-keys=1,size=4M
hostpci0: 0000:01:00,pcie=1
ide0: local:iso/virtio-win.iso,media=cdrom,size=612812K
ide2: local:iso/windows11.iso,media=cdrom,size=6561616K
machine: pc-q35-8.1
memory: 8192
meta: creation-qemu=8.1.5,ctime=1717567965
name: windows11
net0: virtio=BC:24:11:2F:73:26,bridge=vmbr0
numa: 0
ostype: win11
scsi0: local-lvm:vm-100-disk-1,iothread=1,size=256G,ssd=1
scsihw: virtio-scsi-single
smbios1: uuid=808887e3-4b20-491b-b748-06bf112ed02c
sockets: 1
tpmstate0: local-lvm:vm-100-disk-2,size=4M,version=v2.0
vmgenid: 36dd65f0-4414-40fe-aa06-4733ad057fdd
root@pve:~# dmesg | grep "DMAR"
[ 0.004032] ACPI: DMAR 0x000000007633C000 000050 (v01 INTEL EDK2 00000002 01000013)
[ 0.004054] ACPI: Reserving DMAR table memory at [mem 0x7633c000-0x7633c04f]
[ 0.128610] DMAR: IOMMU enabled
[ 0.286961] DMAR: Host address width 39
[ 0.286962] DMAR: DRHD base: 0x000000fed91000 flags: 0x1
[ 0.286968] DMAR: dmar0: reg_base_addr fed91000 ver 5:0 cap d2008c40660462 ecap f050da
[ 0.286970] DMAR-IR: IOAPIC id 2 under DRHD base 0xfed91000 IOMMU 0
[ 0.286971] DMAR-IR: HPET id 0 under DRHD base 0xfed91000
[ 0.286971] DMAR-IR: Queued invalidation will be enabled to support x2apic and Intr-remapping.
[ 0.288400] DMAR-IR: Enabled IRQ remapping in x2apic mode
[ 0.482046] DMAR: No RMRR found
[ 0.482046] DMAR: No ATSR found
[ 0.482047] DMAR: No SATC found
[ 0.482048] DMAR: dmar0: Using Queued invalidation
[ 0.482339] DMAR: Intel(R) Virtualization Technology for Directed I/O
root@pve:~# dmesg | grep "IOMMU"
[ 0.128610] DMAR: IOMMU enabled
[ 0.286970] DMAR-IR: IOAPIC id 2 under DRHD base 0xfed91000 IOMMU 0
root@pve:~# dmesg | grep "vfio"
[ 118.248404] vfio_pci: add [10de:26b1[ffffffff:ffffffff]] class 0x000000/00000000
[ 118.295996] vfio-pci 0000:01:00.0: Enabling HDA controller
[ 119.557339] vfio-pci 0000:01:00.0: Enabling HDA controller
[ 119.557358] vfio-pci 0000:01:00.0: enabling device (0000 -> 0002)
[ 119.695648] vfio-pci 0000:01:00.0: vfio_bar_restore: reset recovery - restoring BARs
[ 120.319668] vfio-pci 0000:01:00.0: vfio_bar_restore: reset recovery - restoring BARs
[ 120.335904] vfio-pci 0000:01:00.0: vfio_bar_restore: reset recovery - restoring BARs
[ 120.348073] vfio-pci 0000:01:00.0: vfio_bar_restore: reset recovery - restoring BARs
Last edited: