[SOLVED] GPU Passthrough on multiple VMs

TanguyE2VR

New Member
Sep 24, 2020
3
2
3
23
Hi I need help I have 2 graphic cards and I need to do 2 VMs with 1 graphic card for each VM. I've created the 2 VMs and done the gpu passthrough they work well but when I try to launch them together one of them does'nt launch and output this error message

Code:
Task viewer: VM 102 - StartOutputStatusStop
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: warning: host doesn't support requested feature: CPUID.01H:ECX.pcid [bit 17]
kvm: -device vfio-pci,host=0000:0a:00.0,id=hostpci0.0,bus=ich9-pcie-port-1,addr=0x0.0,multifunction=on: vfio 0000:0a:00.0: failed to open /dev/vfio/2: Device or resource busy
TASK ERROR: start failed: QEMU exited with code 1
 
Hello, help please i have the same problem...
i have done to 1 VM the GPU but i need that the GPU to share with 10 VM
here is my GRUB

GRUB_DEFAULT=0
GRUB_TIMEOUT=5
GRUB_DISTRIBUTOR=`lsb_release -i -s 2> /dev/null || echo Debian`
GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=on pcie_acs_override=downstream"
GRUB_CMDLINE_LINUX=""

is that ok? please help me
 
Hello, help please i have the same problem...
i have done to 1 VM the GPU but i need that the GPU to share with 10 VM
here is my GRUB

GRUB_DEFAULT=0
GRUB_TIMEOUT=5
GRUB_DISTRIBUTOR=`lsb_release -i -s 2> /dev/null || echo Debian`
GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=on pcie_acs_override=downstream"
GRUB_CMDLINE_LINUX=""

is that ok? please help me
You cannot share a dedicated GPU with more than one VM unfortunately. EDIT: that might not be strictly true; you may be able to allocate the GPU to more than one VM but only one VM can use the GPU at any time.
 
Last edited: