nvidia-smi reports Unable to determine the device handle for GPU

My Setup is a headless HP Z800 running Vmware ESXi 5.5
Nvidia 1080 using DirectPath is assigned to an Ubuntu 16.04 64bits Guest OS

Ubuntu is running in level 3 (in sysv terms). NVIDIA 1080 will be used only with cudnn.

Nvidia drivers tested 375.26, 370.28.

nouveau is blacklisted:
~sudo lsmod|grep nouveau ~

Driver modules are loaded:
~$ sudo lsmod|grep nvidia
nvidia_drm 53248 0
nvidia_modeset 790528 1 nvidia_drm
nvidia 11943936 1 nvidia_modeset
drm_kms_helper 147456 2 vmwgfx,nvidia_drm
drm 360448 5 ttm,drm_kms_helper,vmwgfx,nvidia_drm

~$ cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module 375.26 Thu Dec 8 18:36:43 PST 2016
GCC version: gcc version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.4)

~$ cat /proc/driver/nvidia/gpus/0000:03:00.0/information
Model: GeForce GTX 1080
IRQ: 63
GPU UUID: GPU-???-???-???-???-???
Video BIOS: ??.??.??.??.??
Bus Type: PCIe
DMA Size: 36 bits
DMA Mask: 0xfffffffff
Bus Location: 0000:03:00.0
Device Minor: 0

$ nvidia-smi -a
Unable to determine the device handle for GPU 0000:03:00.0: Unknown Error

and my logs are full of
NVRM: RmInitAdapter failed! (0x23:0x56:458)
[ 2179.616234] NVRM: rm_init_adapter failed for device bearing minor number 0

Thank you

Fixed. It was another parameter in Vmware

sorry, I am facing the exactly same problem. could you tell me how to solve it ?

This is a VMWare pass-through (directpath) problem.
You must add to your vmx file the following directive:

hypervisor.cpuid.v0 = FALSE

Cheers

1 Like