CentOS7 with two Quadro P2000's has "No devices detected" running X

Graphics / Linux Linux
1

I have a new CentOS7 box Dell Precision 5820 with two Quadro P2000’s

# cat /proc/driver/nvidia/gpus/0000\:17\:00.0/information
Model:           Quadro P2000
IRQ:             90
GPU UUID:        GPU-ce73a90b-9cad-88f5-8247-05ff292098b0
Video BIOS:      86.06.3f.00.30
Bus Type:        PCIe
DMA Size:        47 bits
DMA Mask:        0x7fffffffffff
Bus Location:    0000:17:00.0
Device Minor:    0

# cat /proc/driver/nvidia/gpus/0000\:65\:00.0/information
Model:           Quadro P2000
IRQ:             91
GPU UUID:        GPU-da494be3-1101-0a56-deb9-bedec7c3a322
Video BIOS:      86.06.3f.00.30
Bus Type:        PCIe
DMA Size:        47 bits
DMA Mask:        0x7fffffffffff
Bus Location:    0000:65:00.0
Device Minor:    1

# cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module  390.30  Wed Jan 31 22:08:49 PST 2018
GCC version:  gcc version 4.8.5 20150623 (Red Hat 4.8.5-16) (GCC)

# cat /proc/cmdline
BOOT_IMAGE=/vmlinuz-3.10.0-693.21.1.el7.x86_64 root=UUID=91aae8f9-2f6f-4262-942c-0fe64f64cff8 
ro crashkernel=auto LANG=en_US.UTF-8 nouveau.modeset=0 rd.driver.blacklist=nouveau video=vesa:off

# cat /etc/X11/xorg.conf
Section "Device"
        Identifier  "Videocard0"
        Driver      "nvidia"
        Option "AllowEmptyInitialConfiguration" "on"
        BusID "PCI:65:0:0"
EndSection

I have tried PCI:17:0:0 and no BusID also.

Trying to run X gives

[ 29961.277] (II) LoadModule: "nvidia"
[ 29961.277] (II) Loading /usr/lib64/xorg/modules/drivers/nvidia_drv.so
[ 29961.277] (II) Module nvidia: vendor="NVIDIA Corporation"
[ 29961.277]    compiled for 4.0.2, module version = 1.0.0
[ 29961.277]    Module class: X.Org Video Driver
[ 29961.277] (II) NVIDIA dlloader X Driver  390.30  Wed Jan 31 21:01:36 PST 2018
[ 29961.277] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[ 29961.277] (--) using VT number 2

[ 29961.290] (EE) No devices detected.
[ 29961.290] (EE)
Fatal server error:
[ 29961.290] (EE) no screens found(EE)
[ 29961.290] (EE)
Please consult the The X.Org Foundation support
         at http://wiki.x.org
 for help.
[ 29961.290] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[ 29961.290] (EE)
[ 29961.314] (EE) Server terminated with error (1). Closing log file.

I am installing CUDA/nvidia driver via the repo Index of /compute/cuda/repos/rhel7/x86_64

Any ideas why the nvidia X driver is not seeing the cards?

2

‘No devices detected’ means no monitors detected, which should be overridden by the AllowEmptyInitialConfiguration option, but your BusID is incorrect. xorg.conf takes decimal values, lspci displaying hex values. e.g. 65 has to be 101
If this doesn’t resolve the issue, please run nvidia-bug-report.sh as root and attach the resulting tar.gz file to your post. Hovering the mouse over an existing post will reveal a paperclip icon.

3

Using the correct decimal versions for the PCI IDs fixed the issue. Thanks. I don’t understand why sometimes the BusID is required in xorg.conf and sometimes it is not. It seems the nvidia X driver could easily see the correct IDs from the kernel module.

4

If there are more than one gpu active and no ‘OutputClass’ section available the busid is often required.