NVIDIA A30 no device were found HP Proliant 585 G7 centOS 7

Graphics / Linux Linux
1

Hi we are installing a new NVIDIA A30 in our HP Proliant 585 G7.

We can install the NVIDIA-Linux-x86_64-470.82.01.run drivers

but when we try to run the nvidia-smu we get this error message:
No devices were found

From dmesg | grep NVRM we get:

[ 9.537026] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 470.82.01 Wed Oct 27 21:21:55 UTC 2021
[ 462.145888] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
NVRM: BAR1 is 0M @ 0x0 (PCI:0000:41:00.0)
[ 462.146044] NVRM: The NVIDIA probe routine failed for 1 device(s).
[ 462.146048] NVRM: None of the NVIDIA devices were initialized.
[ 1624.354351] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 470.82.01 Wed Oct 27 21:21:55 UTC 2021
[ 1674.449971] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 470.82.01 Wed Oct 27 21:21:55 UTC 2021
[ 2199.542247] NVRM: GPU 0000:41:00.0: RmInitAdapter failed! (0x24:0xffff:1220)
[ 2199.542617] NVRM: GPU 0000:41:00.0: rm_init_adapter failed, device minor number 0
[ 2200.187143] NVRM: GPU 0000:41:00.0: RmInitAdapter failed! (0x24:0xffff:1220)
[ 2200.187410] NVRM: GPU 0000:41:00.0: rm_init_adapter failed, device minor number 0

and from /proc/driver/nvidia/gpus/0000:41:00.0 we get

Model: NVIDIA A30
IRQ: 96
GPU UUID: GPU-9dd77b2e-aa4e-b963-efbd-17c864b60f6d
Video BIOS: ??.??.??.??.??
Bus Type: PCIe
DMA Size: 47 bits
DMA Mask: 0x7fffffffffff
Bus Location: 0000:41:00.0
Device Minor: 0
GPU Excluded: No

The nouveau driver is off

Any ideas how to solve this problem ?

Kind Regards

2

Please make sure that “above 4G decoding” or “large/64bit BARs” is enabled, CSM is disabled in bios and the system is booting using EFI.
If that doesn’t help, please set kernel parameter
pci=realloc

3

Hi in our bios we don’t have“above 4G decoding” or “large/64bit BARs options

about pci=realloc we added this line in /etc/default/grub and after we updated the
grub with the command grub2-mkconfig -o /boot/grub2/grub.cfg

In particular with pci=realloc and pci=realloc=on it is n’t possible to install the NVDIA drivers… with pci=realloc=off we are able to reinstall NVDIA driver but nothing change we go back to the starting point nvidia-smi => no devices were found

We tried also to follow step by step this guide Install Nvidia Drivers on RHEL | Kinetica Docs
but no changes always the same error

Do you have any other idea ?

Kind Regards

4

IIRC, the bios setting is a bit hidden on HPE servers, in “service options”
https://forums.developer.nvidia.com/t/nvrm-this-pci-i-o-region-assigned-to-your-nvidia-device-is-invalid/81645/4
Please run nvidia-bug-report.sh as root and attach the resulting nvidia-bug-report.log.gz file to your post.

5

nvidia-bug-report.log.gz (126.6 KB)

Hi in attachment our nvidia bug report

Kind regards

6

Now this is some really old machine and the bad news is, IIRC, this won’t work due to the fact that this is an old non-UEFI multi-socket AMD machine and the Linux kernel has disabled 64bit resources for those due to instability. 64bit resources are needed to get the A30 to work.
https://forums.developer.nvidia.com/t/nvidia-smi-shows-no-devices-were-found/167930/4