Tesla T4 - PCI I/O region invalid

[thies.gerken]

Hello,

I’m trying to use a Tesla T4 card.
Unfortunately, the card is not shown in nvidia-smi.
Instead, dmesg shows

[   38.828288] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
               NVRM: BAR0 is 0M @ 0x0 (PCI:0000:04:00.0)
[   38.828289] NVRM: The system BIOS may have misconfigured your GPU.
[   38.828307] NVRM: The NVIDIA probe routine failed for 1 device(s).
[   38.828308] NVRM: None of the NVIDIA devices were initialized.

Reading

NVRM: This PCI I/O region assigned to your NVIDIA device is invalid: NVRM

Hi guys, i have WS C422/SAGE 10G, It cannot detect Tesla T4 even after installing the driver. i have update BIOS version to the newest version. do you have any idea how to handle this?

Cuda and nvidia driver setup failure on Centos7

I am trying to get a K20 setup on a HP DL380p g8 for some cuda and accelerated vms. I have a fresh install of centos7 and get an error about Unable to load the ‘nvidia-drm’ kernel module. I have tried 396 410 and 418 non of them will load. I have also tried 3.10 4.4 and 5.2 kernels. any suggestions would be appriciated

NVRM: This PCI I/O region assigned to your NVIDIA device is invalid

Dear All, We use a lot of ProLiant SL250s Gen8 server with Tesla K40m. The servers boots from networking an unified boot image. We replaced the motherboard in a server, then it cannot load the nvidia driver: [root@hostname ~]# nvidia-smi NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running. [root@hostname ~]# lspci | grep Tesla 08:00.0 3D controller: NVIDIA Corporation GK110BGL [Tesla K40m] (rev a1) 24:…

it seems that I have to enable “Above 4G decoding” in my BIOS. However, I cannot find such a setting.
I would be grateful for any ideas on how to remedy this problem.

System:
Ubuntu 18.04, kernel 5.3.0
ASUS TUF GAMING X570-PLUS (most recent BIOS), 32GiB RAM, AMD Ryzen 3950X
Drivers: 440.64.00 (from Tesla Driver for Ubuntu 18.04 | 440.64.00 | Linux 64-bit Ubuntu 18.04 | NVIDIA)

[generix]

Try enabling “SR-IOV Support”, maybe this will implicitly enable 64-bit BARs. Of course, make sure your OS boots using EFI, disable CSM in bios.
If all that doesn’t help, you’re out of luck with that board.

[generix]

Just a general note, when not used in specific servers, the T4 needs an additional fan for cooling.

[thies.gerken]

With SR-IOV Support enabled and the T4 card installed the system does not seem to boot at all (no POST screen, but also no beeping).
I am using an additional fan, the system boots using EFI and CSM is disabled.
Is there some kind of compatibility list of mainboards (ideally with support for the current Ryzen processors) that are known to work with the T4?

[generix]

No, that list won’t exist because Ryzen means consumer/gaming hardware and Tesla means enterprise/server/workstation hardware. Nobody will test that combo officially.
So most if not all workstation/server mainboards will support this but finding an Ryzen consumer mainboard which supports this will be up to you.