WARNING: infoROM is corrupted at gpu 0000:01:00.0 while using RTX 2080 Ti on Ubuntu 19.04


I am currently using the RTX 2080 Ti AMP Edition on Ubuntu 19.04 LTS. After entering nvidia-smi

Here is the output:

| NVIDIA-SMI 418.56 Driver Version: 418.56 CUDA Version: 10.1 |
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| 0 GeForce RTX 208… Off | 00000000:01:00.0 On | N/A |
| 35% 25C P8 8W / 260W | 431MiB / 10986MiB | 4% Default |

| Processes: GPU Memory |
| GPU PID Type Process name Usage |
| 0 1282 G /usr/lib/xorg/Xorg 24MiB |
| 0 1318 G /usr/bin/gnome-shell 58MiB |
| 0 2462 G /usr/lib/xorg/Xorg 137MiB |
| 0 2627 G /usr/bin/gnome-shell 107MiB |
| 0 28390 G …quest-channel-token=1419327995453926651 101MiB |
WARNING: infoROM is corrupted at gpu 0000:01:00.0

This is doesn’t seem like a cuda-gdb issue. Moving to Hardware GPU board for help.


I have the same issue. Nvidia driver version is also the same.
GPU: RTX2060 graphics card
OS: Ubuntu 18.04

Could you suggest any solution to this problem?

I also have the same issue on a machine with two Quadro P6000’s. The error comes up on both Windows 10 and Debian (unstable), using latest drivers. On Windows I am able to run some of the NVidia provided samples without error, and was also able to get tensorflow-gpu working in anaconda (CUDA toolkit 10.1 I believe). I haven’t yet been able to get tensorflow-gpu working in anaconda under Linux.

Is possibly this just a harmless warning? I read on other comments that there is no publicly available information about this. This is very difficult to digest for a multi-thousand dollar card. We need to know whether or not it’s a problem necessarily, or not!

I’m replying to my own message, as I downloaded the NVidia Display Port firmware flash tool and ran it to find out that it recommended updating the display port firmware. Do, I did that, and upon rebooting the inforam message is not longer present! That update doesn’t update the Video Bios, but whatever it did it seems to have solved the problem. I’ll write back if it somehow reemerges.