Gpu has fallen off the bus Ubuntu 18.04

Hi guys, I have a small miniature vehcle, which seems like this
https://www.digitalwerk.net/adas-modellauto/
The hardware board is NVIDIA GeForce GTX1050Ti
recently a GPU bug comes to my miniature vehicles.

GPU at PCI:0000:01:00: GPU-c0d34353-a805-2aaf-30fe-6abdce0d043f
[ 8992.085325] NVRM: GPU Board Serial Number:
[ 8992.085330] NVRM: Xid (PCI:0000:01:00): 79, pid=0, GPU has fallen off the bus. Like12:10 pm

Sometimes the whole Screen freeze. It more happens when my vehicle begin to move through the remote desktop(I use no machine). I can connect the system with SSH and check the dmesg. It shows the GPU fallen off the bus.

Then after I reboot it, it works well if I only put my miniature vehicle in the table.

I have searched in the forum and tried every method. Unfortunately, I have no success.

I also get the nvidia-bug-report.log, can anyone help me to check what happened to my miniature vehicles? Thanks a lot for your guys help

nvidia-bug-report.log (1.2 MB)

dmesg seems like this:

@generix
hi generix, you are the most nice guy in this forum! Very thanks to leave so much comments about this related issue. Can you have a look at what I met? Thank you so much

1 Like

Most common XID 79 failures on desktop systems are overheating or lack of power. Since the 1050Ti is bus powered only and your cpu is also a low-power design, maybe try reseating the gpu in its slot (even multiple times) and check airflow and temperatures using nvidia-smi.
What kind of psu is built into the system?
Low chances are defective gpu or mainboard. A bios update is always worth a shot.

1 Like

Thank you so much @generix . The PSU I used is looks like this

That looks mighty small, good for about 120Watts, I guess. Just the nominal power draw of cpu+gpu would be about 110W. Plus accessories like mainboard, memory, disks, whatever, this is not sufficient. Especially when cuda is used and the gpu is power-spiking. Rather use something like 200W+

Yeah the problem by me is, it works very well last 6 months. However, this bug suddenly comes and I have never met that before

Like now it is failed again. Finally return the image below

If you’re continuously overtaxing a psu, it will finally fail.

Then there’s the question where the psu gets its power from, now worn-off batteries?

With power supply or batteries.
Batteries are 6-cell with each cell 22,2V and 5.200mAh

Taking a look at that vehicle and your initial post again, this rather sounds like the slot broke due to vibrations over time, since you said “when the vehicle begins to move”. So “fallen off the bus” can be taken literally in this case.
So first check would be to disassemble and reseat the gpu.

1 Like

Might also be some broken voltage regulators splitting battery/input power between motors and mainboard psu, since the starting-current drawn by e-motors is highest.

Have done nothing just put it on the table, then the problem came.
I will definitely try to reseat the gpu.
Thanks anyway @generix I know which direction should I go

it seems I solved it by reseating the GPU on PCIE slot.
Thanks @generix