Questions[nvidia](server)

I have a GPU server which requires cuda for example for machine learning tasks.

unfortunately, as soon as I install the NVIDIA drivers and cuda, apparently a variant of gnome is installed as well. This gnome variant can almost do nothing, the shell is not working and apparently it comes with some desktop settings which are really annoying me.

For example per default, it sets shutdown after 20 minutes inactivity. For a Server, this is very suboptimal to call it this way.

Do you know how to install the drivers and the cuda toolkit without gnome or (less preferrably) remove gnome after the fact and all its stuff (such as the 20 minute idle setting) which I definitely do not want on my Server?

We have a Ubuntu 20.04 server with Nvidia GPUs and want to change the Power Mode / GpuPowerMizerMode to Prefer Maximum Performance.

One way to do this is

nvidia-settings -a "[gpu:0]/GpuPowerMizerMode=1"

However, that would require an X session with access to the respective GPU; editing /etc/X11/xorg.conf would not work for the same reason.

Is there another way to read and set the GpuPowerMizerMode of a Nvidia GPU in Ubuntu?

We have a SuperMicro GPU server with:

• 2x Intel(R) Xeon(R) CPU E5-2660 v4 @ 2.00GHz
• 512GB memory
• more than enough disk space
• X10DRG-O+-CPU (BIOS Version : 2.0a [current])
• X9DRG-O-PCIE PCI-E expander card
• 8x GTX 1080

It is setup with Ubuntu 16.04.1 LTS, NVIDA driver 367.57 and CUDA-8.0. When it runs, it runs temporarily fine. It is however completely useless with the stock Kernel (v4.4) -- the system almost immediately freezes when doing something non-trivial on any GPU. We therefore suspected a hardware issue, but cooling is fine, and a second almost identical machine (just different maker of the GPUs) shows the exact same behaviour.

To make it run fine for some time, you have to downgrade the Kernel to v3.14.1-trusty (tested almost every version before that one). But there are still random freezes, usually with nothing in the logs. Sometimes the whole machine freezes, other times just any GPU-related processes.

There seem to be other [1] people [2] having this issue, but no solution there.

Is anyone having the same experience with this type of machine?

Update: The machines seem to run stable (regardless of any software) if the cards are inserted only on one side of the PCI-E expander, which means all cards are driven by the same CPU. Another machine however seems to run stable with 8 cards (uptime of about 4 months right now) with Kernel 3.19 after months of having the problems described above. Bizarre.

[1] https://devtalk.nvidia.com/default/topic/958927/gpu-job-fail-/

[2] https://devtalk.nvidia.com/default/topic/959699/linux/nvidia-smi-periodically-crashes-system-on-ubuntu-16-04-lts/

I'm trying to set up a cluster of 8 computers plus a main file server. Ideally, I'd like to set this up in a pxe-boot, quasi-diskless/quasi-stateless environment (i.e. the only local storage is /var, where things like torque configuration will go). Each of the 8 compute nodes has 4 NVIDIA Tesla K40m's, but the root file server has no GPU.

Ideally, I'd like to be able to create the complete installation on the file server (at /node) then PXE-boot that to the compute nodes, but, I haven't found a way to install the NVIDIA drivers without an NVIDIA GPU on board. I found one question on NVIDIA's forums about how someone unsuccessfully attempted this...

Alternatively, I could install the NVIDIA drivers to one of the compute nodes (one is currently running CentOS on it's local disks) to (for example) /usr/local/nvidia and keep track of what files it creates and create a tarball of that to copy to the file server installation.

Lastly, I could just maintain eight separate installations, but, I don't like this from a long-term maintenance perspective (each compute node will be running torque jobs so I'd like the nodes to look more-or-less identical).

In summary, what I'm asking for is this:

1. Can I install the NVIDIA drivers without an NVIDIA GPU on board?
2. Is there some other way I should be going about this?

For reference, we're running CentOS 7.

[root@compute-3 /]# uname -a
Linux compute-3 3.10.0-514.2.2.el7.x86_64 #1 SMP Tue Dec 6 23:06:41 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux

We have a ProLiant DL580 Gen8 Server and want to install Gigabyte GForce GTX 980 ti Display Card in PCIE slot, When we connect 8 pins sockets power, server could not turn on, and when power socket not connected, server starts but the graphic card could not detect in server, now my question is:

does this server supports this graphic card? if yes, what's the problem?

our power cable is regular cable 8 pins.