Unable to use NVIDIA GPU on Ubuntu 18.04 for computer vision task (GeForce GTX 1060)

My machine is a ASUS ROG SCAR-GL703GM-EE033T with a NVIDIA GeForce GTX1060. I installed Ubuntu 18.04 on it and the necessary softwares needed to run deep learning applications using its GPU.

I installed:

  • Cuda 10.0
  • cuDNN 7.5.0
  • tensorflow-gpu 1.13.1

and everything looks fine, as when running the following command in a Python terminal:

tensorflow.test.is_gpu_available() 

it outputs the characteristics of the NVIDIA card and “True”. So the installation looks correct.

However, when running an application for computer vision that uses pytesseract, I’m surprised by the long run-time that may indicate the GPU isn’t really used. To verify that, when the code is running I display in another terminal the nvidia-smi command that outputs: this screenshot of nvidia-smi command when code is running

As you can see the NVIDIA driver is 418.67. It is not the one recommended by NVIDIA website (here with GeForce / GeForce 10 Series / GeForce GTX1060 / Linux 64-bit), but I tried unsuccessfully to install driver 430.67 with following message:

nvidia-installer log file '/var/log/nvidia-installer.log' creation time: Thu Jul 11 17:56:00 2019 installer version: 430.34  PATH: /usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/snap/bin  nvidia-installer command line:     ./nvidia-installer  Unable to load: nvidia-installer ncurses v6 user interface  Using: nvidia-installer ncurses user interface -> Detected 12 CPUs online; setting concurrency level to 12. ERROR: An NVIDIA kernel module 'nvidia-uvm' appears to already be loaded in your kernel.  This may be because it is in use (for example, by an X server, a CUDA program, or the NVIDIA Persistence Daemon), but this may also happen if your kernel was configured without support for module unloading.  Please be sure to exit any programs that may be using the GPU(s) before attempting to upgrade your driver.  If no GPU-based programs are running, you know that your kernel supports module unloading, and you still receive this message, then an error may have occured that has corrupted an NVIDIA kernel module's usage count, for which the simplest remedy is to reboot your computer. ERROR: Installation has failed.  Please see the file '/var/log/nvidia-installer.log' for details.  You may find suggestions on fixing installation problems in the README available on the Linux driver download page at www.nvidia.com. 

I then tried with driver 418 that ended up with tensorflow recognizing available GPU, so I continued with it.

Also, as you can see in screenshot no process uses GPU. And as you can see in the screenshot that Cuda version displayed is 10.1 althought before installing 10.0 that looked working best I erased all traces of previously installed 10.1. Why would nvidia-smi get this 10.1 version of Cuda?

Do you know, with my current configuration/installation, how to permanently make GPU available for my code running? Do you see a missing point in my installation?

Thanks for your help, I tried to provide as much details as I could!

Théo

Help finding correct power cable for GTX 1060 SC in Mac Pro 5, 1

I recently purchased an old 2012 Mac Pro 5, 1.

I have upgrade several parts including installing an Nvidia GTX 1060 SC 3GB.

I am a video editor by profession, so I quickly installed Premiere Pro. After a few hours the computer shut off suddenly, then rebooted on its own.

This has happened ~15 times now. Generally when I have multiple video applications (e.g. After Effects, Media Encoder) open.

My first thought was drivers, due to the system report that noted the graphics card was the issue.

But I’m now predicting the graphics card is drawing too much power and causing the reboot.

Looking at this guide, it notes that I need to use both mini PCIe power ports to power the card. However my GTX 1060 SC 3GB has a 6 Pin connector, not an 8.

I can only find dual mini 6 pin to single 8 pin cables.

Can someone help me find the correct cable for this?

Or am I crazy, and putting more power to the card would fry the system or just not be the issue?

Other Note: I installed these other things in the Mac Pro:

  • SSD in a PCIe slot
  • 4 HHDs in the hard drive bays to make a RAID
  • 4 port USB 3.0 hub in a PCIe slot
  • BluRay Drive

NVIDIA Driver Installation Issue (Not able to Install latest version) GTX 1060 Zotac

So, whenever i tried to install NVIDIA driver through software and updates application, I used to get blank screen while booting through normal mode.

latest driver for NVIDIA is 418 so far, but in the update menu it just shows up driver version 390.

My questions:

How do i get rid of blank screen while booting?.

Is there any other reliable source to install NVIDIA driver with latest updates?.

Is that okay to run GPU without native it`s driver configuration?.

Second monitor is gone after reboot in GTX 1060 Nvidia

I have MSI GTX 1060 graphic cart on my Ryzan 2700x pc.

GTX1060 has HDMI and DVI output.

One of my monitors (LG BRAND) has HDMI input and it is connected to GTX1060 via HDMI cable as primary and works well with no problem.

I have an old monitor (SAMSUNG SYNCMASTER) that has only RGB in input port. so I purchased a DVI TO RGB convertor (UGREEN brand) in order to connect the second monitor to GTX1060 graphic cart.

here is convertor image https://dkstatics-public.digikala.com/digikala-products…

every thing work well and I have extended dual monitor. But when I reboot the PC, the second monitor (DVI) is not detected !!

And when I unplug DVI from graphic card and plug it again, then second monitor detects and it works with no problem.

I check NVIDIA control panel, Bios and etc but I couldn’t find what is the reason? why DVI port is not working after system boot !!?

I tested another scenario, I removed HDMI cable from graphic and did reboot the system and pc booted via monitor connected to dvi correctly.

it seems when both hdmi and dvi monitors are connected to graphic card, only hdmi monitor detected and dvi is not.

it is very hard to unplug and plug cables physically each time after rebooting and also it may damages for ports for a long time.

Please help me. sorry for my poor English.

Display artifacts on fresh install with GTX 1060

Just installed 18.04.1 LTS, updated, upgraded, added in the Nvidia repo, and using the 396 driver. With all of this done, I’m getting random black line artifacts coming across the screen if I’m on a bright screen on my main monitor.

I’m using a GTX 1060, with two monitors, one connected via DVI, the other connected via HDMI. The DVI monitor is fine, it’s also an older monitor, but haven’t had any artifacts or anything on it. It only happens on my main display.

I have Windows installed on the same system (gaming), and the problem isn’t there, so I’m assuming it has something to do with software/firmware. I’m not sure where I should progress at this point. Anyone have any ideas?