June 6, 20197 yr That tells me nvidia-smi is working though as that's how I get the information displayed there.watch nvidia-smi is the command and it runs on Unraid not in the docker container.Sent from my Mi A1 using Tapatalk I had the same issue with my Quadro K2000. The solution for me was to change my PCI-e port to ”Gen 2” in bios, default it was ”Gen 3”. I can add that I use a Supermicro motherboard.Good luck .
June 6, 20197 yr Just to clarify, this plugin only installs version 418.56 of the drivers? Any further updates that are scheduled? Cheers.
June 6, 20197 yr 1 hour ago, ezhik said: Just to clarify, this plugin only installs version 418.56 of the drivers? Any further updates that are scheduled? Cheers. This is covered in the first two posts
June 7, 20197 yr 25 minutes ago, CHBMB said: This is covered in the first two posts A gentle RTFM. I see... The answer was: -- 6. We will produce one Nvidia build per Unraid release, we will not be updating the drivers multiple times for each Unraid version, unless there is a critical bug that demands this. So please don't ask. --
June 7, 20197 yr 3 minutes ago, ezhik said: A gentle RTFM. I see... The answer was: -- 6. We will produce one Nvidia build per Unraid release, we will not be updating the drivers multiple times for each Unraid version, unless there is a critical bug that demands this. So please don't ask. -- Ha! You found it....
June 8, 20197 yr On 5/27/2019 at 11:59 AM, Xaero said: @CHBMB I too see this high power consumption. I know why it's happening, too. Basically, the nvidia driver doesn't initialize power management until an Xorg server is running. The only way to force a power profile on Linux currently is to use nvidia-smi like so: nvidia-settings --ctrl-display :0 -a "[gpu:0]/GPUPowerMizerMode=2" Which requires a running Xorg display. I've been trying to dig around in sysfs to see if there is another place that this value is stored, but there doesn't seem to be. It looks like the cards are locked into performance mode... Perhaps this is worth bringing up to nvidia? In the meantime, I'm going to continue digging to see if I can find a way (perhaps an nvidia-settings docker?) to force the power state. I'm not exactly sure how much extra power my GTX 1050 is chewing sitting idle without being in a power saving mode, but my guess (based on readings from a "Kill A Watt" with the GTX installed vs. not.) is around 15 to 20 watts. 15 to 20w x $0.15 per kwh = ~ $20 to $25 per year. I'd certainly be willing to donate at least that much towards getting this done. Edited June 8, 20197 yr by brainbone
June 9, 20197 yr I'm planning on buying a GPU for use with this plugin and a Plex container. Can I buy any GPU listed here and be pretty much assured it will work, or is there any kind of compatibility list to watch out for?
June 9, 20197 yr 3 hours ago, knalbone said: I'm planning on buying a GPU for use with this plugin and a Plex container. Can I buy any GPU listed here and be pretty much assured it will work, or is there any kind of compatibility list to watch out for? Make sure you pay attention to what's highlighted in green/yes. The 1030, for example, is a GPU you should NOT get. This list has more detailed information on GPUs with Plex.
June 9, 20197 yr 34 minutes ago, brainbone said: Make sure you pay attention to what's highlighted in green/yes. The 1030, for example, is a GPU you should NOT get. This list has more detailed information on GPUs with Plex. Wow, that site is great. Thank you so much! Edit: sorry for breaking the rules. Sent from my Pixel 2 XL using Tapatalk Edited June 9, 20197 yr by knalbone
June 9, 20197 yr 2 minutes ago, knalbone said: Wow, that site is great. Thank you so much! I guess the only other question is, does this plugin include [...]? Please edit or delete that last post. The first post in this thread forbids talking about that! Edited June 9, 20197 yr by brainbone
June 9, 20197 yr 2 minutes ago, brainbone said: Please edit or delete that last post. The first post in this thread forbids talking about that! Oops, sorry. Edited
June 11, 20197 yr 1 hour ago, Patrick Cheah said: How do you uninstall the Nvidia drivers and reinstall stock drivers? You can do so through the plug-in. Select a stock build
June 11, 20197 yr 1 hour ago, Patrick Cheah said: How do you uninstall the Nvidia drivers and reinstall stock drivers? Go into the plugin and select a stock build from the dropdown on the right. Click on download & install, then reboot when prompted. EDIT: @aptalcabeat me to it Edited June 11, 20197 yr by Cessquill Too slow, man, too slow.
June 11, 20197 yr Thank you. aptalca & Cessquill. Just tried that stock dropdown and getting errors. Every line of the download showed failed and was asked to reboot. After reboot the Nvidia drivers are still installed.
June 11, 20197 yr 3 hours ago, Patrick Cheah said: Every line of the download showed failed and was asked to reboot. How much RAM do you have?
June 12, 20197 yr 27 minutes ago, Patrick Cheah said: 2GB RAM. That is probably not enough to do the install from the webUI. Make a backup of the config folder from flash, prepare flash as a new install, copy that config folder back to flash. You may have to follow a similar procedure anytime you need to update Unraid since you have so little RAM.
June 12, 20197 yr 5 hours ago, trurl said: That is probably not enough to do the install from the webUI. Make a backup of the config folder from flash, prepare flash as a new install, copy that config folder back to flash. You may have to follow a similar procedure anytime you need to update Unraid since you have so little RAM. Thank you very much. Followed your instructions and everything is back to normal.
June 12, 20197 yr Has anyone successfully gotten telegraf working with inputs.nvidia-smi? Even adding all the nvidia variables to the container I still get the following error in my telegraf logs: Quote [inputs.nvidia_smi]: Error in plugin: fork/exec /usr/bin/nvidia-smi: no such file or directory
June 12, 20197 yr @IamSpartacus Is that command being run inside a Telegraf container? If so it would explain why it can't find the nvidia-smi binary
June 12, 20197 yr Just now, CHBMB said: @IamSpartacus Is that command being run inside a Telegraf container? If so it would explain why it can't find the nvidia-smi binary Yes inside telegraf. But when I run 'which nvidia-smi' inside the container it returns /usr/bin/nvidia-smi and I see it there when I cd to /usr/bin/ inside the container as well.
June 12, 20197 yr 6.7.1-rc2 and rc1 video transcoding broke for my GTX 760. Currently reverted back to 6.7.0 and everything is working again. The only docker using the GPU is the Plex server (pms-docker) running the latest software as of today. I was getting an s1002 error when the need for video transcoding was taking place. Changing plex to not use hardware fixed the issue, but reverting back to 6.7.0 allows me to use hardware acceleration again.
June 12, 20197 yr 2 hours ago, jbreed said: 6.7.1-rc2 and rc1 video transcoding broke for my GTX 760. Currently reverted back to 6.7.0 and everything is working again. The only docker using the GPU is the Plex server (pms-docker) running the latest software as of today. I was getting an s1002 error when the need for video transcoding was taking place. Changing plex to not use hardware fixed the issue, but reverting back to 6.7.0 allows me to use hardware acceleration again. Is your GPU detected? Not sure there's much I can do about this to be honest, unless it's broken for everyone, we have very little control over the upstream stuff.
June 13, 20197 yr No questions or complaints today. Just wanted to say I've got this running with my new GPU and Plex container and it works great. Thanks for all the hard work Linuxserver team!
Archived
This topic is now archived and is closed to further replies.