Jump to content
XPEnology Community

Nvidia Runtime Library


disone

Recommended Posts

7 hours ago, dj_nsk said:

DS918+ uses the built-in graphics (iGPU). If it is in your CPU, then you can install Jellyfin directly (without Docker)

 

4 hours ago, Ilfe98 said:

Thanks for your reply. It could be usefun even in other case have the nvidia driver installed. There's some way?

If it's not possible how i can install jellyfin w/o docker and use igpu

Link to comment
Share on other sites

  • 1 month later...

Hello,

 

Has anyone made any progress in getting nvidia card detected in docker correctly ?

 

Tried following the instructions provided in this thread, but they're over a year old now and can't get docker to see the GPU (1650).

 

Tried the guide on this site and their driver starts fine, but after running their command to fix permissions for the nvidia driver and then attempting to start the driver package again, i see a kernel panic in dmesg output and the driver never starts.

 

https://blog.kkk.rs/archives/17

 

DVA3221 / DSM 7.2.64570-U3 / GTX 1650 on ESX 8

 

Link to comment
Share on other sites

20 hours ago, irishj said:

Hello,

 

Has anyone made any progress in getting nvidia card detected in docker correctly ?

 

Tried following the instructions provided in this thread, but they're over a year old now and can't get docker to see the GPU (1650).

 

Tried the guide on this site and their driver starts fine, but after running their command to fix permissions for the nvidia driver and then attempting to start the driver package again, i see a kernel panic in dmesg output and the driver never starts.

 

https://blog.kkk.rs/archives/17

 

DVA3221 / DSM 7.2.64570-U3 / GTX 1650 on ESX 8

 

 

Ignore this, tried it again on a clean system and now I can get hardware support in docker for the 1650 card running in a VM on ESX 8

 

 

Screenshot 2023-11-16 125043.png

  • Like 1
Link to comment
Share on other sites

  • 2 weeks later...
9 hours ago, anhyeuviolet said:

 

I am trying this method on Baremetal but look like it is only working on VMs.

 

It's working on both baremetal and VM.  I have two boxes, one is virtualized and the other is baremetal and both have a GTX1650 and it's working on both of them.

 

Install as usual, add the custom nVidia driver and ensure it's running, then SSH into the machine and run the command in terminal "vgpuDaemon fix" as root.  Stop and restart the custom nVidia driver and then the GPU should appear in the 'info center' in control panel.

 

Screenshot2023-11-27174758.png.20225e9e0a32d9bb239ac5ee167cbf8c.png

  • Like 1
Link to comment
Share on other sites

On 11/28/2023 at 12:48 AM, irishj said:

 

It's working on both baremetal and VM.  I have two boxes, one is virtualized and the other is baremetal and both have a GTX1650 and it's working on both of them.

 

Install as usual, add the custom nVidia driver and ensure it's running, then SSH into the machine and run the command in terminal "vgpuDaemon fix" as root.  Stop and restart the custom nVidia driver and then the GPU should appear in the 'info center' in control panel.

 

Screenshot2023-11-27174758.png.20225e9e0a32d9bb239ac5ee167cbf8c.png

 

Thanks for your feedback.

However the GTX1650 is naitive as I know.

 

I am trying to get my GTX1660S works but still not success.

Edited by anhyeuviolet
Link to comment
Share on other sites

As I already wrote, at least for DVA1622, I was able to confirm GPU compatibility is hardcoded in Surveillance Station binary/library.

No matter you make it work for HW decoding with Plex with 10th gen addon etc... Surveillance Station AI tasks crash no matter what.

 

It would not be surprising the same exists for DVA3221 and Nvidia library/drivers...

I bet as long as the GPU is natively detected (so compatible) by DVA3221 it will work, but any custom drivers/addons not hardcoded in Surveillance Station will fail...

  • Like 1
Link to comment
Share on other sites

6 hours ago, anhyeuviolet said:

 

Thanks for your feedback.

However the GTX1650 is naitive as I know.

 

I am trying to get my GTX1660S works but still not success.

 

Do you see the GPU when issuing the command nvidia-smi ?  

 

Is the custom nVidia GPU driver installed and running in package manager ?

 

Please share the exact install steps you followed, in order.

Edited by irishj
  • Like 1
Link to comment
Share on other sites

16 hours ago, irishj said:

 

Do you see the GPU when issuing the command nvidia-smi ?  

 

Is the custom nVidia GPU driver installed and running in package manager ?

 

Please share the exact install steps you followed, in order.

Thank you very much ! Your previous post with note about "vgpuDaemon fix" that fix my issue.

nvidia-smi now is showing CUDA core and Driver version as: 5XX.

 

However, Surveillance stations AI will not work, ATM (or even in the future).

 

20 hours ago, Orphée said:

As I already wrote, at least for DVA1622, I was able to confirm GPU compatibility is hardcoded in Surveillance Station binary/library.

No matter you make it work for HW decoding with Plex with 10th gen addon etc... Surveillance Station AI tasks crash no matter what.

 

It would not be surprising the same exists for DVA3221 and Nvidia library/drivers...

I bet as long as the GPU is natively detected (so compatible) by DVA3221 it will work, but any custom drivers/addons not hardcoded in Surveillance Station will fail...

You're correct.

Now I just sit tie and wait till the next release and hope they can support AI functions.

Link to comment
Share on other sites

2 hours ago, anhyeuviolet said:

Thank you very much ! Your previous post with note about "vgpuDaemon fix" that fix my issue.

nvidia-smi now is showing CUDA core and Driver version as: 5XX.

 

However, Surveillance stations AI will not work, ATM (or even in the future).

 

You're correct.

Now I just sit tie and wait till the next release and hope they can support AI functions.

From who ? Synology ?

Why would they add support to any other card than GTX1650 ?

Unless they release a new DVA product with some new card, don't expect it.

Link to comment
Share on other sites

24 minutes ago, anhyeuviolet said:

Nope. A Chinese Dev is working on newer drivers.

FYI:

https://blog.kkk.rs/archives/17

So you did not understand my point.

 

No matter what driver will actually work. He could make latest 4090 RTX works on NAS system.

As long as you/he/anyoned do not rebuild Surveillance Station SPK to accept these cards, you will only use these drivers to have HW transcoding capabilities for software like plex etc... but never for Surveillance Station AI features.

 

No matter if nvidia-smi detect your card. if it is the same as DVA1622, Surveillance Station will generate some dump/core crash files on the filesystem in loop.

 

If you are lucky, DVA3221 Surveillance Station SPK is not hardcoded like DVA1622 is... and then my point is irrelevant.

Edited by Orphée
  • Like 1
Link to comment
Share on other sites

5 hours ago, anhyeuviolet said:

Thank you very much ! Your previous post with note about "vgpuDaemon fix" that fix my issue.

nvidia-smi now is showing CUDA core and Driver version as: 5XX.

 

However, Surveillance stations AI will not work, ATM (or even in the future).

 

You're correct.

Now I just sit tie and wait till the next release and hope they can support AI functions.

 

Great, at least now you can use hardware transcoding for Plex / Emby / Jellyfin :)

 

If you are going to use h/w transcoding, I'd advise using the patch for nVidia drivers which allows more than 3 simultaneous NVENC video encoding sessions , which is a driver limitation imposed by nVidia on their consumer GPUs.

 

https://github.com/keylase/nvidia-patch

 

Clone the repo, run the patch, then restart the custom nVidia driver and you're good to go.

  • Like 1
Link to comment
Share on other sites

On 11/30/2023 at 5:18 PM, Orphée said:

So you did not understand my point.

 

No matter what driver will actually work. He could make latest 4090 RTX works on NAS system.

As long as you/he/anyoned do not rebuild Surveillance Station SPK to accept these cards, you will only use these drivers to have HW transcoding capabilities for software like plex etc... but never for Surveillance Station AI features.

 

No matter if nvidia-smi detect your card. if it is the same as DVA1622, Surveillance Station will generate some dump/core crash files on the filesystem in loop.

 

If you are lucky, DVA3221 Surveillance Station SPK is not hardcoded like DVA1622 is... and then my point is irrelevant.

 

I understood your point very well.

 

From the latest post in Chinese, the dev is testing the "beta driver" that supports AI functions of Surveillance Station.

Just wait until the next release.

 

On 11/30/2023 at 6:49 PM, irishj said:

 

Great, at least now you can use hardware transcoding for Plex / Emby / Jellyfin :)

 

If you are going to use h/w transcoding, I'd advise using the patch for nVidia drivers which allows more than 3 simultaneous NVENC video encoding sessions , which is a driver limitation imposed by nVidia on their consumer GPUs.

 

https://github.com/keylase/nvidia-patch

 

Clone the repo, run the patch, then restart the custom nVidia driver and you're good to go.

Yup, at least I can share the heavy encoding tasks to the GPU and wait until the better release.

Edited by anhyeuviolet
  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...