OpenVZ Forum


Home » General » Support » CUDA support inside containers (How can I run CUDA workloads in multiple containers?)
Re: CUDA support inside containers [message #52621 is a reply to message #52619] Mon, 14 November 2016 18:04 Go to previous messageGo to previous message
khorenko is currently offline  khorenko
Messages: 533
Registered: January 2006
Location: Moscow, Russia
Senior Member
Hi,

we tried this several years ago and it seemed to work with no additional development from our side,
so you can just try and tell us if it works now or not.

The basic steps were:
1) install both on the Node and inside Containers
a) nVidia drivers for CUDA support
b) CUDA compiler kit providing nvcc compiler
(there were some linker problems, but they were resolvable)

2) provide Containers access to nvidia0, nvidia1, nvidiactl devices

3) run a cuda-enabled program inside a Container

That's it. Smile

Hope that helps.


If your problem is solved - please, report it!
It's even more important than reporting the problem itself...
 
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic: CVE-2016-7910 CVE-2016-7911
Next Topic: Can you rebuild vz/private files?
Goto Forum:
  


Current Time: Fri Aug 29 18:18:46 GMT 2025

Total time taken to generate the page: 0.13061 seconds