I'm not familiar with the technology Amazon has here. But is there any opportunity here to engage this service with Octane now or in the future?
https://aws.amazon.com/ec2/elastic-gpus/
Amazon Elastic GPU's
Moderator: juanjgon
-
upinspiringmotion
- Posts: 2
- Joined: Fri Apr 24, 2015 1:17 am
I don't think it works with Elastic GPU's since those GPU's are not NVidia CUDA graphic cards. Those are virtual GPU's.
The question that I'm trying to find out for more than a week is: Does Octane work in any of the instances provided by Amazon EC2?
The question that I'm trying to find out for more than a week is: Does Octane work in any of the instances provided by Amazon EC2?
The Elasctic Graphics product at AWS is a network attached graphics adapter that offload OpenGL calls via a custom (amazon devloped??) API hook in Windows. It currently is not a CUDA or similar capable system, but only a display accelerator. What it does, effectively is take OpenGL calls to the OS, sends them to the networked graphics hardware and delivers rendered frames back to your instance. There is no compute capability right now.
OpenGL 4.3 and GLSL 4.3 is all that's supported according to the documentation I've been able to find so far. It's primarily for use when you're setting up an AWS based virtual workstation for graphics intensive apps under Windows. I imagine any CPU native 3D app that uses OpenGL for the interface will benefit from this, but it's not going to assist with any GPU based compute needs.
ORC is the way to go for remote (non-on-premises networked) Octane renders right now.
All that may change in the future, but I imagine it'll depend entirely on Nvidia and others getting sane about their licensing on the use of their cards in such use cases. Until then, ORCs are your friend.
Scott
OpenGL 4.3 and GLSL 4.3 is all that's supported according to the documentation I've been able to find so far. It's primarily for use when you're setting up an AWS based virtual workstation for graphics intensive apps under Windows. I imagine any CPU native 3D app that uses OpenGL for the interface will benefit from this, but it's not going to assist with any GPU based compute needs.
ORC is the way to go for remote (non-on-premises networked) Octane renders right now.
All that may change in the future, but I imagine it'll depend entirely on Nvidia and others getting sane about their licensing on the use of their cards in such use cases. Until then, ORCs are your friend.
Scott