I have laptop with NVIDIA Quadro T2000 GPU but this GPU is not support with the Omniverse.[see screenshot]
So I am trying with creating AWS EC2 Instance with G4dn.2xlarge instance type with windows server 2022.
Note: I want to Omniverse XR which only support Windows OS.
After creating EC2 instance in AWS, I have install NVIDIA driver from https://www.nvidia.com/Download/index.aspx .
Driver Details are as follows:
Product Type: Tesla
Product Series: T-series
Product: Tesla T4
Operating System: Windows 10
CUDA Toolkit:11.7
Driver is successfully installed I can see the status of installed, But still when I am trying to open any app in Omniverse, for ex. Omniverse create/XR, each and every app got crashed as soon as we launch.
I can see following logs in the terminal:
Image
Related
Description: I set up an AWS EC2 Windows Server 2019 instance. The instance I used is g4dn GPU instance class
Purpose: I want to use this instance for 3d modeling
Problem: When I open the 3d software. It says opengl 2.0 not supported by your system
My failed "solutions":
Install Nvidia driver. As I initially thought it's because GPU is not present
Use teamviewer. Based on online tutorial. They say connecting to Windows instance via RDP, the default opengl is 1.1, and teamviewer will resolve that. However, when I connect using teamviewer and check, the opengl is still 1.1
I tried AWS AMI Windows server 2019 with driver installed. However, it has the same problem as the instance I manually setup
I am trying to execute a codebase (detect and track by facebook) inside the docker container that requires GPU access. My docker image is linux based with CUDA toolkit installed however I cannot see any GPU devices in proc filesystem.
The host machine is windows with CUDA 9 toolkit and drivers installed.
When I try to execute the code it says:
CUDA driver version is insufficient for CUDA runtime version
On searching, I realized, I can leverage nvidia-docker plugin to map the host nvidia drivers to container OS but by the looks of it, I couldn't find any support for windows.
Is there any other way around to execute the codebase inside docker container on my windows OS?
It seems that this is not possible on any Windows OS.
https://github.com/NVIDIA/nvidia-docker/wiki/Frequently-Asked-Questions#is-microsoft-windows-supported
I try to run a Kinect, using ROS indigo and Ubuntu. Ubuntu is running in a virtual machine (Parallels) with Mac OS as an host.
I tried to test the Kinect via rosrun openni_tracker openni_tracker
with the error:
InitFromXml failed: Can't create any node of the requested type!
Therefore I assume that the system can't access the Kinect. However, using lsusb returns:
Bus 004 Device 008: ID 045e:02b0 Microsoft Corp. Xbox NUI Motor whenever i connect the Kinect
The following does not solve the problem:
http://answers.ros.org/question/9737/openni_tracker-initfromxml-failed-cant-create-any-node-of-the-requested-type/ (simply has no effect at all)
Kinect / Primesense (Xtion) ROS Ubuntu through Virtual Machine (VMware) (different vm and host, also it seems to be a different problem)
If anyone could help me make open_tracker see the device, that would be very helpful.
I'm in the process of setting up Torch on OSX on a 2015 macbook pro with a Radeon GPU using this library (cltorch) for OpenCL support
I can successfully run Torch scripts now, but running this test script which outputs the device and platform being used I get:
Using Apple , OpenCL platform: Apple
Using OpenCL device: Iris Pro
Obviously, I want torch to run on Radeon instead of the integrated Iris, but I have no idea how to do that.
You can use cltorch.setDevice to choose the device, like:
cltorch.setDevice(2)
When I try to call D3D12GetDebugInterface or CreateDXGIFactory2 with the flag DXGI_CREATE_FACTORY_DEBUG, the calls are failing on my laptop, but not on my desktop. Other calls work fine on the laptop and I can render stuff -- I just can't initialize the debug layer. The desktop has an oldish GeForce 650 Ti, and the laptop is a Dell XPS15 laptop (latest gen), which has a GeForce GTX 960M. Both have the latest drivers (361.43). Any ideas what could be going wrong?
I posted on the NVIDIA devtalk forum first, but cross-posted to stackoverflow because the official NVIDIA forum seems pretty dead.
For Direct3D 12, both the debug layer and the WARP12 device are part of the Graphics Tools feature-on-demand which is not part of the default install. You need to enable it on your desktop system.
This FOD package is specific to your version of Windows 10. If you had it enabled but then upgrade the OS (say from 10240 to 10586) the existing Graphics Tools feature-on-demand is disabled. You need to re-enable it so the newer matching version can be installed.
See Visual C++ Team Blog