Hello,
Our lab bought the Nvidia K80 GPUs running in Dell rack server RE730 as host machine.
I wanted to know whether docker containers of deep learning libraries available on NGC registry will work with Tesla K80 GPUs.
Kindly guide me towards the tutorial available for configuring docker environment with Tesla K80 GPUs.
If there is some hands-on workshop conducted by NVIDIA team in India than please let me know.
Hi,
NGC deep learning containers only support NVIDIA Volta and Pascal GPU’s. Your system’s K80 GPU’s are a previous generation that are not compatible with the latest NGC deep learning framework containers.
Here is the official statement of supported platforms.
[i]NVIDIA® GPU Cloud (NGC) containers leverage the power of GPUs based on the NVIDIA Pascal™ and Volta™ architecture. NVIDIA DGX systems as well as Volta and Pascal-based virtual machines hosted on cloud service providers utilize these GPUs.
Aside from these classes of hardware, NGC containers can also run on PCs installed with the following NVIDIA products:
NVIDIA TITAN V (Volta)
NVIDIA TITAN X (Pascal)
NVIDIA TITAN Xp (Pascal)
NVIDIA Quadro GV100 (Volta)
NVIDIA Quadro GP100 (Pascal)
NVIDIA Quadro P6000 (Pascal)
The document describes how to set up an NVIDIA TITAN PC to run NGC containers. Open the command line on the Ubuntu desktop and paste the code blocks into the command line.
Prerequisites
These instructions assume the following:
You have a single card from the supported list installed on your PC.
You have Ubuntu Desktop 16.04 with 4.10 kernel installed. Ubuntu Desktop can be downloaded from https://www.ubuntu.com/download/desktop/.[/i]
The complete doc is here: https://docs.nvidia.com/ngc/ngc-titan-setup-guide/index.html