Yesterday, Microsoft announced NVIDIA GPU Cloud (NGC) support on its Azure platform. Following this, data scientists, researchers, and developers can build, test, and deploy GPU computing projects on Azure.
With this availability, users can run containers from NGC with Azure giving them access to on-demand GPU computing that can scale as per their requirement. This eventually eliminates the complexity of software integration and testing.
The need for NVIDIA GPU Cloud (NGC)
It is challenging and time-consuming to build and test reliable software stacks to run popular deep learning software such as TensorFlow, Microsoft Cognitive Toolkit, PyTorch, and NVIDIA TensorRT. This is due to the operating level and updated framework dependencies. Finding, installing, and testing the correct dependency is quite a hassle as it is supposed to be done in a multi-tenant environment and across many systems. NGC eliminates these complexities by offering pre-configured containers with GPU-accelerated software.
Users can now access 35 GPU-accelerated containers for deep learning software, high-performance computing applications, high-performance visualization tools and much more enabled to run on the following Microsoft Azure instance types with NVIDIA GPUs:
- NCv3 (1, 2 or 4 NVIDIA Tesla V100 GPUs)
- NCv2 (1, 2 or 4 NVIDIA Tesla P100 GPUs)
- ND (1, 2 or 4 NVIDIA Tesla P40 GPUs)
According to NVIDIA, these same NVIDIA GPU Cloud (NGC) containers can also work across Azure instance types along with different types or quantities of GPUs.
Using NGC containers with Azure is quite easy. Users just have to sign up for a free NGC account before starting, then visit Microsoft Azure Marketplace to find the pre-configured NVIDIA GPU Cloud Image for Deep Learning and high-performance computing.
Once you launch the NVIDIA GPU instance on Azure, you can pull the containers you want from the NGC registry into your running instance.
You can find detailed steps to setting up NGC in the Using NGC with Microsoft Azure documentation.