NVIDIA GPU Cloud now supports Microsoft Azure

Source: NVIDIA blog post. Microsoft Azure and NVIDIA GPU Cloud now talk to each other.

Microsoft Azure is now a supported platform with NVIDIA GPU Cloud (NGC). This allows developers to make use of on-demand GPU computing that scales as they need, while eliminating the complexity of software integration and testing.

Running popular deep learning software such as TensorFlow, Microsoft Cognitive Toolkit, PyTorch and NVIDIA TensorRT — requires software stacks that run smoothly and reliably, NVIDIA explains.

"For high-performance computing (HPC), the difficulty is how to deploy the latest software to clusters of systems. In addition to finding and installing the correct dependencies, testing and so forth, you have to do this in a multi-tenant environment and across many systems," said NVIDIA's Chris Kawalek, Senior Product Marketing Manager, NVIDIA GPU Cloud in a blog post.

"NGC removes this complexity by providing preconfigured containers with GPU-accelerated software. Its deep learning containers benefit from NVIDIA’s ongoing R&D investment to make sure the containers take advantage of the latest GPU features. And we test, tune and optimise the complete software stack in the deep learning containers with monthly updates to ensure the best possible performance."

"Containers from the NVIDIA GPU Cloud (NGC) container registry are now supported on NVIDIA Volta and Pascal-powered Azure NCv3, NCv2 and ND*. This brings together the power of NVIDIA GPUs in Azure cloud infrastructure with the comprehensive library of deep learning and HPC containers from NGC," elaborated Brett Tanzer, Partner PM Manager, Azure Specialized Compute at Microsoft in a complementary blog post.

Tanzer said that the NGC container registry includes NVIDIA-tuned, tested, and certified containers for deep learning software. "Through extensive integration and testing, NVIDIA creates an optimal software stack for each framework – including required operating system patches, NVIDIA deep learning libraries, and the NVIDIA CUDA Toolkit – to allow the containers to take full advantage of NVIDIA GPUs."

Explore:

Deploy NGC containers with Azure, through NVIDIA GPU Cloud Image for Deep Learning and HPC in the Azure Marketplace. This image provides a preconfigured environment for using containers from NGC on Azure.

• Launch a compatible NVIDIA GPU instance on Azure. 
• Pull containers as needed from the NGC registry into a running Azure instance. (A free NGC account account is required) 
• Read the Using NGC with Microsoft Azure documentation

Containers from NGC on Azure NCv2, NCv3, and ND virtual machines can also be run with Azure Batch AI by following GitHub instructions.

*Kawalek elaborated that the Microsoft Azure instances, or virtual machines, would be equivalent to physical machines with the following hardware:

• NCv3 (one, two or four NVIDIA Tesla V100 GPUs)
• NCv2 (one, two or four NVIDIA Tesla P100 GPUs)
• ND (one, two or four NVIDIA Tesla P40 GPUs)