Print this page
Published in AI

Microsoft expands Nvidia GPU capabilities

by on25 September 2018


Tweaked for AI and Cloud

At Microsoft Ignite, Microsoft announced an expansion of its NVIDIA GPU capabilities, offering for the first time a powerful addition to its lineup that features 8 Nvidia Tesla V100 Tensor core GPUs interconnected via Nvidia NVLink GPUs to use for compute-intensive AI projects.

Corey Sanders Corporate Vice President, Azure, announced the preview of two new N-series Virtual Machines with Nvidia  GPU capabilities.

"GPUs are ideal for compute and graphics-intensive workloads, and they have been helping you innovate through scenarios like high-end remote visualisation, artificial intelligence, and predictive analytics", he said.

A new version of the NV-series designed to support remote visualisation workloads and other graphics-intensive applications. The new NVv2 VMs will feature Nvidia GRID technology and  Tesla M60 GPUs, up to 448GiB of RAM, and will support Premium SSDs. The NVv2 VMs are in preview now in US West, US South Central, and Europe West.

NDv2 VMs which will be previewed by the end of the year, is a  new addition to the ND-series, NDv2 VMs focus on DL training and inferencing as well as machine learning. The new NDv2 will feature eight Tesla V100 Tensor core GPUs interconnected via NVLink GPUs and 40 Intel Skylake cores aimed to deliver high-quality results even faster.

Vole is announcing the preview of two new N-series Virtual Machines with Nvidia GPU capabilities. It reckons GPUs are ideal for compute and graphics-intensive workloads, and they have been helping  "innovate through scenarios" like high-end remote visualisation, artificial intelligence, and predictive analytics.

NDv2 VMs is a new addition to the ND-series, NDv2 VMs focus on DL training and inferencing as well as machine learning. The new NDv2 will feature eight Tesla V100 Tensor core GPUs interconnected via NVIDIA NVLink GPUs and 40 Intel Skylake cores aimed to deliver high-quality results even faster. NDv2 VMs are expected to be in preview by the end of 2018.

Vole is also announcing two new H-series VMs for High-Performance Computing (HPC) scenarios. These new VMs are optimised for performance and cost and are aimed at HPC workloads like fluid dynamics, structural mechanics, energy exploration, weather forecasting, risk analysis, and more.

HB VMs also to be previewed by the end of the year are VMs which feature 60 AMD EPYC cores and 240 GiB RAM. It's claimed yhey have the highest amount of memory bandwidth (260GBps) in the public cloud. This is valuable for calculations required in fluid dynamics and weather forecasting.

HC VMs will be previewed by the end of the year are VMs optimised for computationally intensive workloads. The HC will feature up to 352 GiB RAM, 44 Intel Skylake cores with clock speeds up to 3.7GHz, and will support Intel’s rich ecosystem of HPC software tools.

Last modified on 25 September 2018
Rate this item
(0 votes)