3 min read
We are excited to announce the public preview of two new N-series Azure Virtual Machines with GPU capabilities. GPUs are ideal for compute and graphics-intensive workloads, and they have been helping our customers innovate through scenarios like high-end remote visualization, machine learning, HPC simulations, and predictive analytics.
The first offering is part of the Azure VM NV-series, which have been architected to support powerful remote visualization workloads and other graphics-intensive applications backed by the NVIDIA Tesla M60 GPU.
NV-series VMs allow customers to run any graphics intensive workload or application in the cloud and deliver the experience to any device, anywhere, anytime. The new NVv2 will feature up to 448GiB of RAM basically doubling the amount of memory present in the prior VM generation (NVv1) and will include an upgraded CPU class. The NVv2 VMs will also support Premium SSDs and they have been optimized for applications like CAD, gaming and interactive 3D design. And lastly, NVv2 will continue to come with NVIDIA GRID licensing cost built into the VM price.
Some of our existing customers – like Nimble Collective – are looking forward to bringing their customer experience to the next level with these new offerings:
“Nimble Collective chose to position the Nimble Studio platform on top of Azure for a wide variety of reasons, not the least of which is the power of the new NVv2 virtual machines that they are bringing to market. The original NV offering easily supported the complexities of the Nimble Studio infrastructure, but the NVv2 VMs will remove any remaining obstacles – around application startup times in particular, where NVv2 will now support premium SSD. Users are online and productive faster than before.
Additionally, because of Nimble Studio's demand on graphics systems, the NVv2 VMs use of the NVIDIA M60 GPUs will hit our market's sweet spot for performance.”
Rob DeMillo | Chief Technology Officer, Nimble Collective
The second offering will be part of the Azure VM ND-series, which is focused on machine learning, HPC and inference scenarios. The new NDv2 will feature 8 NVIDIA Tesla V100 NVLINK interconnected GPUs and 40 Intel Skylake cores and 672 GiB of system memory. NVIDIA® Tesla® V100 is the world’s most advanced data center GPU ever built to accelerate machine learning and high-performance computing workloads more broadly. We continue to be inspired by how these technologies have been able to make remarkable progress in a variety of fields: from cancer therapy to helping virtual personal assistants converse naturally and predicting the next hurricane.
We continue to work with our customers and partners to enable them to bring to market true innovation. For example, solutions like Altair ultraFluidX™ – a simulation tool for ultra-fast prediction of the aerodynamic properties of passenger and heavy-duty vehicles as well as building and environmental aerodynamics – and Altair Altair nanoFluidX™ – a particle-based fluid dynamics simulation tool to predict the flow in complex geometries with complex motion – will be able to benefit from our new Azure NDv2 VMs:
“Altair ultraFluidX and Altair nanoFluidX are industry leading fluid dynamics simulation tools that leverage the latest NVIDIA GPU solutions. The increased throughput and lower cost of GPU solutions is going to allow for a significant increase in simulations, which can be used to further impact the design process. We look forward to getting early access for our GPU based simulation solutions to Microsoft’s new NDv2 instances that feature 8 NVIDIA Tesla V100 NVLINK interconnected GPUs and 40 Intel Skylake cores.”
Piush Patel | Vice President, Corporate Development for HPC and Cloud – Altair.
Additionally, we continue to collaborate with our Speech and Language Group in the Microsoft Cloud and AI team to take advantage of the latest capabilities this new Azure VM series can provide:
Voice is the new interface driving ambient computing as speech recognition has been transforming our daily lives from digital assistants, dictation of emails and documents, to transcriptions of lectures and meetings. These scenarios are possible today thanks to years of research in speech recognition and technological jumps enabled by neural networks where Microsoft has been at the forefront. With the continuously expanding fleet of Microsoft Azure GPU accelerated VMs, like the ND v2 packing an 8-way NVIDIA Tesla V100 NVLINK interconnected GPUs together with a Skylake based processor, we are excited of the possibilities to boost the training speeds of our large scale speech recognition models powering Cortana, Skype translator, Windows and Office. Looking forward to their launch!”
Xuedong Huang | Technical Fellow, Lead Speech and Language Group in Microsoft Cloud and AI